Foreign Exchange Options and Risk Management

Foreign Exchange Options and Risk Management

Market Dynamics, Models and Human Behaviour

Demetri Papacostas and Francesco Tonin

Risk Books is a trading name of Infopro Digital Risk Limited
© Infopro Digital Risk (IP) Limited 2018
All rights reserved.

Infopro Digital
Haymarket House 28–29 Haymarket
London SW1Y 4RX
Tel: +44(0) 20 7316 9000
Sites: www.riskbooks.com

© 2018 Infopro Digital Risk Limited

ISBN 978-1-78272-262-5

British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library

Publisher: Nick Carver

Commissioning Editor: Sarah Hastings
Managing Editor: Lewis O’Sullivan
Designer: Lisa Ling
Copyeditor: Powan Publishing Ltd

Typeset by Mark Heslington Ltd, Scarborough, North Yorkshire

Printed and bound in the UK by PrintonDemand-Worldwide

Conditions of sale
All rights reserved. No part of this publication may be reproduced in any material form whether by photocopying or storing in any medium by electronic
means whether or not transiently or incidentally to some other use for this publication without the prior written consent of the copyright owner except in
accordance with the provisions of the Copyright, Designs and Patents Act 1988 or under the terms of a licence issued by the Copyright Licensing Agency
Limited of Saffron House, 6–10 Kirby Street, London EC1N 8TS, UK.

Warning: the doing of any unauthorised act in relation to this work may result in both civil and criminal liability.

Every effort has been made to ensure the accuracy of the text at the time of publication, this includes efforts to contact each author to ensure the accuracy
of their details at publication is correct. However, no responsibility for loss occasioned to any person acting or refraining from acting as a result of the
material contained in this publication will be accepted by the copyright owner, the editor, the authors or Infopro Digital.

Many of the product names contained in this publication are registered trade marks, and Risk Books has made every effort to print them with the
capitalisation and punctuation used by the trademark owner. For reasons of textual clarity, it is not our house style to use symbols such as TM, ®, etc.
However, the absence of such symbols should not be taken to indicate absence of trademark protection; anyone wishing to use product names in the public
domain should first clear such use with the product owner.

While best efforts have been intended for the preparation of this book, neither the publisher, the editor nor any of the potentially implicitly affiliated
organisations accept responsibility for any errors, mistakes and or omissions it may provide or for any losses howsoever arising from or in reliance upon
its information, meanings and interpretations by any parties.
 Contents

About the Authors

Acknowledgements

Foreword

Abbreviations

Introduction

PART I: THE FX MARKETS, THE GENESIS OF A NEW WORLD ORDER

1 A Brief History of the FX Market’s Evolution

In the Beginning
The Fun Begins
Technology and Regulation: Key Catalysts for Change
Conclusion

2 Foreign Exchange Markets

The Who, What and Where of FX
A Huge Market
FX Mostly Trades in Europe
It Happens in the UK, but the US Dollar and US Banks Rule
FX Fixings: A Storm in a Teacup?
Culture, Collusion, Collision, Fines and Jail Terms, More Transparency
Conclusion

3 Predicting FX Movement
Forecasting FX: Practical Considerations
Longer-term Planning
Follow the Money: Using Market Sentiment to Predict FX
Putting it all Together
Some Thoughts on the Applications of Currency Forecasting
Conclusion

PART II: THE MARKET AND TRADING FLOOR DYNAMIC AND THE FORMULAS
THAT DRIVE IT

4 Basic FX Instruments
The Anatomy of an FX Trading Floor
Foreign Exchange Spot
The Forward and NDF Markets
FX Swaps
Cross-currency Basis
 Conclusion

5 Trading Floor Dynamics

Making an Ethical Price, in Theory
Shades of Grey
How the Sell Side is Compensated
Conclusion
Appendix: the Indictment

6 FX Options: An Intuitive Approach

The Hedger’s Paradox
A Crystal Ball and Some Practical Considerations
Option Expiration, Calculating P&L
What is Better, an Option or a Forward?
Some Option Fundamentals
Conclusion

7 Famous Formulas, Fame and Fortune

On the Hunt for the Formula with NYC Hustlers
Pricing Simple Optionality
Real-World Complications
Choosing the Right Crystal Ball
Conclusion

8 Getting to the Formula and the Correct Probability Distribution

A Quest for the Correct Future Distribution
Circling in on the Famous Formula
Volatility Briefly Explained
Is the Black–Scholes Formula Correct?
It’s All About the Premium
Conclusion

9 The Greeks – A Practical Approach

DELTA: δ
GAMMA: Γ
THETA: Θ
Vega Or Kappa: Volatility
Interest Rate Sensitivity: Phi Φ, Rho Γ and More Advanced Greeks
Conclusion

10 Portfolio Management and Second-order Greeks

Option Portfolio Management: the Theory and the Reality
Delta Hedging as a Mechanism to Lose Money
Every Option Has Its Own Volatility: The VOL Smile
Risk Reversals and Their True Implication
Risk Reversals, Butterflies and Portfolio Management
Conclusion

11 FX Options Trading Book & Risk Measurement

Delta, Gamma and Vega Considerations in Hedging the French Elections
Risk Across the Institution: VaR
Risk Across the Institution: Scenario Analysis and Stress Testing
Conclusion
PART III: DRIVERS AND TOOLS FOR HEDGERS AND SPECULATORS

12 Hedging FX Risk at Corporations

The Corporate Treasurer and the FX Conundrum
Corporate Hedging Best Practices

13 You Have Options

Hedging Structures for Corporate Hedgers
Final Thoughts on Corporate Hedging Structures

14 Situations Gone Mad, From the Most Complex to the Simplest

Complex Trades Tend to Have High Hedging Costs
Creating Attractive Trades
What Constitutes Good Performance?
Two Traders, Two Very Different Approaches
Structures Gone Mad: TARNs
Market Situations Gone Mad: the Case of Euros Versus Swiss Francs
Summary of Extraordinary Situations

15 Speculators and Hedge Funds: How Do Portfolio Managers Make Money?

Do Successful Portfolio Managers Have Talent?
How Do Portfolio Managers Make Money?
To Make Money, Get an Edge
Are Edges Ethical?

16 Speculating and Hedging: The Fundamental Differences

How Is Speculating Different From Hedging?
Why Bother With Options?
Relative-Value Trades
Clever Ways to Increase Speculative Leverage
Portfolio View
A Strategy for Every View

Epilogue
Conclusion

References
 About the Authors

Demetri Papacostas is head of the Market Specialist Group at Bloomberg. His FX career spans 35 years, with
extensive experience developing and running profitable FX and derivative businesses for major financial
institutions. His FX derivative product specialisation includes trading, structuring and selling to corporations, hedge
funds and asset managers, as well as to the private wealth arms of money centre banks. Prior to Bloomberg, Demetri
was a managing director at JPMorgan, where he ran the commercial bank FX sales team in New York and FX
derivatives sales nationally, and he previously also managed its currency derivative structuring team. He also
established a global FX derivative sales presence for ABN AMRO, and previously managed currency and precious
metal options business for Royal Bank of Canada, Security Pacific (Bank of America), UBS and Marine Midland
Bank.
Demetri is a well-known speaker at financial forums and webinars, has been quoted in the Wall Street Journal,
Profit & Loss magazine and Bloomberg Business Summit 2014, CFO Edition, and has been published in
Bloomberg’s Markets and appeared on Bloomberg TV to discuss the FX markets. He received a BA in economics
and computer science from New York University, and an MBA from Pace University, with numerous honours and
distinctions for academic achievement.

Francesco Tonin is an FX market specialist at Bloomberg, based in New York. Most recently, he was the North
America head of FX structuring at Citigroup, having been recruited initially at Morgan Stanley for a sales position in
interest rates derivatives. Francesco then worked at Deutsche Bank in FX corporate sales, and subsequently in FX
exotic derivatives trading, and at Credit Suisse held an FX structuring role. He has given talks together with
corporate speakers such as General Motors, Fluor Corporation, Dell Computers, Knowles Corporation, to the
national Association for Financial Professional conference, Texpo, Windy City, Minnesota AFP, New York Cash
Exchange, the IMF, Princeton University and Columbia University. Francesco has written a couple of hundred
articles on financial matters, with contributions to Bloomberg Markets magazine and Bloomberg Briefs.
His mathematical research activity focused on the Cauchy problem for partial differential equations with
holomorphic coefficients, for which he received research grants from the EU, the French Ministère des Affaires
Etrangères, the University of Padova and the University of Pisa, in addition to the Italian National Research Council
and a Fulbright Scholarship offer. Francesco holds a BS in mathematics from the University of Padova, a PhD in
mathematics from the University of Torino and an MBA from Columbia University, and has also held a tenured
research position at the University of Padova.
 Acknowledgements

Demetri Papacostas
Acknowledgements and dedications, by definition, are personal statements that will most likely be read only by
people that feel they have a personal connection with the author, and will potentially look for a rightly deserved
mention. That becomes a difficult endeavour when I consider that every single thing I’ve contributed to this book
has been learned over the last 30-plus years from people who generously and patiently shared their knowledge.
Trading floors were equal-opportunity enablers for people to be decent, or not. Many chose the latter. However,
overwhelmingly I had the unbelievable good fortune to work with many wonderful people who remain my friends.
In the global FX market, people were particularly close. One’s relationships were the focal point, as we all
switched jobs and navigated the mega-mergers between the handful of global shops. Therefore, any attempt to
mention those who contributed to my FX options knowledge would be a fool’s errand. I dedicate this book to all of
you, and thank you from the bottom of my heart. Thank you for your collaboration, patience and friendship. Truth
be told, the subject matter always played second fiddle to my allegiance to each and every one of you.
As we discuss in the book, where each of us ends up is often nothing more than the roll of the dice. To that end, I
have had some wonderful individuals load the dice in my favour with their belief in me – in particular, Bob Mark,
Fred Stambaugh, David Puth, Mark Babunovic, Rob Lichten, Phil Cunn and the late Gary Turkel (who is greatly
missed). Thank you.
The easy part of acknowledgements is recognising those whose influence was such that this book, or my FX
career for that matter, would never have been possible. At the very top of that list is my wife and other half of my
soul, Susan, my children, Katerina and Erik, Nicholas and Zoe, and my sister Maria Kombogiannis. Their
encouragement and misplaced belief that I can move mountains (or even write a book) were always an
unquestioning force that moved this project forward. In that category I would add my dearest friends, Matt Daniel
and Fr Elias Villis, who, along with Susan, are my biggest fans and advocates. I have always wanted to send Susan
and Matt to my job interviews to speak on my behalf. A special thanks also to George Zimmar for demystifying
book writing and for his encouragement.
This book would never have been completed, and certainly not have been as interesting or entertaining, if it was
not for my genius of a co-author, Francesco. His mathematical acumen is surpassed only by his considerate analysis
of human nature and sense of humour. I am convinced there is no subject on which Francesco has not read
extensively and analysed thoroughly. Franscesco and I would particularly like to thank our friend Alison Fletcher for
her comments on the FX code of conduct, on the CNY developments, and for reminding us to mind our manners and
keep the appropriate tone throughout the book.
Finally, the only reason I thought of writing this book at all was because my employer encouraged it. An amazing
company that has transformed financial markets by breaking up cartels of knowledge and forcing transparency
across all financial endeavours for the last 35 years. This past decade, at Bloomberg, I have worked with the
smartest, most creative people of my career. Among them Dharrini B. Gadiyaram, who kindly reviewed it.

Francesco Tonin
I always thought that you never read the acknowledgements before reading the book, and you only (maybe) read
them if you have liked the book. So, jump right to the beginning and come back to this once you are done.
This book was so much fun to write, and it was born of a joy for creative work that I picked up from my wife
April, an illustrator, and my daughter Vittoria, a budding videographer and writer, who showed me the way by
example. This project was also special because of the chance it gave me to work with Demetri, a magnetic
personality with infectious enthusiasm. He is a really special man. He explained the project, and then asked if I
wanted to be part of it. “Demetri, you already knew my answer before you asked”, I responded, and the game was
on. My only regret is not having even more time for projects together.
There are many to which I am indebted, and this can only be a short list. Matthew Daniel, at Wells Fargo, is one
of the greatest intellects I have ever met – a very charismatic person who instilled in me the pleasures of
understanding technical materials and the subtle human reasons behind otherwise arbitrary accounting rules. To the
list of great curious minds and scientific intellects I have to add Krishna Kumar, formerly at Omega Advisors and
MKP Capital. Kris is a volcanic mind, a polymath who has the very rare ability of reading any mathematical
research paper in 10 minutes and getting the main ideas behind it. When your mathematical mind is ready for a great
conversation, there is nobody better than Kris.
I also want to mention another sharp mind, Yevgeny Frenkel, formerly at Morgan Stanley, a character greater
than life to whom I turn in the darkest and most challenging moments. He does not know how important it has been,
in those moments, to talk to him. His world view is extremely smart and should be the subject of a book. Probably a
novel. Another great accounting intellect to whom I am personally indebted is Rob Baer, formerly a managing
director at Bank of America in Chicago. He taught me a lot of accounting, but also a great approach to one’s own
career and to life. He has a fantastic sense of humour and of friendship.
My boss at Bloomberg, Steve Mendolia, is another of the reasons I like my job so much. He has the managerial
genius of being able to get 110% of what you can give by praising you. He is able to carve out for each member of
his team a special place where they can thrive and fulfil their professional inclinations. Many more people at
Bloomberg have had a deep influence on the concepts that constitute the skeleton of this book, among them two
other mathematical virtuosos, Fabio Mercurio and Mario Bondioli. Their conversation is the best dish served at
Felidia, that great pillar of Italian cooking in midtown Manhattan.
Among other colleagues at Bloomberg I should mention the pyrotechnic economist Stephen Jonathan, who will
always be the best read person around, and with whom I share a passion for chocolate. Of course, I should mention
and thank Alison Fletcher, who helped the creation of this book immensely. I always quip that working at
Bloomberg offers many perks, but I am the only employee who enjoys the perk of sitting between Steve and Alison.
I was also always very impressed by the strategic thinking of another of my colleagues, Owen Minde, who is
younger than I but who regularly teaches me how to approach the challenges of selling. As does Mike Briody,
another of my teachers. I am learning every day the virtues and the persistence of his approach to selling, and his grit
for “getting it done”.
Bloomberg would not be the same place without Bruno Dupire and Arun Verma, who work in a wonderful
thinktank called Quantitative Research. Their vision and thought leadership reverberates all across that financial
community known as “the quants”. They have made us all feel at home and members of the same team. Their
creation of the monthly Bloomberg Quant Seminar has fostered the community in a powerful and unique way.
Among the other great people that I have had the pleasure to work with during my banking career, I would like to
mention just a few: Joe Hedberg and Kelly Wang at Bank of America, Simon Hards, Pradeep Janjee and Grace Koo
at Credit Suisse, Andrew Baxter and Lynn Corsetti at Deutsche Bank, Robert Giacomelli, Steve Leach, Debbie
Connelly, Michele Ghidoni, Caren Pacifico and Mark Veale at Citigroup, Alex Maja and Andrius Gerasimovas at
Morgan Stanley.
Travelling to that great city, Chicago, would be much less fun for me if it were not for a couple of great friends I
can always count on: Heather Alger at PNC and Brigid Brennan at US Bank. They are always there for me when I
need some help. And of course I always enjoy a good deep-dish pizza, but that is another story.
Sadly, life in foreign exchange has been more dangerous of late, given all the investigations brought by various
agencies and the convictions. This is a fascinating topic that borders on philosophy, when it does not involve more
tragic aspects of the human condition. My limited knowledge of the legal aspects of market dynamics is entirely due
to the educational efforts of Lawrence Gerschwer and Nicole Love at Fried, Frank, Harris, Shriver & Jacobson in
New York. They have taught me all I know about the intersection of law and markets, and have a wonderful sense of
the unexpected consequences life can throw at us.
My previous life as a mathematician was a lot of fun as well, and of course much of it shows today in every
comma I write. Some of the very influential mathematicians that have shared their thoughts with me over the years
are Francesco Bottacin and Andrea D’Agnolo at the University of Padova, and Giovanni Taglialatela at the
University of Bari, Alberto Scalari, formerly at Barclays, and of course Mohammed Chadli and Olivier Berni,
formerly at the Université Pierre et Marie Curie in Paris. The years I spent in Paris working on horrendous formulas
were so special to me because I had them as my office mates.
Finally, I may be in a minority, but I always thought that a sense of humour is one of the highest rewards of
intelligence. Therefore, to complete the recipe, I added a pinch of my mamma’s terrific sense of humour to the mix,
because nothing can be good that is not funny!
 Foreword

The foreign exchange market is the biggest of the capital markets, with an average daily turnover in excess of US$5
trillion. That is such a mind-boggling number; to put it into perspective, consider that in one week its turnover is
sufficient to cover world trade for a whole year. It is also one of the most exciting markets. It is a battle of wits
between computers, central bankers, traders, hedge funds, corporations, asset managers, debt managers and
speculators, from Monday morning in Wellington to Friday evening in New York. Given the globalisation of
savings, all of the factors that drive debt and equity markets can influence the foreign exchange market and more.
The FX market is more akin to the three-level chess played in Star Trek than the two-dimensional version we may
be more familiar with. There is the cash or spot market, multi-segmented derivatives markets (eg, forward, futures
and swaps), an options market, and all sorts of combinations thereof. Liquidity is shifted through these various
channels, and there is great variation among the currencies.
The foreign exchange market is also a new market even though both money and exchange are as old as recorded
history. However, the modern FX market was made possible by the breakdown of the Bretton Woods monetary
system. That 1944 agreement had pegged the US dollar to gold, while the other currencies were pegged to the dollar.
It was not meant to be a strictly fixed exchange rate system, but it became ossified because of political
considerations. The end of Bretton Woods and the brief attempt to re-base it (via the Smithsonian Agreement)
ushered in the modern era of floating exchange rates.
FX is also post-modern in terms of the relationship between the signal and signifier. The thing of value is
signified by something of little value, such as seashells or metal, which outside of jewellery and ostentatious
displays, has limited use value. Paper money then became the thing of little use value that represented that thing of
value. And now electronic blips have largely replaced paper money, which is no longer backed by the thing of little
use value, to represent the thing of value. The signal has become free of the signifier and has taken on a life of its
own.
The global liberal order that has arguably been threatened by a new rise of nationalism since the great financial
crisis was not born at Bretton Woods, but evolved through positive government action beginning in earnest in the
late-1970s and early-1980s. Severing the dollar’s relationship from gold and allowing currencies to fluctuate was not
sufficient. Capital markets were purposely liberalised and de-regulated. It was a global phenomenon associated with
the UK prime minister Margaret Thatcher and US president Ronald Reagan.
In a world in which capital is highly mobile, FX exposure is an important way for investors to seek
diversification, which seems the closest thing to a Holy Grail of investing. Foreign exchange is also central to the
other capital markets. Given the relative volatility, the currency component can average as much as two-thirds of the
return portfolio of international bonds and one-third of the return of a portfolio of international equities. There are
two key areas of foreign exchange: understanding the drivers and direction of the market, which is economic,
political and financial in nature; and the operation of the FX market, which is about the players, motivations, tools
and rules of engagement, and is the focus of this important new book.
The complex relationship of evolving business practices and technology, including financial innovation,
regulation and accounting rules, is remaking the FX market. Foreign Exchange Options and Risk Management is
therefore focused on helping both lay people and professionals navigate through these treacherous and exacting
waters. Imagine the efficiencies that have evolved – for instance, the EURUSD exchange rate, the most commonly
traded currency pair in the world, is often quoted to a thousandth of a penny. It is probably unfair to conceptualise
the FX market as a zero-sum exercise, as not all participants are operating in the same timeframe, nor are they all
profit-maximisers.
While speculators and hedge funds may seek a profit in the foreign exchange market, others, such as corporate
treasurers, are looking to lock in certainty. Some are hedging, with what is gained in the hedge being given back in
the underlying, ideally. What is a multi-sigma event for a day trader may be a new buying opportunity for the
weekly trader. For global fund managers, it is often a transactional vehicle. To buy the foreign stock or bond, foreign
currency is needed. For others, such as bond fund managers, it is a risk that can be hedged to keep a pure bet on
interest rates. It can also be used to convert to the denomination of the investor (share class) or a benchmark.
 Technological advances are also rapidly changing how the business is conducted, and how trades are executed
and recorded. Increasingly, computers are working directly with other computers, initiating trades, sending messages
such as confirmations to counterparties. Human relationships are at risk of being disintermediated by technology.
New ways of thinking about foreign exchange are required to offer a compelling value-added proposition.
Regulation that forces enhanced transparency is also moving financial institutions in the same direction.
The authors of this important book, Demetri Papacostas and Francesco Tonin, have produced a use-focused
guided tour of the FX market, especially from the perspective of asset managers and corporate treasurers. However,
speculators and those who seek alpha from the foreign exchange market will also gain a better understanding
regarding the motivation of the other side of their trades. Speculators take the risk from other speculators and from
commercials (those with an underlying business need, including hedging). They offer a very intuitive explanation of
the Black–Scholes formula, at the heart of option valuation models, which even readers without a mathematics
background will find engaging.
One of the things that make this book a most enjoyable read is that the personal experiences of the authors are
brought to bear to illuminate crucial points through practical examples. Human decision-makers are at the centre of
their story. Ultimately, it is one of people in a fiercely competitive market, evolving in response to a changing
landscape shaped by the implementation of new technologies in the new post-2008 regulatory environment, and
challenged by their moral fortitude. So, sit back, relax and enjoy this journey into the world of foreign exchange. As
a spoiler alert and word of caution, the enthusiasm and passion of the authors may be contagious.

Marc Chandler
Global Head of Currency Strategy
Brown Brothers Harriman
 Abbreviations

AI Artificial intelligence
AOCI Accumulated other comprehensive income
ATM At-the-money
ATM Automated teller machine
ATMF At-the-money forward
AUM Assets under management
BFIX Bloomberg FX fixing rate
BIS Bank for International Settlements
Bp Basis point
CEO Chief executive officer
CFaR Cashflow-at-risk
CFETS China Foreign Exchange Trade System
CFO Chief financial officer
CME Chicago Mercantile Exchange
CV Cumulative value
CVaR Conditional VaR
DTCC Depository Trust & Clearing Corporation
EaR Earnings-at-risk
EBS Electronic Broking Services
ECB European Central Bank
ERM Exchange Rate Mechanism
ETF Exchange-traded fund
Euribor Euro Interbank Offered Rate
FDI Foreign direct investment
FICC Fixed income, currency and commodity
FSB Financial Stability Board
FX Foreign exchange
G10 Group of 10
GAAP Generally accepted accounting principles
GDP Gross domestic product
GFXC Global Foreign Exchange Committee
IOSCO International Organization of Securities Commissions
IPO Initial public offering
ITM In-the-money
iVaR Incremental value-at-risk
KIKO Knock-in and knock-out
KYC Know your customer
Libor London Interbank Offered Rate
LTCM Long-Term Capital Management
M&A Mergers and acquisitions
MARS Multi-Asset Risk System
MiFID Markets in Financial Instruments Directive
MTM Mark-to-market
MVaR Marginal VaR
NAFTA North American Free Trade Agreement
NAV Net asset value
NDF Non-deliverable forward
NEER Nominal effective exchange rate
OIS Overnight indexed swap
OTC Over-the-counter
OTM Out-of-the-money
OWS Occupy Wall Street
P&L Profit and loss
PBOC People’s Bank of China
PM Portfolio manager
PPP Purchasing power parity
QE Quantitative easing
REER Real effective exchange rate
SBC Swiss Bank Corporation
SEC Securities and Exchange Commission
SNB Swiss National Bank
STP Straight-through processing
T-bill Treasury bill
TARN Targeted accrual redemption note
TCA Transaction cost analysis
TWAP Time-weighted average price
VaR Value-at-risk
“There is only one holistic system of systems, one vast and immane, interwoven, interacting, multi-variate, multi-
national dominion of dollars. Petro-dollars, electro-dollars, multi-dollars, reichmarks, rins, rubles, pounds, and
shekels.

… It is the international system of currency which determines the totality of life on this planet. That is the natural
order of things today. That is the atomic and sub-atomic and galactic structure of things today!”
– Arthur Jensen, played by Ned Beatty Network (1976)
 Introduction

This book examines the behaviour of the major players in the FX market at a time of great change brought about by
unprecedented transparency and efficiency. The huge advancement and broad availability of massive computer
processing capability, combined with the globally coordinated imposition of regulation, is allowing a look under the
hood for the first time. This transformation is throwing relationships that have existed for the last 40 years into
turmoil, and causing massive dislocation of people. The FX market is looking more and more like the equity market,
and the rocky path to the promised land of frictionless, fully transparent FX markets is becoming closer than ever.
We examine the impact of this revolution based on real-world situations, and through a detailed explanation of
how these instruments work and how their characteristics impact the behaviour of market participants. Specifically,
the book explores the basic use of FX, and particularly FX options, by the major market participants in a practical
and intuitive way – how and why they mitigate risk, how they profit and how they speculate using FX-related
instruments. The approach is practical and at times minimally quantitative so as to reflect how we’ve taught and
advised these same market participants for the last quarter-century. For those who want to see the calculus behind
the instruments, we make the mathematics available but in no way critical to appreciating the rest of the story.
Topics are addressed from a few perspectives and are revisited multiple times.
The book is designed to be interesting to a wide spectrum of readers, from the person casually involved in
markets, to the person attracted to a deeper understanding of FX and FX options, as well as those who want to
understand the numbers behind the practical. So, if you see technical parts that make you roll your eyes, skip ahead,
we will connect the dots later on.
In the first three chapters that comprise Part I, we examine the forces shaping the FX market today, the changing
character of FX trading, hedging and the tearing down of implicit agreements between hedgers and risk-takers. We
then dive into the details of who is trading what and analyse who the key players are. We also briefly look at the
business of forecasting FX and understand its strengths and limitations. Part I, as well as the rest of the book, offers
a palpable understanding of the culture that existed in FX and how that has dramatically altered. Consequently, we
touch on the scandals that have afflicted the market and excellerated change. These are revisited throughout the book
as they resurface in the more complex structures discussed in Parts II and III.
In Part II, we look to provide more context on the specifics, and run through a practical explanation of the FX
spot, forward and option markets. This is followed by an examination of how options work from a couple of
different perspectives so that the reader can gain an intuitive feel for what these instruments do. We analyse the
contribution of different players and discuss how these players are coping with greater risk. We explore some basic
FX option structures and what they can tell us about the market, pointing out the shortcomings inherent in the
models but also how they are being misused.
Part III contrasts two very different FX market participants. On the one hand, we focus on corporations who in
general are not in it to maximise profit, and hedge funds who are attracted to the FX market specifically because
there is excess profit. Corporations and asset managers represent the biggest sources of inefficiency, so we will
examine why their approach to foreign exchange gives rise to excess profits and why they need to be concerned as
the regulatory and technological landscape evolves. This part concludes with a look at the business and
mathematical models that provide best practice solutions, and shows how to use currency options in a smart and
responsible manner.

WHO IS THIS BOOK FOR?

As mentioned, we hope that elements of this book will be a thought-provoking read for a wide swath of readers. At
one end of the spectrum, the stories and human element that drives behaviour and trading can be useful for the
casual reader interested in capital markets. At the other end, a quantitatively inclined reader who wants to
understand the mathematics behind FX derivatives and the optimisation of risk management with FX options should
find value in the book. It is also intended for those in the middle, the financial professionals with a practical interest
in foreign exchange. They are typically interested in how derivatives can be used to solve daily issues faced by
corporate treasury, portfolio managers and analysts in asset management communities. Hedging FX is mostly
handled internally, very often ignored or farmed out to overlay and hedge fund managers. Consequently, all the
parties involved need to understand how the risk is dealt with by their competitors and partners so they can better
decide what makes sense for their stakeholders.
Those who will find this book useful are the staff of corporate treasury, asset management and speculative
communities, but also their vendors. For example, it will be helpful for bank sales desks, as well as software vendors
that want to provide analytic tools to this community. We believe that what is missing from the market is a
commonsense book that speaks to a huge audience across banks, corporations, hedge funds, asset managers and
high-net-worth individuals. These tens of thousands of market participants understand the fundamentals of foreign
exchange, and are generally aware of the fundamentals of options. For example, if you understand the relation
between a put option on a stock and the relative call option, you also understand how it pertains to foreign exchange,
gold or crude oil. What does not carry as easily across asset classes are the nuanced issues of liquidity, market
participant dynamics and specific instrument constraints.
Foreign Exchange Options and Risk Management recognises the varied reasons different professionals are
interested in the foreign exchange markets. It aims to be a practical guide with many references and connections to
actual events – so, for example, when explaining how to use collars in FX, we will make reference to how the
concept was perverted with insane leverage by Japanese companies. When talking about targeted accrual redemption
notes (TARNs), we will demonstrate how these instruments brought down many companies during the global
financial crisis. When teaching FX options we liken our approach to that of teaching someone to drive. To be a great
driver you do not need to understand the chemical process by which petrol combusts, nor exactly how the internal
combustion engine works. However, you do need to understand a little about the explosions that occur and a lot
about how the steering wheel, accelerator and brake pedals interact to control the vehicle.
Finally, students learning about capital markets will find this book a great resource for understanding all things
FX. Discovering the mysteries of the FX market should work like an hourglass. At first nothing makes sense,
conventions are hard to grasp and everything seems to happen for no reason. Everything is out of focus and very
wide open. Gradually, with study and practice, with thought experiments and discussions with colleagues, things
start to fall into place, and the great symphonic dance of the numbers becomes harmonious. Ideas begin to emerge
from the cacophony of trades. Now the market becomes lively and full of engaging thrillers: monetary policy and
supply of corporate bonds, corporate buybacks and exchange-rate-induced inflation all play together like a perfect
ensemble.
At this point our vision becomes laser sharp and completely focused, and we are at the tightest point of the
hourglass. And then our understanding starts to take us off road. Better questions pop into our head, and often there
are no obvious answers. This happens for two reasons. First, deep understanding always ends with four famous
words: “I do not know”. The issues that we really understand well, those that we have spent a lot of time trying to
dissect, must be those with no answers. The second reason, specific to this book, is time. The industry is changing as
forces that have until now been kept at bay are finally prevailing, and there is no telling what will happen. In the
middle of the book, the reader will learn they can purchase a crystal ball to predict the future, and how to calculate
how much they should pay for it, but there is no crystal ball on sale for the questions that they will find waiting at
the end of the book. At that point, they will find themselves at the other end of the hourglass, in complete ignorance,
and together with us will again utter those words: “I do not know”.
Lastly, a note to the reader. This book has two authors – when the first person is used, it can represent either of
the authors, and it will not be important to distinguish which one. We hope to have blended our individual voices
into a new unique expression. Nevertheless, besides the period-specific references that betray our age difference, it
may be fun for you try to assign an author to the anecdotes.
We welcome our readers’ feedback at [email protected] or via Twitter at @FX_Risk.
 1
A Brief History of the FX Market’s Evolution

Humankind has a very long history of using different currencies to facilitate trade, going as far back as 10,000 years
ago. Our existing system of currency creation and exchange, based on state authority and credibility, is very new. In
this chapter, we will briefly explore the beginnings of this system following WWII, and its evolution into a fiat
system of currencies. We will also highlight the transformative impact on this market of regulation, and the
astonishing increase in computing horsepower.

IN THE BEGINNING
In 1944, at Bretton Woods, the US was about to become the only real economic victor of both world wars. It used its
dominance to bring to the table all relevant countries in its sphere of influence. The US created a global accord that
stabilised the global financial system and secured its economic hegemony for the foreseeable future. At the heart of
this system were a set of rules and ethos that discouraged speculation, and provided the maximum possible certainty
of exchange rates and a mechanism for the massive global debts to become manageable. Global debt was a huge
problem after WWII – for example, US government debt had risen from 20% of GDP in 1933 to 112% of GDP in
1945 (see Figure 1.1), while the UK’s public debt was over 200% by the end of WWII.
The two cornerstone rules of Bretton Woods were that US$35 can be exchanged for one troy ounce of gold, and
that the US dollar was pegged to fixed exchange rates versus other currencies (see Figure 1.2). Trade imbalances
were corrected primarily by gold reserve exchanges. As Japan, Europe, and particularly West Germany, France and
the UK became economic powerhouses again, they were no longer dependent on the US. By 1971, countries were
demanding gold for the dollars they held. In August 1971, President Richard Nixon announced that the US would no
longer exchange dollars for gold, effectively ending the Bretton Woods agreement. What followed was both a sharp
decline of the dollar and a sharp appreciation of the other major currencies, sometimes under chaotic conditions.
 A new powerful force entered the market – FX volatility. Central banks stepped in and acted as the main buffer to
volatility. They often expanded their rationale for intervening to include reaching inflation targets and increasing
their countries’ competitiveness. They used various mechanisms, including the direct purchase and sale of their
currencies versus the US dollar in ways that would try to intimidate the market into submission. The effectiveness of
these actions is debatable, but for the most part what is known as “non-sterilised intervention” has been more
effective than “sterilised intervention”.1 The cat was out of the bag! A period of currency speculation and
uncertainty had become a way of life. This uncertainty transformed how firms and investors managed their
businesses. It made basic planning much more difficult, as an unpredictable factor such as FX had a substantial
impact on the bottom line of anyone dealing with the global markets.
Market participants responded slowly at first in recognising the impact of this new world to the bottom line.
However, within a decade a huge marketplace developed, led mostly by banks, to create tools that mitigated this
risk. The 1970s was a period of rapid change for FX, with attempts to create bands within which currencies could
float. These attempts did not pass the test of time, and all major currency pairs effectively became floating (market
forces driving the exchange rate at any point in time). Spot markets (the actual exchange of one currency for
another, usually settled within two days) exploded, fed by speculation.
 According to the economist Bernard Lietaer, author of The Future of Money, as recently as 1975 roughly 80% of foreign exchange
transactions involved the real trading of a product or a service. The remaining 20% were speculative; bets made on the value of currencies
going up or down – buy it before it rises, dump it before it drops. By the late 1990s that ratio had changed dramatically. In 1997 the
percentage of foreign exchange that involved transactions in the real economy was only 2.5%. Today, the picture is even starker. According
to the Global Policy Forum, in 2011 only 0.6% of foreign exchange could be traced to genuine international trade in goods and services. Of
the rest, a minimum of 80% was directly attributable to exchange rate speculation. The ratio of mud to brick has reversed entirely. (Andreou,
2013)

The first derivative of the spot market, the forward market, also grew, led by the development of future contracts on
the Chicago Mercantile Exchange (CME). There was something very special about the FX market – it quickly
became huge in terms of volume and was the least regulated of global markets, involving a multitude of players with
diverse interests and objectives.

THE FUN BEGINS

In the early 1980s, I worked for Marine Midland Bank (later bought by HSBC) in New York. Marine Midland,
having a long tradition of merchant banking and historic ties to trade finance, was one of the pioneers of exploring
alternative ways its clients could hedge the new risk to their trade flow, generated by the volatility in the FX
markets. In 1982 Scott Dillman, who created and ran an FX option trading desk to complement the bank’s spot and
forward trading, traded the first over-the-counter (OTC) FX option.2 The technology of how to think about, manage
and profit from this derivative of the FX forward market was nascent.
The use of options was not new in the broader market. It had its theoretical underpinnings in the 1973 Black–
Scholes model published in “The Pricing of Options and Corporate Liabilities” (Black and Scholes, 1973). FX was
different; the unregulated, very liquid nature of the asset class pushed the model assumptions to their limits quickly.
The basic assumptions of the model concerning constant volatility across time and strikes was limiting, but the
methodology for hedging an option by buying and selling the underlying currency in the forward market was
sufficient to spur its expansion. By 1984 the more daring corporates were ready to get into the game. Apple
Computer started using options to hedge its short Japanese yen position and, more importantly, was happy to talk
about it to The New York Times (Sterngold, 1984) as a tool that “significantly reduces our foreign exchange risk”,
and saying “They are perfect for giving you peace of mind over movements in the market.”
 As a consequence, a huge business developed for banks in brokering transactions and risk related to anyone
involved in global business. As part of offering hedging alternatives to corporations and investors, banks and other
participants, mostly hedge funds, found the leeway to create lots of opportunity for speculation. The market was ripe
for large profits for the banks because participants such as corporations and asset managers are not driven to
maximise FX profits. Along with limited price transparency, the conditions were perfect for outsized profits relative
to the risks. Corporations in particular are driven to reduce uncertainty (volatility) around cashflow relating to
earnings. In the process, they may be paying large transaction costs and not maximising profits from their FX
exposure (a full discussion of this can be found in Chapter 12).
As transparency and familiarity increased in the basic FX option hedges, banks were very creative in coming up
with new instruments that offered greater flexibility and customisation that mirrored differing market perspectives
and varying risk/reward scenarios. The newer, more complex, instruments also happened to be more opaque and
allowed for outsized rewards for the risk taken. They were fun, heady times; I recall excitedly reading articles by
Scott Dillman such as “Delta: The Option Tool for Hedging and Trading” and then “Life after Delta: Gamma” in the
Euromoney Supplement. Of course, the concepts of delta and gamma trading are still taught and “delta” as a way to
denote change or difference has even entered into everyday speech. In subsequent chapters, we will examine the
creation of these more complex instruments and how they can be used to effectively hedge and manage risk.

TECHNOLOGY AND REGULATION: KEY CATALYSTS FOR CHANGE

The discussion in this book revolves around a set of tools that are designed to work within the floating fiat currency
framework and the conventional understanding of finance and world trade. In some ways, it feels that writing about
these instruments and their use to hedge risks is an exercise in writing history rather than a current set of tools and
activities to mitigate risk. There has been unprecedented and accelerated change since the early 2010s. Change in
market dynamics are hugely influenced by advances in technology and an increasingly global regulatory framework
that is trying to recreate a Bretton Woods-style damper particularly on bank speculation.
Beyond the FX market, we are in the most dramatically transformational period in human history in terms of the
data and information available to every market participant and to every corner of the world. What is equally
transformational is that we are developing technologies that make processing this data into useful information
inexpensive and widely available. When you have two billion people connected with almost cost-free
communication and computer processing power that is massively larger than anything we have ever known, the
potential for unfathomable disruption is a fact of life. The shocks to the system will be foundational. Rapid and
immense changes in technology, communication, computing capability and the processing of incomprehensible
amounts of data lead us to solutions that will be delivered to us by machine-learning algorithms. Human beings will
simply accept the correct answer, and then try to put together theories as to why the machine is right. Imagine an
algorithm that has consistent returns trading FX with a high Sharpe ratio. A speculator is incentivised to implement
first and try to figure out later why it is so successful.
Although some of this change is still not accepted, the tidal wave is about to sweep everybody away: both the taxi
driver stubbornly refusing to use Waze to navigate the unpredictability of Manhattan traffic and the FX spot trader
still thinking that they can “feel” the market, after hundreds of light speed algos have already traded ahead of them.
Beyond the palpable change we suspect we are unable to fully grasp the multi-dimensional disruptions that are
about to overtake us. There are challenges to the very core of our understanding of economic behaviour that will
require us to rethink the foundations of economic orthodoxy. We have accepted, for at least a couple of centuries,
that economic theory should be based on a system that is defined by players acting rationally to maximise their own
interest in a world of scarcity. The most basic economic theories are modelled around supply and demand leading to
an equilibrium price. How do all the hedging techniques change in a world of free and frictionless communication,
close to zero transaction costs and where knowledge is driven by volunteer crowdsourcing? Change is coming at
blinding speed and it is coming outside the traditional models. The English-language Wikipedia had 500,000 articles
in 2005. By 2016, it had over 40 million articles in all language editions, equivalent to 18,000 print volumes of the
Encyclopedia Britannica. It updates in almost real time with a permanent staff of approximately 300 full-time
employees (as of 2018).
 The other edge of the technological sword has forced tremendous reductions in the front-office staff of financial
institutions, while productivity and profitability remains high. According to Bloomberg, front-office staff at the top
12 banks was reduced by 21% between 2011 and 2016 (see Figure 1.3), primarily in fixed income, currency and
commodity (FICC) trading operations.
In addition, the make-up of the remaining FX front-office staff has changed. Below is a more extensive quote
from the Bloomberg article (Nguyen, 2017):
Younger, lower-paid employees make up a greater percentage of trading desks today than they have in years. Part of banks’ broader effort to
cut staff, boost electronic trading and lower costs following the global crisis, the “juniorisation of Wall Street”, as some call it, has been
especially acute in the foreign exchange market. The world’s 12 largest global banks cut front-office staff by about 25% in Group-of-10
currency markets over the past four years, according to Coalition Development Ltd. That’s coincided with a shift to automation, which
slashed staffing needs and spawned a new, and small, generation of quantitative traders whose decisions are driven by mathematical models.
For every managing director with about 10 years or more on the job, there are as many as seven less-experienced staffers on currency desks,
Coalition said. The ratio was one-to-four just five years ago.

Similarly, bank back offices, which employ the most staff at banks, are on the verge of a massive reduction as banks
look to cut costs. Technology offers unprecedented opportunities with more conventional strategies, such as
increasing the straight-through processing (STP) of transactions and by the revolutionary application of blockchain
technology. Blockchain in particular threatens to transform how all back-office transactions are conducted,
confirmed and settled. According to Accenture,3 blockchain technology could reduce infrastructure costs for eight of
the world’s 10 largest investment banks by an average of 30%, translating to US$8–12 billion in annual cost savings
for those banks (Accenture, 2017). The main cost for the banks is data reconciliation, which is a labour-intensive,
error-fraught process. Blockchain technologies promise to carry this out for a negligible cost and significant
improvement in data quality. Let’s do some back-of-the-envelope calculations to understand the magnitude of these
potential changes. The company review website Glassdoor has found that in New York City the average back-office
salary is approximately US$35,000 a year. Therefore, if only US$8 billion is saved, mostly in salaries, that means
around an additional 230,000 people will lose their jobs.
The other key catalyst of change that will have an impact sooner rather than later is the regulatory tsunami hitting
global markets. Regulators are determined to force automation, transparency and uniformity globally. This will
initially promote activities that are obvious but nevertheless transformative. For example, there are similar Excel-
based and proprietary systems in every corner of banking and hedge fund operations, and every corporate treasury
area. They will all be replaced with comprehensive STP systems that require little or no human interaction. Consider
the following three activities.

Smaller asset managers are used to sending an Excel spreadsheet or email to the transacting bank to allocate an FX
trade to different funds.
Hedge funds may send a stack of documents separately to each bank to open a new account to conform to know-
your-customer (KYC) requirements.
Finally, the maintaining of collateral by counterparties that relate to short-term money market trades or more
complex derivative trades is a sea of daily emails and faxes that need to be checked against contractual
obligations to determine the appropriateness of collateral.

All three activities, which require armies of staff both on the bank/dealer and the client side, are now required by
Dodd–Frank4 to be fully automated. One can come up with many reasons this pace of change could be delayed or
reversed. Entrenched interests may be able to block or usurp advancement, or the world can retreat back into a
nationalistic, isolationist model. After all, humankind has also advanced the ability to send us all back to the Middle
Ages! We have more powerful weapons, potentially accessible by many, and an unprecedented ability to deploy
them remotely (not that humans ever had problems destroying each other close up). Of course, we have also
developed cyber-weapons that can bring any financial market to its knees. Assuming we discover our better angels
and continue down the path of using technology and resource abundance to benefit all, we may in fact be facing a
new world in very short order.

CONCLUSION
A trade-focused system, created by (and some would say for) the US, bound the developed world together from
WWII to 1971. As the system succeeded in making its participants a lot richer, it came under pressure by those very
same participants, which were no longer happy to play second fiddle. It was forced to evolve into a more flexible,
less predictable world, where interest rate policy and controlled currency moves were instrumental to lubricate the
new relationships. As is always the case when we experiment in this way, there were unexpected consequences. The
FX lubricant took on a life of its own, and became so large and important that it very often is the tail that wags the
dog. Add to the mix a transformative time in human technological capacity, and the picture of what a future FX
landscape looks like is very volatile and unclear. Regulators globally are not happy with the situation – not only in
FX, but across all capital market activity. The most powerful economic states are determined to cap and control this
speculative behemoth. Their efforts are having a serious impact on all participants and market activity. In the next
chapter, we will examine the size and character of the FX market.

1 Intervention is when the central bank buys or sells its own currency. The intervention is “sterilised” when this is accompanied by, respectively,
purchases or sales of financial assets. Sterilisation is done so as not to alter the money supply, which is otherwise impacted by the purchase or sale of
the currency.
2 Over-the-counter contracts are private deals between counterparties which are legally binding but are not registered with any institution, and until
Dodd–Frank, not to be publicly disclosed.
3 Accenture website, accessed 17/01/2017. https://newsroom.accenture.com/news/blockchain-technology-could-reduce-investment-banks-infrastructure-
costs-by-30-percent-according-to-accenture-report.htm
4 The Dodd–Frank Wall Street Reform and Consumer Protection Act (Pub.L. 111–203, H.R. 4173).
 2
Foreign Exchange Markets

For those not involved in the FX market, it must be very difficult to get a palpable sense of what it is. Exchanging
one currency for another happens when you go to the bank or the automated teller machine (ATM) and exchange
money for your vacation; it happens on the second floor of some shopping mall in Singapore by some guy in a kiosk
– no computer, and he makes a tighter price than JPMorgan; it happens on the fancier streets of Buenos Aires, where
every five steps someone is yelling “cambio”; it happens when IBM sells its artificial intelligence (AI) services to a
Chinese municipality; and it happens when the Federal Reserve Bank buys and sells currency to manipulate FX rates
and implement interest rate policy.
FX is both omnipresent and a self-perpetuating creation. Although it permeates every aspect of global
interactions, the heart of the system is really a market concentrated into a few major players, located in a few central
locations and trading just a handful of currencies. In this chapter, we will explore who trades FX, where it happens,
how big the market is, and look at just a few of the conventions and traditions. Some of these qualities have led to
behaviour that has caused soul searching by market participants, the public at large and the regulatory authorities.

THE WHO, WHAT AND WHERE OF FX

Foreign exchange is the lifeblood of global trade. Although the goods and services trade component may be a small
part of the total, in terms of volume transacted, global trade is very sensitive to the current level, potential changes
and liquidity of the FX markets. There is no corner of the world that is not impacted by FX. Even where there is a
global understanding that some transactions will be conducted in a common currency, typically in US dollars, there
are further FX transactions. Consider, for example, crude oil transactions that are denominated in US dollars. When
Saudi Arabia sells crude to a European country in exchange for dollars, the Saudis will need to invest their dollars
somewhere – and if it is not entirely in the US, they will need to sell their dollars and buy other currencies. The
European country will have had to give up euros to acquire dollars to purchase the oil, and then will most likely
need to refine it in another country, which may require yet another FX transaction.
These transactions happen seamlessly because FX markets are the most liquid of all financial markets. In the
developed world, FX has mostly been free of regulation, or rather effective at self-regulating, and dominated by a
handful of global banks. Consequently, there was no definitive information on the size and details of the market.
Each bank jealously guarded this information and, after regulatory scrutiny, many fines and some jail terms for
collusion (as described later in this chapter), the market is now in the midst of trying to figure out how to share
information beyond what they are required to share by the user-owned and directed clearers, such the Depository
Trust & Clearing Corporation (DTCC). There used to be a dearth of detailed information on the market players,
liquidity and the real market price, which undoubtedly gave a competitive advantage to the biggest players. The
salespeople at the largest banks saw most of the volume from the biggest market participants. They used this
information, first and foremost, to earn a higher spread, and secondly to give it selectively to their best clients so that
they could execute at a better price.
However, this fog has lifted considerably for certain areas. For example, most participants, with a little badgering
to their bank counterpart, are now able to get a price that is reflective of the best price in the market given their
credit quality. Nevertheless, there is a long way to go in terms of getting a total market view like we have in equities.
There is just one systematic view into the opaqueness of the total market. The Bank for International Settlements
(BIS), an institution formed in 1930, is the central bank for the world’s central banks. Every three years it conducts
an extensive global study of the FX markets by asking its member central banks to survey the FX participants in
their jurisdiction. The last available study was released in 2016 (BIS, 2016a), and the following description and
charts of the FX markets are taken from that study.
A HUGE MARKET
The FX market is massive, and the US dollar continues to dominate (see Figure 2.1). Consider some of the
highlights from the BIS report below.

Trading in FX markets averaged US$5.1 trillion per day.

Spot transactions were US$1.7 trillion per day.
FX swaps (described in the next chapter) reached US$2.4 trillion per day.

To many market participants, it may come as a surprise that most of the transactions in the market are swaps and not
FX spot deals. The other interesting item is the relatively small turnover in FX options. You may wonder why we
are devoting a significant portion of this book to such a small part of the story. Although options are a small part of
the total volume, each option transaction has a large multiplier effect on spot and forward trading. Furthermore,
options offer such a refined tool to both hedge and speculate with great subplots that make working with them fun
and interesting.

FX MOSTLY TRADES IN EUROPE

Most FX trading happens in Europe, specifically in the UK. London has dominated the physical location with the
deepest markets. The reasons for that have to do with time zones and history. The UK time zone as it relates to Asia
and the Americas is key. The UK nicely allows Asian dealers to stay late and US dealers to come in early during
London’s morning and mid-day. A market that was so closely held by a few institutions relied on the personal touch.
Traders knew each other across the world. They talked constantly on the telephone, Reuters dealing and on
Bloomberg IB during their day and from their homes in the evenings. Traders established reputations and trust that
carried across borders and institutions. This has been a self-perpetuating cycle in a world where relationships and
voice deals were crucial.
The other reason has to do with the historic domination of UK trade across the world in the last couple of
centuries, and the more laissez-faire attitude of the UK authorities. UK leadership in the FX markets goes back to
the mid-1800s. Dealers still refer to the UK pound as “cable” or “caibo”. UK spot dealers were more likely to come
from East London than Eton. When I first got to the trading floor, I was “bubble and squeak” – Cockney rhyming
slang for Greek. Their presence across the trading floors was legendary. “Cable” refers to the very first transatlantic
cable laid in 1858 and the first FX trades via the cable, soon thereafter, between the US dollar and the UK pound.
 The perception that you would get a more liquid market and a better price during London hours was dogma. I
recall for decades a large California computer/printer company would come in at 3am to hedge their corporate
exposure. No matter how hard I tried to convince them that, given the very inefficient way they decided what and
when to trade, and the spread they paid because they lacked transparency, they should just do it during regular New
York hours. However, someone in treasury had heard that London was more liquid and that was that for years and
years.
London kept its primacy through thick and thin, rising from 29% market share in 1995 to around 37% share
today. It will be interesting to see how this evolves following Brexit, the momentous move to electronic trading and
the use of algorithms described in the coming chapters. Many large banks have announced plans to move hundreds
of jobs out of the UK as a result of Brexit, but it is unclear if this includes FX dealers and salespeople (Martin and
Finch, 2017).

IT HAPPENS IN THE UK, BUT THE US DOLLAR AND US BANKS RULE

The US dollar is the conduit for most transactions around the world – it is the preeminent vehicle currency, being on
one side of 88% of all trades! There are many reasons for this, the most important being the credibility of the US
economic system and the extension of that credibility to the dollar. This causes many countries to hold significant
amounts of dollars as a reserve in their national accounts (see Table 2.1). In addition, some countries unofficially
allow the use of the US dollar as legal tender, while many commodities (such as crude oil) are in fact priced and
exchanged in US dollars.
Shifts in volume can be great validators of trade shifts and financial strength of various countries around the
world. For example, look at the following shifts since the 2013 BIS survey.

The euro, Japanese yen and Australian dollar all lost market share, while many emerging market currencies
increased their share.
The Chinese renminbi doubled its share, to 4%, to become the world’s eighth most actively traded currency and the
most actively traded emerging market currency, overtaking the Mexican peso. The rise in the share of renminbi
was due primarily to the increase in trading against the US dollar. In April 2016, as much as 95% of renminbi
trading volume was against the US dollar. Despite the increase in renminbi’s share, it is still a relatively small
 percentage for such a large trading partner. However, it is expected to continue to grow as the renminbi
internationalises.

Figure 2.3 shows the key groups that trade FX, as identified by the BIS. Note that:

the share of trading between reporting dealers grew over the three-year period, accounting for 42% of turnover in
April 2016;
banks other than reporting dealers accounted for a further 11.2% (51% × 22%) of turnover;
institutional investors were the third-largest group of counterparties in FX markets, at 16%;
in April 2016, sales desks in five countries – the UK, the US, Singapore, Hong Kong SAR and Japan –
intermediated 77% of FX trading, up from 75% in April 2013 and 71% in April 2010; and
non-financial institutions accounted for only 7% of the total volume.
While we can get some broad understanding of this FX trading world, we still do not have full transparency. There
are many problems with drawing nuanced conclusions. For example, there is double-counting of the institutional
investors and non-financial customers (such as corporations). Nevertheless, the above breakdown is fascinating.
Note the dominant position banks have on the overall volume, and the relatively small part of non-financial
institutions, which we can assume includes hedges related to trade. When market power is concentrated in the hands
of a limited number of players, the potential for conflicts of interest is clear.

FX FIXINGS: A STORM IN A TEACUP?

The one subject that has brought the topic of FX to the attention of the wider community is FX fixings. Suddenly,
friends and family who are not in finance are expressing an opinion on the topic. Popular newspapers were full of
stories about market manipulation and the size of the FX market in general. So what are FX fixings, and how did
they make the headlines?
When a client agrees to an FX deal with their banker, ordinarily they agree on a rate (price) at the time they
execute the trade. The client contacts their bank (usually by telephone or electronically) and requests a price; they
then have the choice to accept or decline the price quoted to them by the bank. There is no “official” or centralised
market price due to the over-the-counter nature of the FX markets. Unlike exchange-traded markets, there is little
transparency. It is a transaction of two willing counterparties, and the rate at which they dealt is not disclosed
externally. For reasons that we will go on to discuss, clients began to demand more – in particular, the ability to be
able to deal at a more visible rate (see Panel 2.1).

PANEL 2.1 WHO USES FX BENCHMARKS?

Despite recommendations by the International Organization of Securities Commissions (IOSCO) and the
Financial Stability Board (FSB), as well as subsequent moves by several central banks away from providing
FX reference rates, end-users still have a real need for fixings.

Corporates: FX fixings can provide an independent source for determining price and contract values,
 fixing tax liabilities, providing the rate for a mergers-and-acquisitions (M&A) transaction or valuing
balance-sheet assets and liabilities for corporate treasuries and their accounting departments.
Asset managers: FX fixings are key for asset managers as they are used as a benchmark for FX
transactions and the valuation of multi-asset-class portfolios. For example, index fund managers who
need to buy stocks in a foreign currency may have to buy the foreign currency at a benchmark rate
stipulated in the rules of the index. Similarly, holders of foreign assets may need to translate foreign
currency gains or losses back into their home currency.

For some years, many central banks have published reference rates for their own currencies against numerous
others; often, these were just snapshots of a quote from screen at a set point in time each day (eg, 12 noon).
However, the universe of currency reference rates published by any one central bank could be limited, and central
banks tended to publish rates at different times from one another. As a result, commercial organisations began
offering “fixings”, which took place at known times and according to published methodologies (see Panel 2.2).
Importantly, such rates were then published and accessible not just to the two counterparties, but to the entire
market. End-users would inform their bank’s FX salesperson, ahead of the fixing time, how much of which
currencies they needed to buy or sell “at the fix” – ie, at the fixing rate for each respective currency pair. The FX
salesperson and the client would agree any margin upfront, which would be added to the fix. The FX spot trader
would then work out each day the net position of what they needed to trade on behalf of all of their clients. As a
result, a client eventually had a choice of fixings to deal at.

PANEL 2.2 THE FIX

What’s the fix?

A fix is a number calculated and published by a calculation agent, who is usually not a bank or market
participant to avoid conflicts of interest. This number is supposed to represent the true market value of that
currency pair at a specific point in time.

How is the Fix calculated?

An arm’s-length calculation agent is needed so that the bank could not claim an unreasonably high rate to
deliver the currency it promised. A fix is a convenient method to facilitate business. To calculate it, the
calculation agent usually polls a number of dealers for quotes, takes the mid-point of each of these quotes,
throws out a few of the higher and lower ones and then takes the average. Often the rate is not just taken at a
point in time, but is recorded over a few minutes before and after the hour, and all these observations are then
averaged based on some algorithm. The simplest of these is averaging over time. To more faithfully represent
the market flows, at about the fixing time such averaging can be volume weighted. To achieve that
accurately, you would have to know the volume traded at every millisecond in the entire market for that
currency pair, which nobody knows.
The solution to volume-weighting the fix is to use the volumes available on just one trading venue, which
then is able to calculate their own fix. Remember that there is no ticker tape in FX, and the market is
fragmented in very many trading venues. As a consequence, volume data is only available from a few
sources, and therefore volume-weighted fixings can only be calculated by those with access to those sources.
A second complication arises with the volume-weighted fixings themselves. Suppose you are a market
maker at a bank and a client asks you to deliver to them UK sterling, and to use a volume-weighted fixing
from a particular source. In other words you will agree to deliver the currency to them at a rate equal to a
future fixing! How can you be sure that you will be able to procure the currency at a rate at least close to
what the fixing rate will be? You will have to start buying the pounds that you will then deliver to the client
at the beginning of the time period that the fixing comprises, but it will not be enough for you to buy the
same amount of pounds every so many seconds. That would work for a time-weighted fixing. However, for a
volume-weighted fixing there is a problem – for the average cost you pay to buy the pounds to match the fix,
you would have to know in advance what proportion of the volume during the fixing interval is being traded
at each second. Of course, you cannot predict the future.
Given that you cannot predict the future, you might then decide to react to the past, and buy more right
after the volume increases and buy less right after the volume decreases. But, here is the funny fact – most of
 the time you will not be given volume information by any trading venue or the trading venue that will
publish the fixing.

Where’s the money?

We have just scratched the surface of how complicated and circular this process can get. From the above, one
can understand that offering clients to trade at the fixing rate is risky, and it is therefore understandable that
the bank will want to negotiate to deliver the pounds at the fixing rate plus a fee. Here is the crucial point.
The fanatical attention to gaining market share has driven banks to lower those fees over the years to the
point where this business does not make sense. Unless … There are a few other ways that a bank might have
to squeeze a few bucks off this fixing business. We will look at these in later chapters when discussing the
dynamics of the trading floor and the misunderstood practice of front-running.

So, how did this common practice, which started as a mechanism to increase FX rate transparency, begin to unravel?
It is important to understand the landscape at the time. Prior to FX fixings grabbing the headlines, the London
Interbank Offered Rate (Libor) fix scandal turned everyone’s attention to the market manipulation that was
occurring in the interbank interest rate market. Large fines were levied for breaches of market conduct, which
proved to be only the beginning. While the Libor fixings were sometimes moved by fractions of a basis point due to
inaccurate submissions, the fallout was catastrophic. Billion of dollars of fines were levied, bankers were jailed and
the general public lost any trust they had in the banks. This was made personal as mainstream news journalists
explained how people’s mortgages had been affected. Ironically, borrowers often paid less as Libor actually fixed
lower than it otherwise would have. However, the market’s credibility evaporated. Even banks that voluntarily
disclosed wrongdoing were not shown any leniency, with senior executives at numerous firms losing their positions.
The regulators made it clear that this would not be tolerated. To put the news in context, the week after the first
Libor-related fines were levied (US$453 million), a large pharmaceutical company was fined US$3 billion to settle
fraud and bribery charges relating to drugs that could have had a fatal effect. However, this led to far fewer
newspaper column inches than the Libor scandal.
Almost concurrently, a similar situation to the Libor scandal was unfolding relating to FX benchmark fixes. It
emerged that FX traders had also been artificially influencing where certain FX benchmark rates were fixed. In
addition, client information was being shared, something that every FX salesperson had been taught not to do. When
the story broke, the public reaction was likely bigger than it otherwise would have been had the Libor scandal not
occurred. Again, record fines were imposed by the regulators and traders were criminally charged (see the next
section for details). While the FX traders may have had zero knowledge of what their Libor-manipulating colleagues
had been up to, the public had had enough and everyone was tarred with the same brush.
A long way from a storm in a teacup, the FX-fixing scandal left the reputation of bankers generally in tatters and
at the same time saw the largest set of regulations ever imposed on the once self-regulated FX markets. The US
Dodd–Frank regulations now required that all “US persons” report all non-deliverable forwards trades and FX
options trades to a repository within 15 minutes of trading; the FSB published a “Global Code of Conduct”1 for FX
market participants and the IOSCO published recommendations specifically pertaining to FX benchmarks. This
prompted several central banks to exit the provision of reference rates to the market (see Figure 2.4).
So how do fixings work now? The fixing scandal has not reduced the need for both corporations and asset
managers to have access to independently set FX benchmark rates. They are increasingly turning to data vendors to
provide benchmarks while ensuring that the following requirements are met in any choice of benchmark rate.

The benchmark should be IOSCO-compliant. This means that the benchmark’s methodology has been vetted and
conforms to IOSCO’s strict standards. This could include sign-off by a third-party independent auditor.
The choice of benchmark ensures the ability to access optimal liquidity in the currency pairs the user has exposure
to. This is particularly important if the corporation or asset manager has exposure to emerging markets
currencies, which traditionally suffer decreases in liquidity when their home market is closed. One such example
is choosing a fix with multiple timestamps throughout the day, which allows the corporate or asset manager to
request a fix for Asian currencies during the Asian time zone.
The rate uses an acceptable methodology. One such example is a time-weighted average price (TWAP) over a
period. This is a further consideration to reduce the likelihood that rates are susceptible to irregular price action
due to abnormal trading activity in the market at a single fixing time.
Importantly, such methodologies should maintain the anonymity of price sources and ensure the rotation of those
 sources so as to widen the scope of the market from which the benchmark rate is derived.
In addition, some jurisdictions (notably London and Singapore) have regulated certain benchmarks, and require
both benchmark administrators, and submitters to benchmarks, to obtain specific licensing for these activities.

Figure 2.4 shows the Bloomberg FX fixing rate (called BFIX) for the US dollar against Thai baht fixing at 30-
minute intervals throughout the day in local time.

CULTURE, COLLUSION, COLLISION, FINES AND JAIL TERMS, MORE

TRANSPARENCY
Most of us do not find FX markets relevant to our daily existence. The truth is their efficient functioning, and as
importantly their credibility as a fair marketplace, is key to global trade, employment, the level of interest rates and
even geopolitical considerations that affect decisions concerning war and peace. At this point, we need to expand on
the upheaval, referred to above, that has rocked these markets following the great recession (also known as the
global financial crisis) of 2007–08. Currency market participants, along with their fixed-income brethren, have been
accused of attempted manipulation of market prices and key benchmarks, misuse of confidential information,
opaque pricing to take advantage of their clients and collusion between dealers. Large fines have been levied against
almost every major dealer (see Table 2.2). More fines have been levied after 2015, but this gives the reader a sense
of the breadth of institutions involved. The Boston Consulting Group estimates that from 2008 through to 2016,
financial penalties to financial institutions are roughly US$321 billion.2
Furthermore, it has been reported that “Citicorp, JPMorgan Chase & Co, Barclays Plc and Royal Bank of
Scotland Plc agreed to plead guilty to felony charges of conspiring to manipulate the price of US dollars and euros,
according to settlements announced by the Justice Department in Washington” (McLaughlin, 2015). As part of an
overall US$5.8 billion settlement with multiple regulators and based on a recommendation from the government,
they agreed to pay the roughly US$2.5 billion to the Justice Department. In handing down the penalties at a
sentencing hearing, the judge said “when the market is rigged, folks who play by the rules are suckers”. He went on
to say: “Frankly, I would encourage the government to consider prosecution of individuals.” Citigroup will pay
US$925 million, Barclays US$650 million, JPMorgan US$550 million and RBS US$395 million and UBS Group
will pay US$203 million. UBS had received immunity from prosecution in the currency case because of its
cooperation with authorities. However, the bank’s misdeeds in that case were found by prosecutors to have breached
an earlier agreement over its role in manipulating benchmark interest rates (Larson, Schoenberg and Dolmetsch,
2017).

What is going on here? We are talking about those foundational financial institutions. The same institutions that
used to reside in buildings that look like the Parthenon and where the very air was meant to connote confidence and
integrity.3 From speaking to many ex-colleagues, and frankly having participated in the very same market
environment, it can all feel very unfair from an individual trader/salesperson perspective. This behaviour has been
going on for decades. Everyone who participated knew the game and how it was played. We find it difficult to
reconcile the different views of the world between the dealers trading and selling in the FX markets every day and
the expectations and perceptions of someone outside this world.
As noted, these opaque markets have operated as a closed self-regulating and self-serving system for decades. The
major banks have made hundreds of billions trading FX and provided a platform that facilitated global trade with
significant stability, even during the collapse of most other markets in 2008. The individuals involved in these
markets were driven first and foremost by their employer’s promise to share some of the billions with those
individuals that could outsmart the other players. At the ground level it was a game. Many people that came to these
markets made seven figure bonuses by outmanoeuvering another trader, and expediting unique corporate deals that
related to trade transactions or cross-border mergers. It was difficult to see the harm at the ground level, and the
institutions were geared to sanitise the whole thing.
At other times one felt a ting of guilt. Dealing with philanthropic, selfless organisations, such as Doctors Without
Borders, you realised that your profit came out of donations. We were able to soothe our conscience by taking a
little less profit than we would from a major multinational. For the most part, I suspect most dealers that were
charged and arrested over these scandals are still scratching their heads, muttering how “everybody did this and
every single bank head knew exactly what was going on”. I suspect they are in complete disbelief at the perceived
double-standard, at the retroactive and uncustomary interpretation of the law. They are incredulous that a bunch of
bureaucrats who know nothing of their world are standing in judgement. There was an editorial written in 2012
(Smith, 2012) that does a good job of highlighting how client transactions were viewed. This was not just Goldman
Sachs, and the author goes over the top in painting himself as a victim of circumstance, but the ethos of the trading
floor is accurate. In Chapter 5, we will revisit the topic of trading floor dynamics, and examine the type of activities
and compensation schemes that led to these behaviours. However, we do not intend to re-litigate these issues here.
The intent is to acknowledge their existence and recognise their effect.
The huge fines, combined with the reputational damage, have led to a total re-evaluation of the business from
within and without. Regulators who in the past stayed away from interfering in the FX markets are poised to force
transparency similar to that of the equity markets and, most importantly, they want to clamp down on the ability of
banks to speculate through the introduction of the Volcker rule.4 New rules of conduct have been promoted by the
BIS (BIS, 2016b) and industry groups, all major central banks have commissioned studies to understand what is
proper conduct for these markets (Bank of England, 2015) and all major dealers have clamped down on the culture
that spawned these fines.
Here are some of the most obvious consequences. Bank compliance departments, sales and trading desks are
petrified at being linked to any collusive behaviour. Getting huge margins on any one deal in the past used to get
you a commensurate bonus; today it gets you an audience with compliance. Compensation in general has decreased
for the frontline staff, and the banks have made huge investments in pushing all volume through electronic forums.
Liquidity moved away from the major banks and is now being offered by non-bank counterparts such as hedge
funds and risk-loving asset managers.

CONCLUSION
The FX market is one of the largest, most liquid, most stable, most decentralised, least regulated (given its
importance) markets in the world. Its impact is felt across borders (by definition), fiscal and monetary policy, and
even spills into geopolitical considerations. What is stunning, given all these superlatives, is that it is concentrated
and driven by a few major banks and is geographically centred in London and New York. The bulk of the
transactions are speculative, revolving around a small core of fundamentally driven activity arising from trade and
hedging investments in overseas assets. The global financial crisis of 2007–08, and the hunt for explanations,
culprits, abusers and solutions in setting interest rates, swept FX in its wake.
Benchmarking FX was one key activity for banks making money in FX, but a closer look into that process led to
shedding light into the broader FX market issues of concentration and participant conduct. The closer look revealed
some ugly warts in how bank clients and less-connected participants were treated. That has driven all concerned to
revamp the entire system. Using technology and legislation, the FX market has become significantly more
transparent, more efficient, less profitable and possibly even more concentrated.

1 See http://www.globalfxc.org/fx_global_code.htm
2 https://www.bcg.com/publications/2017/financial-institutions-growth-global-risk-2017staying-course-banking.aspx
3 Remember the famous anecdote that the marble in the lobby of one of those buildings was sourced so as to perfectly mimic the texture and colour of
dollar banknotes.
4 The famous Volker Rule, being one of the most notable pieces of the Dodd–Frank act, establishes that the sales and trading operation of a bank shall
not engage in speculative trading using the banks money.
 3
Predicting FX Movement

Forecasting FX is a critical preoccupation of all FX market participants. Whether you are a spot dealer concerned
over the next 10 seconds or a corporate chief financial officer (CFO) deciding whether you want to build a plant in
Australia that will have a 20-year productive life, your actions will be contingent on your expectations of future FX
rates. This chapter will not try to explore with any rigour the spectrum of approaches, nor evaluate the best way to
predict FX. We will quickly summarise some of the more common approaches, and then focus on the available tools
used by the main players. There are countless books in print that delve into the topic from different perspectives,
including the classic on exchange rate determination by our colleague, Mike Rosenberg (Rosenberg, 2003).
Following Mike’s approach, we can begin to break down predictive movements for defined future periods: short,
medium and long term.
Short- to medium-term movements include intraday and multiday forecasts up to a few months. The most
common approach used by traders is to look at charts and apply one of a myriad different studies and methodologies
to predict what the market will do. This is part of what is called “technical analysis”. It is an attempt to predict future
price movements by looking at past prices. This is an immense area of market focus with decades (maybe centuries)
of research. For our purposes, suffice it to say that most market participants will turn to technical analysis and charts
as the first knee-jerk reaction to understanding where FX is going.
Technical analysts will be furious with us for allocating such a small part of the book to their topic, and for saying
that it is mostly used for short-term, and maybe medium-term, predictions. Many studies have been developed to
predict long-term movements over many years (see Figure 3.1). On various software platforms you can see
extensive lists of technical analysis charts. Most readers will already have an idea of how technical analysis works,
but, for the non-initiated, let us look at a couple of very basic examples. The first example is a moving-average
study. You compute the average of the exchange rate during the (insert your magic number) 55 days prior. This
gives you a moving average. If the moving average is below the current exchange rate but then it jumps above it,
this is meant to be a bearish signal. The second example is a Fibonacci level. If the rate rose but now it is retracing
that move, when the rate has reached a fall equal to 61.8% of the original up move, it is said to have reached the
Fibonacci retracement. At this point, a rebound higher is a bullish signal, but a continuation of the retracement is a
bearish signal. The number 61.8% is supposed to be significant as it approximates the Golden Ratio (for an entire
book on the golden ratio, consult Atalay, 2006). The claim of technical analysis is that both the moving-average rule
and the Fibonacci retracement are universal rules of variation in natural phenomena and this is why they are
successful in predicting rate changes. It is the authors’ personal opinion that there is no logical foundation to these
arguments. The academic evidence about the effectiveness of technical analysis in predicting rates changes is
equally nonexistent. We will need an extra server to handle all of the hate emails following the above comments!
 Another way to predict FX movements is to have strategists typically work for banks analyse and digest massive
amounts of information. They typically examine absolute and relative economic growth, as well as other
macroeconomic factors, to determine what will move the exchange rate, in which direction and how fast. These
strategists (usually with a PhD in economics) will then add to the mix relative interest rate levels and projections of
interest rate movements, before considering flows from one country to the next. They look at trade flows,
investment flows, mergers and acquisitions. Finally, they will consider long-term, very often theoretical,
considerations having to do with purchasing power parity (PPP) and real effective exchange rates (REERs). Add all
this into the mix and you will know exactly where FX rates are going – easy, right?
In this section, we will try something much less grandiose and focus on the practical tools used by market
participants to understand what the market and all these brilliant technical analysts and strategists are saying
collectively. We will examine the outlets of the collective output and how the market participants use them. This
will include tools to assess the strategists directly, some analytical tools to examine PPP and REER for yourself.
Then we will focus on market activity, specifically reflected in:

investment flows and future contracts;

option pricing;
various exchange-rate levels made special by option activity; and
in a later chapter, we examine the curious case of the Chinese renminbi forward, which for a while did not reflect
interest-rate differentials but reflected market sentiment of where the currency was going.

FORECASTING FX: PRACTICAL CONSIDERATIONS

If the introduction to this chapter did not convince you, take our word for it, forecasting FX is notoriously
challenging. The main fountain of knowledge when it comes to predicting FX is the banks – usually large global
massive bank desks and their group of strategists and chief economists. However, there is often a disconnect
between what the end-user requires and what the sell-side banks are able to deliver. In particular, some asset
managers and industries with long lead times within the corporate sector require long-term forecasts to plan their
hedging and investment strategies. Banks are rightly reluctant to forecast out beyond two years, and yet a currency
upward and downward cycle can be significantly longer than that. It is not unheard of for banks’ corporate
salespeople to be told to determine their own forecasts when a client asks for a multi-year timeframe.
It can be anyone’s guess where a currency is going given the multitude of factors that can influence it over time.
We also need to be mindful of who is actually making the forecast and what factors they are taking into account. In
FX non-public information is virtually useless in forecasting, and there is an unimaginable amount of public
information available. Contrast this with single-stock equity research, where for small cap stocks there is very little
public information, but any non-public information is determinant to the forecast. The amount of information that
could conceivably be useful to the forecasting process is quite limited. This is the reason why forecasting FX is way
more difficult than forecasting stock prices. As most forecasters are trained in macroeconomics, their forecasts are
predominantly influenced by the economic health of each country as measured by macroeconomic indicators. As a
result, one factor that very few such forecasters take into account is market sentiment. One reason that FX is so
difficult to forecast is that the human mind is wired to find narratives, and experiences great difficulty in aggregating
a large number of factors into one resulting effect. This is one of the themes that will haunt you throughout the book:
what the mind believes, has nothing to do with the evidence in front of its eyes.
Let’s take the Japanese yen as an example. After the unfortunate events of the 2011 Japanese earthquake, the FX
markets traded more off market sentiment than it did off economic fundamentals. Any natural disaster is inherently
costly for a country – it consumes many resources to rebuild and the event is likely to have an impact on export
capacity. Economics training tells us this is bad for Japan, so the yen should fall. However, there is an opposite
factor at play. The Japanese yen is considered a “safe haven” currency – that is, a currency that the markets flock to
in times of world tension. In the early days following the disaster, the markets were confused and it was unclear
which factor would have the greater influence. Eventually, the yen began to rise, relentlessly once the herd caught
on. A strong currency, which would have impeded its export sector, was the last thing that Japan needed while it was
trying to get back on its feet. The yen is not alone; the challenge at any particular point in time is to realise which
factor has the most influence over the currencies that end-users are exposed to, and that each single currency in a
portfolio will not be influenced in the same way. Taking into account only economic fundamentals – as is often seen
– may involve only half of the picture.

The Bank of Japan intervention to sell Japanese yen, which was eventually aided by other Group of 10 (G10)
central banks across the world, was the first concerted G10 intervention in over a decade. This highlights another
difficulty in predicting the FX markets: central bank meddling based on geopolitical considerations of the moment.
Central bank intervention to influence currency levels has a long storied history, which we will not address here;
suffice it to say billions have been made and lost as a consequence. Probably the most infamous episode concerns
George Soros’s battle royal against the Bank of England in 1992. This series of events made Soros billions and put
his name on the map. Very briefly, on September 16, 1992, speculative FX market players took on the Bank of
England interventionist policy and the speculators won. Speculators “broke the Bank of England”. In a revealing
moment about the importance of the FX market, the UK was forced to withdraw from the government-approved
ranges for FX, a system called European Exchange Rate Mechanism (ERM), which it had belatedly joined in
autumn 1990.
The UK economy, which had low interest rates and high inflation, diverged significantly from the West German
economy. Speculators noticing the inconsistent requirements of the ERM and the reality on the ground began
challenging the ERM range. The Bank of England, feeling the pressure, increased its interest rates to the teens to
attract assets to the pound, but speculators, George Soros among them, began heavy shorting the currency. The UK
lost that fight and withdrew from the ERM at a cost of billions in trying to keep the pound strong. The exchange rate
went from US$2/£ to US$1.5/£ in a matter of months (see Figure 3.2). Soros supposedly made somewhere between
US$1 and US$2 billion on the deal, and a name as the top currency speculator in the world.

LONGER-TERM PLANNING
What does an asset manager think of FX risk? Unless they trade currencies as a separate asset class (which is rare),
currency trading is undertaken to immunise the underlying investments from currency risk. An asset manager puts
on a currency trade to facilitate the initial investment into foreign equities or bonds, and then hedges longer term the
potential sale of that asset down the track. The outlook for currencies is not usually the primary driver taken into
account when an investment decision is made in the underlying asset. However, as events such as Brexit and the de-
peg of Swiss franc from the euro have shown us, currency moves are capable of significantly increasing – or,
conversely, wiping out – returns when translated back to the original investment currency. Similarly, a corporate
CFO planning out the revenue stream of a long-term investment will not be thinking of FX first and foremost.
However, they would be ill-advised to ignore where in the cycle the FX rate is as it can make or break their
investment.

PPP, REER, NEER and flows

Examining a currency’s “fair value” is an often-used tool for long-term currency valuation, and usually involves the
economic principle of PPP. Purchasing power parity invokes the “law of one price” – that is, that tradable goods
should be the same price the world over, and if they are not then goods will be exported to where they are locally
more expensive, or conversely imported from where they are cheapest. It is important to make the distinction that
this principle holds for tradable goods only: a haircut may be cheaper in Toronto, but if my head is in New York I
will be unable to avail myself of this discount. Once importing or exporting takes place, the currency is either sold
or demanded and this, in turn, drives the price of that currency down or up with respect to the counter currency, until
such time as there is no advantage and prices reach equilibrium. As you can imagine, it takes time for this corrective
process to work. This is not a measure that usually works in the short term. If you lived in New York, you saw this
in action in its retail manifestation when the UK pound was above US$1.70/£ or the Brazilian real stronger than
R$2.0/US$ (see below). UK and Brazilian friends would literally bring an extra suitcase when they came to NY to
stuff it with jeans and electronics for the return trip!
There are a few sophisticated, and a few funny, measures of PPP. Some measure prices at the producer price level
but most do so at the retail level. The most well-known of these measures is The Economist magazine’s Big Mac
index. The logic is simple and probably accounts for its success. To prepare a Big Mac anywhere in the world, you
need the exact same ingredients. Therefore, the price implied by the exchange rate should be an accurate reflection
of PPP. This measure was so popular that, when the Argentine government wanted to create the perception of low
inflation, it manipulated the price in the local McDonald’s. What’s great is that there was a food arbitrage spotted as
a consequence (Politi, 2011). Figure 3.3 shows how under- (or over-) valued each currency is against the US dollar
using the PPP (Big Mac index) measure. According to this chart you should travel to Malaysia when you have Big
Mac cravings.

Consider how slowly PPP valuations change. Figure 3.4 shows how the valuation of the yen versus the US dollar
changed over a decade. Where in 2011 the yen was close to 20% overvalued, it took two years to go to fair
valuation, and another two years for the yen to be undervalued by 30%. As an aside, this points to two very
interesting observations about currencies and why hedging is so important. First, these are huge moves! Few other
asset classes of such import move so much. Second, FX is like a huge aircraft carrier – once it starts heading in a
certain direction, it takes a long time to stop the momentum and a long time to turn it around.
Along a similar line of analysis, economists often want to consider a country’s exchange rate relative to a group
of countries as opposed to our previous single country evaluation. The idea is to understand the country’s broader
trade strength versus its trading partners. The instrument to achieve this is the nominal effective exchange rate
(NEER), which is the weighted-average rate at which one country’s currency exchanges for a basket of foreign
currencies. Since that is too straightforward, economists have decreed that if you want better analytics, you must
adjust NEER with an inflation measure to arrive at the real NEER or REER. In all fairness to economists, this is a
much more important number to understand when conducting economic analysis or designing trade policies.
In addition to the trade-related flows mentioned above, another factor used by asset managers, and one that should
really be used more widely, is portfolio flows. Portfolio flows examine how foreign holdings of underlying assets,
such as domestic equity and bond markets, change over time. The data tends to actually be better for developing
market countries than it is for G10. There is often daily data produced by the local stock exchanges on what
percentage of their stock market is held by foreign investors. This data is often a good early warning sign that a
currency is about to move, since once an investor has decided to sell out of, for example, South Korean equities, it is
rare and (depending on the currency restrictions or controls in place) sometimes illegal for the foreign investor to
hold onto that currency after divestment. Therefore, currency end-users can receive a heads-up of a potential move
by watching if international investors are beginning to sell domestic bonds and/or equities. One factor to check on is
whether there is not simply an asset class switch taking place. For example, if an asset manager is still long Chile but
purely making an investment and then switches across from equities to bonds, they will not need to sell the Chilean
peso. By watching other assets classes, FX end-users can gain potential forecasting insight.
There are different levels at which portfolio flows are monitored depending on the transparency available in that
country. The monitor in Figure 3.5, for example, shows portfolio investment changes as they are made available.
As the markets for passive investing have seen explosive growth, another area one can examine for signs of
movement are the exchange-traded fund (ETF) flows (see Figure 3.6). ETFs have become the instrument of choice
for developed world investors to invest outside their market of expertise. As of March 2018, there were US$6.6
trillion in ETF assets under management (AUM). Of those, US$3.1 trillion were for instruments outside the US.
When a US investor buys a Brazilian corporate-based ETF, the authorised participants will buy the underlying
securities from the Brazilian exchange that comprise the ETF and deliver those shares to the ETF provider. To do
that, they have to buy reals and give US dollars in exchange.
FOLLOW THE MONEY: USING MARKET SENTIMENT TO PREDICT FX
Increasingly, market players are looking to assess market sentiment as a means to determine FX direction. Given
that FX market participants themselves put their “money where their mouth is”, it is not surprising they are paying
some attention to the views of others who have skin in the game. This has historically been difficult to do. As we
know, FX is predominantly an OTC market and it is therefore challenging to get into the heads of the traders. It is
not a perfect science but there are several indicators available to help get a good understanding of which direction
the market is betting on.

Futures predict the future

One such example is the “Commitment of Traders” report that is produced weekly by the CME. This report
examines the long and short positions of currency futures contracts traded over the exchange. This is a very small
proportion of the overall FX market, but what is particularly interesting is the fact that it is broken down into
“commercial” (ie, hedgers or those who have to trade no matter what their view is) and “non-commercial” (ie,
speculators who are clearly trading their view). Few indicators offer this level of granularity, and one can definitely
use it as an indicator of market sentiment. Figure 3.7 shows the “Commitment of Traders” report for the euro
contract broken down into non-commercial and commercial market participants.
Here is a small taste of the kind of analytics you can do around this data. Figure 3.8 graphs the net non-
commercial position against the net commercial position. This is how speculators are increasing or decreasing their
overall position over time versus what the commercial interests are doing. Against those two flows, we plotted the
EURUSD exchange rate. Note that, as the speculative flows increase by buying euros, so does the euro. Okay, we
must stop here and beg your forgiveness for offering such poorly contrived arguments. First, you cannot claim
correlation because it “looks like they track”. Second, even if they were highly correlated it does not mean there is
causation. These and many other reasons means this analysis is shaky at best. We offer it only as way to show how
you may start to think about this data in projecting FX rates. Once again you see evidence of our theme: the mind
believes what it wants to believe.
Options distributions and FX forecasting
Another indicator of market sentiment can be found in the price of an option (see Figure 3.9). In Part II of this book
we will go into excruciating detail why this is so, but for now you must take our word for it. To price options you
need to have a probability distribution. If you have a probability distribution, then you can give a probability for
each range on the curve. These probabilities are a direct output of something called “implied volatility” or, for our
purposes here, what the market thinks will happen. You do not need to be an option trader to use this output.
With those provisos, what does the market think will happen? Let’s ask a few questions. For instance, what is the
probability that:

USDMXN will be above Ps20/US$ in six months? (answer: 23.4%);

USDCAD will gain 5% between today and the next three months? (answer: 13.0%); and
the euro will touch parity (US$1/€) six months from now but before two years? (answer: 22.4%).

Pretty cool, right? Cool maybe, but there is no guarantee. This is simply what the traders are paying in premium for
options that relate to those levels. There are a myriad ways to use this information, but the most straightforward is to
answer everyday questions that arise in your business. For example, if one is looking to either hedge against, or
speculate that, the US dollar will appreciate by 5% against the Canadian dollar over the next three months, these
analytics provide useful tools to give insight into the options market’s view of the probability that this will happen.

This is interesting for two reasons. First, even if I am not a participant in the FX options markets, it is good to
gain insight as to what market participants are predicting; second, if I am a user of options, particularly barrier
options, it is helpful to understand the percentage probability that the options market is giving to my barrier being
triggered. One such example is the use of a forward extra structure by a corporate. This involves buying a protection
option and (usually for zero cost) selling a knock-in option to finance it. This is an effective strategy for corporates
who need to be protected against adverse currency movements, but at the same time would like to have some
flexibility to participate in a favourable exchange rate move. However, the downside is that if the currency moves
too far in your favour, you are knocked into a forward that is at a worse rate than a forward that you could have
booked at the time of trading. It is this difference that “pays for” your flexibility. It is therefore prudent to assess
where to set this barrier (knock-in) level. If the options market predicts there is a high percentage chance of
knocking in, your optionality may be wasted and you may have been better off simply doing a forward. We will
examine these and other structures in greater detail later in the book. This is just a tickler to wet your appetite. But
don’t worry, in just a few hundred pages you will be able to understand all those details!
Finally, depending on the nature of the trading of a particular currency pair, option dealers engage in something
called delta hedging (see Part II). In the short and very short term, sometimes the delta hedge activity in the market
forces the FX rate to stay around a particular strike level. Given the transparency offered by the Dodd–Frank
regulations, market participants can get an idea of what strikes may have a concentration of volume. For example, if
at 9:50am New York time spot USDJPY is at ¥112.15/US$, 10 minutes before a huge amount of options struck at
¥112.00/US$ expire at 10am, then there is a high probability that spot will head towards ¥112.00 before 10am.
There is another market-driven number that many in the market mistakenly think indicates market direction: the
“risk reversal”. This number is a measure of how much more the market is paying up for calls over puts, which in
theory should have equal probability of being in-the-money. This is not true. A risk reversal greater than zero only
tells us that, should the spot rate move in that direction, the market expects implied volatility to rise. Stated
differently, the market is worried that moves in that direction will cause more uncertainty.
If the above two paragraphs are Greek to you, don’t worry – by the end of this book, you will know exactly what
we are talking about. You will be explaining risk reversals to your kids as a bedtime story!

PUTTING IT ALL TOGETHER

If you are not interested in spending significant portions of your life thinking about where FX rates are headed, how
can you get to the answers quickly? Traditionally, as mentioned you could call your banker and have them mobilise
their support staff to get you an answer on the economic macro outlook for the country you are interested in, and
then get their strategist to put it all together and bottom-line it for you. As we discussed and will expand upon later,
these groups are disappearing at the big banks. They are simply not making enough margin to support these high-
powered individuals.
However, you’re in luck. There are services out there that put it all together and deliver the answers to you
electronically. These consultative types of support service are now being offered by consultants, but also by vendors
such as Reuters and Bloomberg. For example, you can go to the Bloomberg Intelligence section on your terminal
and get massive amounts of information and analytics at the click of a button. It was only a few years ago that this
type of analysis was put out exclusively by global economic units at the major banks.

Next, you may want to know what your banker strategists are thinking and what this macro analyses means for
specific rates (see Figure 3.10). Figure 3.11, on the other hand, presents a partial page of forecasts out many years
into the future. You can also click on any currency pair and see exactly what a particular bank predicts. Finally, the
system ranks who is the best at predicting a particular rate at a given time (see Figure 3.12). You would be surprised
at who is the best – it’s very often not the biggest names in the market. Okay, now that you know what the strategists
are thinking, bring it all together with market-driven probabilities (see Figure 3.13).
Figure 3.13 is the mother of all predictive charts. Look at the table on the left. In this example, for USDCAD,
given the option market prices, we know that there is a 72.9% chance that USDCAD will be between 1.2681 and
1.4049 between now and the next six months. The chart on the right combines the subjective evaluation. The
histogram denoted by the number ① is taken from all the strategists on the previous figures. Their consensus is that
line ② lies around 1.3600. Note that some strategists actually think the US dollar may skyrocket to around 1.47. The
forward rate, the rate at which I can deal at today (line ③), is 1.3365. The option market mode is below 1.3400.
 Where does this leave us? Beyond confused? Here’s an imagined statement that may describe the above
information:
Economists are expecting a further decrease in the Canadian dollar over the next six months. Some are actually very bearish. However, there
is not a lot of money put on that perspective. The money says the Canadian dollar will appreciate or remain near spot.

SOME THOUGHTS ON THE APPLICATIONS OF CURRENCY FORECASTING

What’s the best reason to get currency forecasts? To use that knowledge to turn a profit, right? Wrong. It’s true that
if we can predict the future we can use that knowledge to make money. However, predicting currency movements
and turning a profit from that is not such an easy proposition. Most participants misunderstand this relationship and
waste their time and money trying to turn random numbers into gold. The ones that make money in the FX market
by speculating have an information, speed or positioning advantage, not more accurate forecasts. The last part of this
book explores this concept extensively.
Let’s look at corporations and non-trading entities. The simplest and best reason to have accurate currency
forecasts is to establish a budget rate. This budget rate can be used to establish revenue numbers that will be
communicated to their broader community (such as analysts). It can also be used to establish the right transfer
pricing levels, set the price for products and services, or help determine when to outsource expenses. In addition, it
is imperative to assume some level of FX when investing overseas, or when buying assets through an acquisition or
building a factory. This is the main reason corporate treasuries around the world routinely poll banks for short-term
and multi-year currency forecasts. While at the core well-intentioned, these forecasting efforts quickly turn to
speculation and away from their original purpose.
There are many well-publicised stories of corporations making widgets, before turning to extensive currency
speculation that then endangered the whole company, and I am sure there is many more that will never see the light
of day. Dell was a famous example from the early 1990s,1 as well as some car manufacturers in the late 1990s and
2000s (Landler, 2004). This type of behaviour still continues and can afflict anyone, even not-for-profit
organisations such as Greenpeace, who issued a formal apology to its donors for speculating in FX and losing €3.8
million in 2014 (Oltermann, 2014).
Short-term currency forecasting is well beyond human capabilities. The sheer number and complexity of factors
playing crucial roles in the determination of the exchange rate make this impossible. It is a deep cognitive fallacy of
the human mind that so-called “experts” interviewed on live television, so confident in their own forecast, can
explain it to you in the short time between two commercial breaks. In the animated movie Ratatouille, you might
remember that the famous chef’s mantra was “anyone can cook”. Our position is “nobody can forecast exchange
rates”! Another of the main themes of our book, which will take you all the way to the very last pages.
At the extreme end of the time scale, very short-term forecasting might seem possible. Most recognise the
tremendous success of high-frequency trading firms. To summarise this approach, trading firms have been taking
advantage of electronic trading platforms by creating algorithms capable of trading hundreds of times per second,
and have been successful at making money.2 However, the ability to make money by trading high-frequency has
nothing to do with the ability to forecast the exchange rate. Profitable high-frequency traders do not create forecasts,
but rather simply react to trades happening in the markets. They are faster than other market participants.
There are many examples of reaction trading, including the Rothschilds’ use of carrier pigeons to profit from
news of the battle of Waterloo in 1815, and some using the Chappe telegraph to trade on the Philadelphia exchange
using information coming from New York that they were the only ones to possess in Philadelphia (Mihm, 2014).
This was happening in the 19th century and is again happening in the 21st century, but has nothing to do with the
ability to forecast the exchange rate.
Although short-term forecasting cannot be successful consistently, long-term trends in the real economy will
eventually cause the exchange rate to bend to their will. The problem is that the time scale is completely unknown.
For example, look at Zimbabwe and Venezuela, two countries once prosperous and then both ruled by a
dictatorship. As the respective regimes started to implement populist, often unhinged, economic policies that divided
their people, it became easy to forecast the demise of the economy and eventually of the currency. In both cases, it
took double-digit years for the theory to play out. Economists were eventually vindicated, but at the price of
immense suffering for millions of people and economic collapse.
The above arguments are not to say that FX exposures should be ignored, nor are we saying we should throw our
hands up in despair. On the contrary, in the third part of this book we will show some ways to use options to deal
with these exposures without relying on forecasts. It sounds like a paradox that you don’t need to forecast the future
to protect yourself, but it is not. You need to build your risk profile in such a way that no matter what happens you
will live to fight another day. Many believe it is their duty to correctly forecast the future. It may be their duty, but it
is also impossible. There are many things that should be done beforehand to reduce the overall exposure to FX
movements. The most obvious is to use natural hedges to reduce your overall risk. Corporations have known this for
a long time. If you have revenue in country A, check if it’s optimal to move expenses in the same country. The
simplest way is to borrow in the same country, thus using your revenue to pay off the loan and never have to go
through the FX markets. BMW has taken this idea to a new level of sophistication and implementation, by
developing a way of doing business that looks to mitigate FX risk as a key priority (Xu and Liu, 2012).
That is the essence of effective hedging for businesses. Stick to your business because business works in a
different way than the financial markets. If you tell the chief executive officer (CEO) that the firm is making loads
of money because they are the only ones to own a plant that produces fast memory chips, and can therefore raise
prices, it will make perfect sense to them. When they turn around and look at profits generated by the treasury, they
apply the same logic and think those profits can be replicated systematically. However, nothing could be further
from the truth.
The most uncomfortable meetings any of us have had with corporate treasury professionals were the ones where
we were asked to teach them how to generate profits in their FX exposures. They would not believe us when we told
them that there is no recipe to make easy money in the FX market. They would just go and ask another bank. Many
times we have brought out FX spot dealers to meet with corporate treasury departments so they can learn how
trading can help them. This is all very human. The practice and commerce of the financial markets seems to have an
obligation to yield money to those who seek it. Just as studying chemistry gives people the ability to forecast the
result of chemical reactions, it seems logical that the study and practice of finance would yield to its practitioners the
ability to forecast exchange rates.
Economic factors such as gross domestic product (GDP) and unemployment are predictable. FX rates are not.
How is USDJPY so much more impenetrable than the GDP of Japan? Should they not be equally difficult to
predict? Very simply, if everybody thinks USDJPY should trade at ¥140.00/US$, the next day USDJPY will trade at
¥140.00/US$, as trading is liquid and frictionless. However, if everybody thinks that within the next two years the
GDP of Japan will rise at an annualised rate of 10%, then it will still take two years for the real economy to hire the
right number of people and to create the capacity for that growth. It will just take its time and may never happen.
But not so for the financial markets, where prices move instantaneously and are not constrained by nature, in the
short term at least.
Those who are most determined to get FX right in corporate treasuries persist in speculative activity by getting
really sophisticated. In many ways, they are trying to outsmart the traders in the market. Treasury professionals are
very busy with payments, accounting, execution of hedges, strategic meetings and risk quantification. On the
surface, the implication is that within a couple of hours they could learn how to beat entire legions of Wall Street
traders at their own game. They try to outperform, with just a couple of hours per day, people who live in a
windowless closet and devote every waking hour to the observation, examination and fanatical dissection of the
EURUSD exchange rate, and read on the Internet everything that it is humanly possible to read about EURUSD. In
that sense, it’s a zero-sum game. If you gain, some trader will lose.
Technological advancements in computing power have enabled people to dream big. It would seem that with the
right software a sure way to make money can now be revealed as computers can go through more data than the
human mind can imagine, let alone process. The most common manifestation of this technical capability is to
backtest strategies. Backtesting is not the answer – backtesting soothes, but does not cure. It feels right to say: “Here
is my exact strategy, and here is how it would have performed had I implemented it.” Even better, now you can ask
the machine: “What was the best strategy within these parameters to make the highest profit?” However, backtesting
used in this way shows a lack of understanding of the FX market. There is a faith in the existence of winning
strategies, which is the same as believing in the ability to forecast FX, which leads to the gold rush of backtesting.
Once they finally find a strategy that would have made a lot of money in years gone by, they believe they have
discovered a secret law of nature that will continue to make money in the future.
Strategies might have worked in the past due to the particular economic situation, or to the positioning and
sentiment of market participants, or to a particular news flow of the time. This does not mean that the same strategy
will work today or at any future time. Finance is not a physics laboratory, where you can run and re-run the same
experiment an infinite number of times and expect the same result. As the markets will never be the same two times
in history, the outcome of the same strategy can never be trusted to be the same again. As Heraclitos said a few
years ago, “No man ever steps in the same river twice, for it’s not the same river and he’s not the same man.”
Having bashed forecasting as a tool to make money, the truth is some market operators have been making money
consistently for many years in the FX markets by speculating, and therefore they necessarily have to be able to
forecast the market. There are FX speculators that make money more than 50% of the time they trade, but this is
attributed to their ability to enter positions only when they detect a greater probability of being right. For example, a
hedge fund might buy the euro only when the positioning in euro futures on the CME reaches a record. This does
not mean that they are able to forecast the exchange rate. Another hedge fund might be selling the euro every time
its 55-day moving average is above the spot rate. Again, this is nothing to do with the ability to constantly forecast
the exchange rate.
For example, the real trick that will allow a hedge fund to continue to make money is the ability to use the CME
positioning or the moving average at different points in time to make money in different ways. They study human
and market behaviour, and are quick to adjust their positions and mindset to reflect the brand-new conditions that
have arisen in the last millisecond. The naïve sauvage, on the other hand, will blindly believe in the existence of
immutable rules carved into the atomic structure of the cosmos that have the power to generate positive FX returns;
the hedge fund portfolio manager will try one strategy and then another, maybe running them in tandem and shifting
the size of the positions in a way that will allow them to exploit the better one. The sauvage thinks he knows, and the
hedge fund manager is very happy to repeat the famous four words: “I do not know”. Some portfolio managers are
able to do this in a very successful way for a while. The number of those who can achieve this for 20 years is very
limited, and only comprise the finest minds.

CONCLUSION
Trying to predict where FX rates will go in the future is a massive endeavour across all FX markets and all non-FX
participants profiles. Many have devoted their life’s work to understanding what makes FX rates change, and to
predicting the direction and speed of change. The most visually appealing approach is to look at charts and apply
technical analysis, with all its sophisticated tools that have been refined over the decades. However, the more
credible forecasts come from economists and FX strategist usually employed by large banks. They peddle their ware
to corporations, asset managers and hedge funds with sophisticated analytics and often very complex econometric
models. They look at the macroeconomic factors that impact economic activity, but also incorporate practical
considerations such as trade and financial flows that can impact shorter-term movements.
This was primarily the domain of bank economists. Banks leveraged access to these strategists to get FX
transactions. As technology and pressure on margins hit the banks, these departments have been scaled back. The
economic analysis and forecasts are more and more being provided by other vendors as a service. Technological
advancement has led to transparency and comparisons that often reveal the futility of trying to predict FX. At the
same time, technological progress has led others to erroneously assume techniques such as backtesting can reveal
the secrets of the universe, of immutable winning strategies that only need to be tapped into by the treasure hunters.
Many FX participants have moved away from going to an enlightened person digesting all the information and
providing the answer, and are instead looking to the market for clues about positioning. This insight can come from
the futures market and, more likely, from the options market.

1 See http://www.nytimes.com/1992/12/01/business/company-news-dell-reports-moredetails-in-a-dispute.html
2 Most readers will be familiar with the literature on this topic, but see the excellent account by our colleague Michael Lewis (2015).
 4
Basic FX Instruments

There are two broad intertwining paths in any asset class: the core, or underlying instrument, and the derivative path.
By definition, the derivative is a derivation of the underlying, so it cannot exist without the core. With that in mind,
this chapter will build some foundations in core FX so we can later move onto the derivative part.
Four related instruments will be discussed:

FX spot market;
forward market;
non-deliverable forward (NDF) market; and
cross-currency swap as it relates to the cross-currency basis.

It can be argued that the forward market is nothing more than spot for delayed delivery, that an NDF is an artificial,
virtual forward and that a cross-currency swap is a series of forwards. Why examine them separately? The reason is
that there are many differences in application and effectiveness that make them very interesting and critical to
understand, especially if we want to advance to more complex derivatives. To get some context, we begin with the
desks on a typical bank trading floor that trades and sells these products.
Although we will be looking at the mechanics of the market, it is critical to appreciate changes to the regulatory
environment surrounding FX and, in particular, the release of the FX Global Code (the Code) in May 2017.
Traditionally, FX is not regarded as a “security” in the sense that equities and bonds are securities – regulation has
not extended to FX. Further, its global nature means it is very difficult to apply standardised regulation to the FX
market. As a result, a group of central banks, as opposed to securities regulators, and market participants from
around the globe have worked together to arrive at a set of principles to which all market participants should be able
to adhere. Several jurisdictions have had their own codes of conduct; the Code will supplant these.
This code has no legal force and is not formal legislation in any jurisdiction. Nevertheless, regulators in several
major markets have already indicated that they will expect FX market participants in their jurisdictions to abide by
the Code. By mid-2018 every major financial institution demanded that both counterparties and vendors agreed to
follow the Code. This will be the closest the world has ever come to having uniform regulation of the FX market!

THE ANATOMY OF AN FX TRADING FLOOR

The main arena for the bulk of the volume in FX is the major global bank trading floor. On a typical bank FX
trading floor there are different desks serving distinct purposes, all working together throughout the day to service
their clients’ FX needs and generate revenue. There are groups that make markets (provide price quotes) and trade
around the risk taken on by the bank, typically arising from client deals. Those are:

the spot desk, which executes spot transactions; for most currencies, these are settled on the trade date plus two
business days;
the forward desk, which manages transactions with settlement dates beyond and before the two-day spot settlement;
and
the option trading desk, which manages the residual option risk; for the sake of accuracy, the option desk will also
trade extensively with the spot and forward desks, trying to minimise the spot and forward risk while keeping
the volatility risk it wants to hold.

There is also a group of desks that talk to clients. Typically, at large banks there would be the hedge fund sales desk,
the “real money” sales desk (ie, asset managers, insurance companies and pension funds), and the major corporate
and regional (smaller) corporate desks. There might also be a desk catering to private, high-net-worth investors or
retail investors. Often those desks are on a different floor or even a different building. Depending on the size of the
bank and its focus, there may also be desks speaking to central banks and other financial institutions (called non-
reciprocal banks).
All these desks are supported by a cadre of individuals with very different skillsets. We will expand on some of
those groups in the next chapter, but let’s start here with the structuring and other supportive functions. A desk
structuring complex trades supports both trading and sales desks, but is mostly there for client deals. The structurers
usually have a quantitative bent and work with the salesperson to create customised strategies for clients. We will
explore some of these strategies later in this book, but suffice to say they can be complex. The purpose of this
activity is to combine spot, forward and option instruments to create a tailor-made risk/reward scenario that
expresses the client’s interest and encourages them to trade with the institution.
Very often, they create strategies to address specific client requirements that may arise due to another transaction
that the bank is advising the client on. For example, suppose US-based company ABC is buying a company in
France. The bank may be financing the acquisition. The total loan they need to make depends on the EURUSD
exchange rate. If the total cost of the company was €100 million at an exchange rate of parity (US$1 to buy €1), the
loan would be for US$100 million. At an exchange rate of US$1.2/€, the loan would be for US$120 million.
Additionally, at US$1.2/€ the company has to rethink its assumptions about whether this was a good acquisition or
not, as a 20% price increase (in US dollar terms) may kill the whole deal. Given that it often takes months and
sometimes years to negotiate such transactions and obtain the necessary approvals, the client may seek advice from
the structuring desk on how to protect itself against adverse currency movements during the negotiation period.
Then there are the strategists, whose job it is to understand the global macroeconomic environment and offer
advice as to how the economies that affect their clients are evolving. We looked at their output in Chapter 3. These
individuals tend to be economists with extensive FX experience. There are also groups of people that look after the
electronic engines that feed liquidity to the traders but also to the clients. Many times on those floors you will find
middle-office people who are charged with confirming the risk taken by the traders and communicating it to the
back office. Increasingly, compliance and regulatory surveillance people are likely to be found on the trading floor
as well.
 That’s a lot of people! Over the years the size, influence and importance of each of these groups has ebbed and
flowed. In the 1980s and early 1990s, the spot dealers were at the centre. At a major bank where I worked in the
mid-1990s there were 30 people on the spot desk: 10 senior traders and many junior and trainee dealers. When the
chief dealer stood up and wanted to execute a large trade, they literally had 15 people snapping to attention and
“making calls” (obtaining price quotes from other banks). Some got prices by phone but most via Reuters Dealing, a
private interbank communications system commonly used in the FX market. Today these same desks may have only
a staff of five, but they are executing many multiples in volume compared with the 1990s. The difference in human
resources required is due to the increase in sheer computing power and the proliferation of electronic trading. In the
late 1990s and 2000s, the focus shifted to sales desks and structuring. High leverage and an ever-increasing list of
available structured products generated the revenues. Post-2008, electronic trading became the growth area and
today it may appear that compliance is the only department expanding its workforce.
The raison d’être of the whole operation is to intermediate, manage and distribute market risk. This is done by
traders making prices to clients via their salespeople, whose job is then to bring in flow for the traders. On occasion
there is an underlying fundamental question about the role of the strategist. The issue can take on a deeper
discussion, which we will not address here, but think about this potential conflict of interest for a bank that makes its
living by the transactions its clients execute – for example, selling euros – but also publishes commentaries and
research opining as to whether the euro will in fact rise. This underlying potential conflict of interest is addressed
head on by regulation in Europe under the requirements of the second Markets in Financial Instruments Directive
(MiFId II).
 Now that we have a sense of the people involved, we can examine FX spot and the mechanics of a spot
transaction.

FOREIGN EXCHANGE SPOT

Consider the first foreign currency that comes to mind, the euro. It has a value, and the value of one euro is
expressed just as if it were a stock price. The euro-to-dollar exchange rate (EURUSD) is the amount in US dollars
that buys one euro today. If that number rises, then the euro is stronger.
Figure 4.2 shows the price in US dollars of one euro since the euro’s birth at the beginning of January 1999. In
addition to the euro, the same quoting convention (called an “inverse quote”) holds for the UK pound (GBPUSD or
£/US$), the Australian dollar (AUDUSD or A$/US$), the New Zealand dollar (NZDUSD or NZ$/US$), the
Botswana pula (BWPUSD or P/US$), the Solomon Islands dollar (SBDUSD or SI$/US$), the Tongan pa’anga
(TOPUSD or T$/US$) and the Western Samoan tala (WSTUSD or WS$/US$). I was once told as a graduate trainee
that one way to remember which currencies were quoted this way is they were all cricket-playing nations. Like all
rules, this one is broken because it leaves out the many cricket-playing nations of South Asia (such as India, Sri
Lanka, Bangladesh and Pakistan), along with the West Indies and Eastern and Southern Africa. For all other
currencies, the convention is to quote the amount of foreign currency that buys one US dollar today (ie, US$1 = n of
the other currency – called a “direct quote”). If that number rises, it shows that the US dollar is strengthening,
meaning the other currency is weaker. For example, USDJPY is how the market quotes the US dollar in terms of
Japanese yen (by linking the two ISO codes). It is quoted as the number of Japanese yen per one US dollar, unlike
the six “inverse” currency pairs mentioned above.
Let’s go through quotes for the most traded currencies and understand how they are expressed (see Figure 4.3).1
Can you guess why the total in Figure 4.3 is 200%? Here’s a hint: can you trade a single currency by itself?

After this section, we will use market conventions to describe currency pairs interchangeably. As an example,
“yen” can refer to USDJPY as where a currency is mentioned without a counter-currency one can assume that the
counter-currency is the US dollar. Additionally, ¥112.05/US$ will be abbreviated to 112.05, as we feel the reader is
more comfortable with that market jargon. Both are market convention when referring to the exchange rate between
the US dollar and the Japanese yen.

Exercise 4.1: If GBPUSD rises then the pound:

1) strengthens 2) weakens

Exercise 4.2: If the Australian dollar weakens then the AUDUSD exchange rate goes:
1) higher 2) lower

Exercise 4.3: If the Japanese yen (JPY) weakens then the exchange rate USDJPY goes:
1) higher 2) lower
 Answer: 1:2:1

Figure 4.4 shows the USDJPY exchange rate. Note that the market convention expresses the amount of Japanese
yen that buys one US dollar. As a consequence, a rising exchange rate corresponds to yen weakness (dollar
strength). In other words, you need more yen to buy one US dollar.

Executing a trade in the spot FX market

When executing a trade in the spot, or cash market, the money will be wired to a bank account that accepts that
currency (ie, the trade will be settled) usually two business days later (unless you are transacting USDCAD,
USDTRY, USDMXN or USDRUB, which settle the next day, or USDKZT, which settles same day). As in any
other financial market, if a trader at a hedge fund needs to purchase euros they call a salesperson at a bank and ask
for a price. Alternatively, they can use the bank’s Internet portal or a multi-dealer platform to ask for the exact same
quote, in which case they will be executing trades with the click of a mouse. After the details are agreed upon, the
trade is concluded, or “done” in jargon. The back office will take care of confirming all the parameters of the trade,
to wire the euros to the client’s bank in two business days’ time, and to verify that the client has wired, on the same
settlement date, the dollars that will pay for those euros, or to reach out with inquiries if the dollars are not received.
There are a couple of types of quotes that one can ask for.

One-way quote: The requester needs to buy euros, discloses the amount and the market-maker trader at the other
end of the telephone will show a price, such as US$1.0567/€. Usually the price will have four decimals if traded
by humans and four or five decimals if quoted by an electronic trading site. In the case of USDJPY (dollar yen),
there will only be two decimals for humans and two or three for computers (an old-style example would be:
“can I have your offer in 10 million euros versus dollars?”, “67”, “done”, “okay, to confirm you buy ten euros at
67 for spot”).
Two-way quote: The requester discloses the euro amount but doesn’t say if they intend to buy or sell, and the
market-maker trader on the other end of the telephone gives two numbers (eg, 1.0563/1.0567). If the requester
intends to buy euros they buy at 1.0567, which is the bank’s ask (or offer) price, but if they intend to sell euros
then they sell at 1.0563, which is the bank’s bid price (an old-style example would be: “can I have a two-way
price in 10 million euros versus dollars?”).

It may look straightforward, but both counterparties have responsibilities to uphold under the Code in this
transaction. Indeed, the aim of the Code is to engender a level of trust among market participants that each is
adhering to a common set of standards of behaviour so as to “promote the integrity and effective functioning”2 of
the FX market. On both sides of the transaction, the Code seeks to address a lack of transparency that was once part
and parcel of dealing FX as a non-exchange market.
From the bank side, the Code directs that banks should disclose to clients that the final transaction price quoted by
the bank may be inclusive of mark-up, how that mark-up is determined and that a price quoted to one client may
differ from prices quoted to other clients. To clearly define markup let’s take the example of when a trader makes a
two-way price like 63/67, and in a scenario where the client buys at 67, then the bank is short 10 million euros and
will have to buy them from some other bank in the market (unless there is luckily another client that wants to sell the
bank 10 million euros at the same time). Then the bank estimates, without certainty, that it will be able to buy those
10 euros at 67 or below. This is risky, but that is why the trader makes a buck. If the salesperson then sells the euros
to the client at 69, the difference 69-67=2 is called the markup. The markup is not part of the risk because,
theoretically, even with a zero markup the bank would still at least break even. This is addressed specifically by
Principle 14 of the Code. To promote transparency, the Code provides illustrative cases that include guidelines on
mark-up – for example, price “needs to be fair and reasonable and can take into account a number of factors
regarding risk, cost and relationship”. Once an overall price was quoted to a client that included mark-up, credit
spread (the credit spread is a charge relative to the risk that the client goes bankrupt before the transaction settles. As
we have seen, this is typically two days for spot trades and is equal to the maturity of the trade for forward trades),
and liquidity premium. The market has become more transparent as to the contributions of these components to the
overall price. Importantly, the bank should disclose the existence of mark-up, and some banks have even gone as far
as to specify the exact amount of the mark-up on each transaction.
This provision addresses, and puts a structure around, practices such as those that were made public in a multi-
billion-dollar settlement between UK and US regulators and a group of major banks in 2015. A trader from one of
the banks involved indicated that mark-up was essentially balancing how much profit could be extracted from a
trade before the client would become dissatisfied and take their business elsewhere:
mark-up is making sure you make the right decision on price … which is what’s the worst price I can put on this where the customer’s
decision to trade with me or give me future business doesn’t change. If you ain’t cheatin’, you ain’t trying.3

To counter cases such as these, the Code specifies the reasons permitted for a difference in price and provides
examples of reasons that do not qualify. In our FX spot example, two separate clients, identical in all aspects other
than level of sophistication, looking to buy euros could not be shown prices of US$1.0570/€ (sophisticated client)
and US$1.0590/€ (unsophisticated client) when the market is in fact trading at US$1.0565/€. The Code example
specifies that: “it is not fair and reasonable” to discriminate “between clients based only on their level of
sophistication”. On a typical sales desk most of the revenues would come from this price discrimination between
discriminating clients (pun intended: very low revenues, business maintained as a service to keep other more
lucrative business) and clients that did not know where the price should be. The author ventures to estimate that at
least 80% of a salesperson’s revenues would come from non-discriminating clients.4
Equally, the price taker (client) has obligations to adhere to when asking their bank for a price. The Code
addresses this in Principle 12, specifically that market participants “should not request transactions, [or] create
orders … with the intent of disrupting market function or hindering the price discovery process”. As an example, on
the price taker side it was not uncommon for the price-maker to ask if the spot trade was the “full amount” – that is,
the amount being shown to the price-maker at the time was the price taker’s full interest or if there was possibly a
residual amount to be dealt later or with other counterparties.
 In our FX spot example of a client looking to buy euros, the client could ask for a EURUSD spot price in €100
million. If the amount is a full amount, the trader would be able to show a price of US$1.0570/€ and be able to
manage their risk on the deal. If, however, the amount is not full and the client intends to trade other amounts with
other banks at the same or a similar time, the trader would want to quote a wider price of, say, US$1.0575/€ to
compensate for the extra risk that this transaction now carries. Previously, depending on the client and the
relationship with their bank, the client either answered this full-amount question truthfully, or not. Regardless, in the
immediate aftermath following the trade it became evident by the price action whether or not the client had in fact
also traded with other price-makers, as traders were unable to sufficiently manage their risk before the market
moved against them.5
In the good old days, etiquette also played a self-regulating role. For example, if a client asked a market maker for
a two-way price and then asked if the market maker could tighten (reduce) the bid/ask spread, they would have been
obliged to trade and it would have been very rude to say “nothing done”. Etiquette alone, however, may not have
been enough. Under normal circumstances, you could assume that a client calling for a price (two-way or otherwise)
might at the same time be on the phone with other banks and only say “done” with the bank showing the best price.
Analogously, the bank could start buying in anticipation of the client closing a trade with one of the banks on the
phone. If the bank won the trade with the client, it had already bought some euros, so it could give them to the client.
If the bank lost the trade, the winner bank would need to buy euros and the price would potentially rise and the
losing bank could sell the pre-purchased euros to the winner bank instead. This bank practice, of buying a little for
the purpose of testing market liquidity before giving a firm quote to the client, has been under enormous scrutiny,
and it is famously called “front running”. We discuss this extensively in later chapters.
Electronic platforms dramatically altered the FX market’s structure. Let’s look at a typical situation in USDJPY.
Figure 4.5 is a screen from the Bloomberg FXGO platform, which shows a request for a two-way price sent to six
fictitious banks: BGDM, BGYN, BGSP, BGEU, BGYL and BGTO. Some banks have made a two-way price while
others have declined. However, the client can clearly see who has the best price and see exactly how much more, or
less, it will cost to deal with one bank versus another. “BB Ind.” shows you a benchmark reference.
The price action described above happens every day on billions of dollars’ worth of FX transactions across all
currency pairs, in all time zones, across many trading platforms. In addition to cutting back the human element in
FX trading, banks are using electronic trading as their main armour against the declining margins and increased
pressure to be transparent. To entice clients, they offer a slew of pre- and post-trade analytics.
As Figure 4.6 shows, only 15% of the deals are done by telephone with a human on the other end, and 55% are
electronic, voice direct (bilateral) versus indirect (brokered). Within electronic trading, there is a clear bias towards
bilateral electronic trading platforms. In other words, banks want clients to use their proprietary electronic systems.
This brings client liquidity directly to their traders or their algorithmic dealing system. Additionally, it is more
difficult to put banks in competition using multiple single-bank portals than it is using just a single multi-bank portal
which compares the prices for you. Within those proprietary platforms, the one huge evolution has been the
explosion of algorithmic trading. Algorithmic trading systems bring the electronic execution of FX trades into a new
realm that appears poised to dominate. According to a Greenwich Associates study (Johnson, 2017):
Execution algorithms have become popular for foreign exchange (FX) trading in recent years, driven by strong uptake among hedge funds,
market makers and speculators. Not wanting to be left out of this trading technology arms race, Greenwich Associates is seeing increased
algo usage by long-term investors and corporate end users. For these traders, algorithms are sophisticated tools that allow them to
intelligently access multiple liquidity sources, reduce information leakage and improve trading performance.

Some of the larger banks continue to engender goodwill through the other services they provide, but it is getting
tougher and tougher to fight the machine. Greenwich goes on to say: “While dealer relationships remain important –
given the research, credit and other non-execution services they provide – longer-term investors are demanding a
level of increased control and transparency into the trading process that only automation brings.” Also, based on
their survey, “Greenwich Associates expects algo usage to steadily increase over the next three to five years as
pending MiFID II regulations and the FX Global Code of Conduct place even more emphasis on best execution and
prompt more market participants to employ Trade Cost Analysis (TCA) and similar analytic tools.”
Especially in the case of portfolio managers, whose speciality might be international equities or bonds but not FX,
it becomes very attractive to be able to offload the responsibility to perform in terms of FX execution to an
algorithm. This is part of the algo success. Algo trading employs many styles and is constantly evolving to create a
competitive edge for different market players. Algorithmic trading is great because it can massage the market and
create the least possible impact on price, or opportunistically take advantage of complementary flows. However, it
usually requires executing the whole trade with just one bank, which then goes unopposed by competition. The
converse happens when instead the client uses a multi-bank platform, where multiple banks compete on the same
trade. In this case the client could claim to achieve the best possible price at that time. A deeper analysis of the
subject is beyond the scope of this book, but it is a very important and relatively new way of executing an FX trade
that will certainly impact the FX market’s evolution.
Now, back to the main story. Let’s review some of the basics of buying and selling in the spot FX market. If a
trader’s price in the EURUSD market is 1.0564/1.0566, where US$1.0564/€ is their bid and US$1.0566/€ is their
ask, then US$1.0565/€ will be the mid-market. If a client of the bank buys euros versus US dollars at US$1.0566/€
then, at least in principle, the mid-market should move to the rate of the last trade. The next quote that the trader will
show will therefore be 1.0565/1.0567 (assuming a consistent spread). If clients keep buying then the market maker
will move their quoted price higher to manage their market risk. That is, they will be making it decreasingly
attractive for a euros buyer to buy from them, but increasingly attractive for a euros seller to sell to them (due to
their increasing bid price), allowing the trader to exit (“square”) their position.
 Let’s move on to the next level of sophistication. Assume the market is 1.0564/1.0566 for all the banks in the
market. Suppose a client calls and asks for a two-way price (they are not disclosing if they intend to buy or sell).
Suppose also that you strongly suspect the client will be buying. Corporate FX salespeople at banks typically know
what their clients want to do. As an example, if the salesperson knows that a certain pharmaceutical company is
asking for a price in Vietnamese dong, they will most likely be “long” dong (that is, have revenue from selling their
drugs in Vietnam) and therefore looking to sell dong and buy US dollars. Again, the Code Principle 12 comes into
play to guide the players in this transaction. In the interests of transparency, the client may ask that the mute button
(on the salesperson’s telephone) be left open so that they can hear the price that the trader makes to the salesperson.
Here, the salesperson should not disclose via any means such as hand signals or otherwise the potential suspected
direction that the client may end up dealing, and this should not be reflected in the price shown. The sentences above
are proof that the FX market is being revolutionised. Having been in this market for decades, we would have never
in our wildest dreams imagined such behavioural requirements. The art of secret hand gestures used to rival any
third base baseball coach!
Indeed, the market maker should not know the identity of the client. They should just be making the most
aggressive (ie, narrowest) price they can (taking into account their own position and views on market direction) and
let the sales coverage person, who knows the client, widen the price appropriately, based on such factors as the type
of client, their credit and the client’s appropriate mark-up. Notice the fine balance, since similar clients should not
receive different markups based on their respective levels of sophistication. In fact, the Code stipulates that “market
Participants should not … provide prices with the intent of … hindering the price discovery process”.6 Several banks
have admitted to this practice in the past via their disclosure notices. One such example is Citibank’s Plea
Agreement of May 20, 2015, where it disclosed that “certain of our salespeople used hand signals to indicate to the
trader to add mark-up to the price being quoted to the client on the open telephone line, so as to avoid informing the
client listening on the phone of the mark-up and/or the amount of the mark-up”.7 These disclosures are very
surprising for opposing reasons. On one hand you could be scandalised at the lack of ethical behaviour, but on the
other hand one wonders how in the world will it ever be policed and monitored? Will the next settlement plea
include eye-blinking or nose-scratching?
Meanwhile, clients (price takers) are also trying to monitor and reduce transaction costs (ie, spread between dealt
and market price). There is a huge business building around transaction cost analysis (TCA). Until now, TCA
analytics were a “nice to have”. However, under the regulatory regimes sweeping the world, and particularly MiFID,
banks and clients must be able to prove that they in fact dealt at the best possible price. In fact, given the trading
complexity and the proliferation of automated execution tools, reducing the transactional friction is a huge topic
across many asset classes, and frankly beyond the scope of this book. Nevertheless, we will give a taste of what
TCA analytics would look like for FX.
Figure 4.7 shows a dashboard from the Bloomberg VWAP functionality, and serves as an example of how FX
clients can use historical and real-time snapshots of price, volume and other trade analytics for a sample exchange
rate, to determine post-trade performance against a benchmark.
In addition to the natural conflict between the market maker and the price taker, intermediating institutions such
as banks and brokers act so as to fulfil their individual policy objectives. As we saw in Chapter 2, only a few large
banks dominate the FX spot market. Since the global financial crisis in particular, the number of banks that will be
willing (and able) to manage large FX spot risks arising from client trades has plummeted. Today, a handful of large
banks are real price makers and the rest are happy to effectively source all of their prices from the former – in
essence, white-labelling market making, and generating revenues by servicing clients and maintaining their list of
relationships.8
The major banks know most of the information on the major currencies, and the smaller (regional) banks have
created a niche for themselves in their country’s currencies. For example, the Canadian banks created a small
community, strong not just with Canadian clients but also globally, dominant in providing Canadian dollar liquidity,
while if you want to deal in kiwi (New Zealand dollar) you may be more likely to call a bank from that country.

Would you like another service with your FX deal?

Bank services around spot FX did not stop with a price. They provided economic analysis at the macro and micro
level, they structured hedges for particular needs, they marked your positions to market9 and even performed the
client’s hedge accounting calculations. For all these services, a tight relationship developed over the years between
the bank capital markets area and many of the clients (see discussion above about publishing research). The careers
of many salespeople were advanced by being approached by a competing institution seeking to enhance its
relationship with a particular client. When these considerations are in the equation of price making, it inevitably
means the buy side (clients) will not trade at the best price.
Banks do not want to spend time and energy competing on price. It is very difficult to consistently have better
prices than the competition on a product that is identical to competitors. Over time, such competition tends to erode
profits by reducing spreads. It is much better for banks to provide clients with economic research, market
commentary, number crunching and any other piece of intelligence, and then just remind them that they need to be
rewarded with more business. By definition, this means they will expect a certain amount of trades in scenarios
where the client may not receive the best price. A key point of the game here is that the client will never really know
how much they are paying for all those extra services.
Consider the example of a bank that has a desk that offered accounting advice as part of an overall relationship
service. This has dramatically changed with regulations such as the Dodd–Frank Act and MiFID, and the massive
scandals that rocked the FX world. All transactions and services should now be priced fairly and on their own
individual value, rather than as part of an overall relationship service. Banks are implementing technology at
breakneck speeds and searching for new, compliant and sustainable business models under the new regime (such as
disclosure of mark-up amounts). Having said that, it appears that the concentration of market making is increasing
rather than decreasing, as mentioned above.
 The relationship between dealers and the buy side is undergoing a dramatic transformation, much like it happened
after the “Big Bang” in equities in London in 1986. In the not too distant past, it would not have been unusual for a
salesperson to take clients for lunch on Friday and, upon returning to the office, take a call from the same client,
asking for a large trade. This was the way the client used to pay for lunch. The buy-side dealers liked this as well.
Beyond the fact that they enjoyed dining at the bank’s expense, their companies benefited from the ancillary
services provided. Marking books to market, getting economists to tell you what is happening in various countries
and getting smart people to structure your hedges is expensive.
To this day, some buy-side traders still like to talk to a sell-side salesperson before executing, so that they can
better understand market conditions. They will tell you that executing a trade is never just the process of asking five
banks and picking the keenest price. Recall how we previously showed that asking five banks actually may be
detrimental to the client’s interests in some cases. There is much more to it, of course – there is also the human
aspect. In the most dramatic marketplace upheaval, yet to run its course, the days for those people are numbered, by
regulation and by the technological tsunami unleashed. In the Dodd–Frank world, by law, salespersons cannot reveal
confidential information about trades done by other clients of the bank. The Code recognises the value of a degree of
information sharing in Principle 22. Such information should be at a general level, and be “effectively aggregated
and anonymised”.10 Salespeople must be careful to provide market colour that is not too specific or detailed. It is
equally difficult, from the buy-side trader’s point of view, to justify why business was directed to a particular bank
in return for information, analysis and services. Remember that every time business is directed away from the best
price, the buy-side trader should be able to justify why they did not pay the lowest price as a matter of fiduciary
duty.
With these changes sweeping the marketplace, banks are curtailing all their extra services. They are firing
economists, getting rid of structurers, refusing to mark clients’ books to market and demanding clients deal
electronically and not call the salesperson.

THE FORWARD AND NDF MARKETS

Most transactions in FX are not spot (or cash) transactions as the cashflows in question happen at a future date. A
typical example may be a corporation that must pay its suppliers in euros in one year. The firm will call a bank and
purchase euros, and they will specify that they need the euros in one year’s time. The bank is happy to oblige, but
what price should they charge for a euro delivered in one year? Should it be the same as the price of a euro in two
days (spot)? Let’s examine what happens and see if we can calculate the cost. To satisfy the client, the bank will buy
the euros spot at today’s spot rate and place them on deposit in a bank in Frankfurt (or any other Eurozone location)
for one year, so that they have them ready for the corporation when it needs them. In one year, the corporation will
pay the bank dollars in exchange for the euros they need.

That’s easy enough. However, there are two complications to this simple plan. First, the bank in Frankfurt will
pay interest on the euros deposited. Assuming interest rates are positive, the bank in New York will need to deposit
a smaller number of euros today and they will grow over one year to the needed amount. Second, the bank in New
York will have to pay for those euros purchased today by delivering dollars. However, the corporate client said very
clearly that they will pay for the euros in one year, not today; no funds are expected from them for one year. The
bank in New York has only one choice. Borrow some dollars, use them to pay for the euros, and in one year’s time,
when finally the corporate client pays for the euros, the New York bank will receive dollars and will repay the dollar
loan. If you divide the amount of euros that will be available in Frankfurt in one year by the amount of dollars that
will be needed to repay the dollar loan contracted in New York, that ratio is exactly the exchange rate to purchase
euros for value in one year’s time (see Figure 4.8). In this figure we have simplified the calculation of the interest
rate by assuming annual compounding. This ratio is also known as the forward rate, or the all-in forward rate.
Note in this example that the higher the interest rates in euro, the cheaper it is to buy euros for delivery at a future
time via a forward trade. This is somewhat counterintuitive, but is due to the fact that if you receive the euros in one
year, for one year you will be foregoing the high euro interest rate. Similarly, the higher the interest rates in dollars,
the more expensive it is to buy euros in one year’s time. This is because the bank will then have to pay the high
dollar interest rate for one year.

Exercise 4.4:
What’s the formula for the calculation of the forward rate in EURUSD:

S = spot rate in EURUSD

rd = US$ one-year interest rate
rf = € one-year interest rate
Fwd = Rate at which you can buy euros vs. US$ for value in one year’s time
 Answer: FWD = S × (1 + rd) / (1 + rf)
In other words, the higher the interest rate in euros, the lower the price to buy euros in the future.

Exercise 4.5:
Repeat the previous exercise in the case where the corporation needs to buy Japanese yen (¥) in the future,
not euros.
Hint: The formula will be different because the yen quotation (USDJPY) is different from the euro quotation
(EURUSD)

Answer: FWD = S × (1 + rf) / (1 + rd)

Figure 4.9 shows what a bank would do to deliver dollars in one year to a client who would pay euros in one year,
and how to calculate the appropriate exchange rate (the one-year forward rate). Look at the “1Y” row. The bank will
borrow euros at a rate of –0.4989 (that is, a negative interest rate of 0.4989%) and lend US dollars at a rate of
2.1891% for one year. The ratio between the resulting numbers of euros and dollars, inclusive of the interest rate
received or paid, is US$1.209101/€. The market is actually trading now at US$1.213575/€. There is a small
difference due to the choice of the interest rate curves used and the cross-currency basis (more on this later).
Not all currencies can be readily traded the way we described above. Some governments want to protect their
financial sector by preventing speculation in their currency. They institute currency controls and make their currency
non-deliverable. This usually means that FX transactions must be justified and documented as having an approved
commercial rationale. For example, a corporation that wishes to enter into an FX transaction must show
documentation of the underlying commercial transaction. Trade-related transactions are usually approved, but hedge
funds are forbidden from investing in the currency. This protects the country’s financial system from “hot” hedge
fund money that may want to rush out of the country. Their solution is to never let those hedge funds in to begin
with. The list of countries with exchange controls is very long, and includes some very popular investment
destinations such as Brazil, Korea and China.
If a hedge fund wants to speculate and buy the Brazilian real, how can it do it? It can enter into a contract, an
NDF, with a foreign bank that will give it the same future gains as it would have received had it been able to
speculate by buying the real. In effect, it agrees to pretend the regulations banning the hedge fund from speculating
do not exist. Since this contract is outside the affected country’s borders, there is nothing that country can do about
it. This is a bilateral agreement, typically between a bank and another entity, and until recently, given the OTC
nature of the FX market, such contracts were not even made public or reported to any entity or regulator. The Dodd–
Frank Act has made it mandatory to report such contracts to the public. Such contracts are not only used for
speculation, but also to hedge broader economic activity – including investments in a country that does not allow for
the free float of their exchange rate.

Exercise 4.6:
Can you figure out the formula for calculating the gain around a non-deliverable forward?

Exercise 4.7:
Can you explain why there has to be a maturity date associated with non-deliverable forward contract? Can
one not just hold the non-deliverable forward contract indefinitely just like one holds a spot position?

Answer to Exercise 4.6: Suppose that the hedge fund wanted to bet that the value of R$10 mio will rise in
six months’ time. The way to figure these things out is not to overcomplicate the topic but think of the
practical steps you would take.
The hedge fund would purchase those Brazilian reals with a forward contract, where the trading price is
f6. At the end of six months it would re-sell the reals back to the market, at a rate equal to the then-prevailing
spot rate s6. Therefore, the total gain for the hedge fund will be R$10 mio × (1/s6 – 1/f6), and this number is
 in US$ (algebra helps us here (1/$/R$ – 1/$/R$) gives us $/R$ then multiply by R$, cancels out the R$ and
leaves us with US dollars).
Let us see the flow of numbers in a table:
BRL Position USD Position
Initial trade, long reals 10mio -10mio/f6
Closing the trade -10mio +10mio/s6
Net result of trade 0 10mio × (1/s6 – 1/f6)

So we can see that in the end the BRL position is nil and the final PNL of the trade expressed in USD is
equal to 10mio × (1/s6 – 1/f6), which can be positive or negative. It will be positive if s6<f6, or in other words
if USDBRL strengthened with respect to the forward rate observed six months before.
Because these contracts are not driven directly and exclusively by interest rates, like a regular forward
contract, we just use the beginning and ending prices and settle the difference. Nice and easy!

Answer to Exercise 4.7: This is a more philosophical question. Let’s take a step back and ask “how is it that
spot trades exist in the first place?” After all, you could imagine a world where people only trade on a
forward basis. The spot contract in a deliverable currency like euros, implies that you receive the currency
and need to decide what to do with it, such as lend it and receive interest. In non-deliverable currencies (like
NDFs) you cannot take delivery and so there is no concept of receiving interest. Holding the currency is a
concept that does not exist in NDFs. So with NDF contracts we must have a fixed maturity.
Suppose that you traded a six-month forward contract in USDBRL where you bought the real (in other
words, you sold USDBRL). After two months you want to close out your contract and take your gains. You
can do that by closing out the original contract, which is now a four-month contract. The difference in the
market rate of the original contract and the closeout will generate a US$-deonominated cashflow for value
date, now the four-month date. This can be positive or negative, but it will be for value date four months
from the date of the closeout. You can realise the gain (if it is a gain) today by asking your counterparty to
pay you today the present value of the gain which is for value four months from today.

What is brilliant about NDF contracts is that there is no need to exchange reals – only US dollars change hands. If
the formula gives a positive number, then it means that the USDBRL exchange rate has decreased, signifying that
the real has strengthened, just as the hedge fund had anticipated. The bank will pay that amount to the hedge fund. If
the formula gives a negative number, then it means that the real weakened (ie, USDBRL rose), and therefore the
hedge fund will pay that amount (or, more precisely, the absolute value of that number, which is the number without
the minus sign) to the bank. The only issue remaining with such an NDF contract is that the bank and the hedge fund
will have to agree on what s6 is. This is not an easy negotiation, especially in illiquid markets.
Of course, the hedge fund will try to push s6 lower and the bank will try to argue that it is higher, each pursuing
their own interest. The market, in its infinite wisdom, has found a solution to this problem. In almost all countries
you will find currency rate fixings published daily. Sometimes these are put together by the central bank, sometimes
by news organisations, or other data providers. As long as the bank and the hedge fund can agree on one of those
fixings, they will write on the NDF contract which one of the published fixings is to be used in the formula for the
calculation of the gain/losses related to the contract. This is the origin of the use of fixings that we explored at some
length in Chapter 2.
One might ask whether the formula used to calculate the EURUSD forward rate can still be used for non-
deliverable currencies. The answer is no. You remember that the formula was discovered assuming that the bank
would borrow US dollars and lend euros for one year. In the case of non-deliverable currencies, the same argument
fails because the bank would not be able to lend reals. The bank would not even be able to purchase reals, as
domestic currency controls would prevent it. In essence, the quotation of NDFs is based on supply and demand for
such contracts and for particular periods of time. In that context, NDF’s are a truer indication of market sentiment
than deliverable forwards. Although the above statement is true, as a practical matter there are a few market
participants who are able to lend reals and borrow US dollars, such as import/export firms. Therefore, there are some
indirect links between NDF quotes, the spot market and the interest rate market beyond pure supply and demand.
 First Paradox: You are an American who wants to speculate in the FX market. You buy US$10 mio versus
the ¥ (Japanese yen) using a forward trade. You are told that such a position has unlimited upside and,
unfortunately, unlimited downside. You discover with disappointment that only the latter statement is true.
Why?

Solution: Write down the formula for the profit and loss (P&L) of your trade. You receive US$10 mio and
pay US$10 mio × Fwd yen (you pay an amount of yen equal to the US$10 mio times the forward rate at the
time of the trade).
In order to realise your profit and loss (P&L) in dollars you have to close out the ¥ position, so you buy
back US$10 mio × Fwd yen and pay with US$10 mio × Fwd/Sf dollars, where Sf is the USDJPY exchange
rate at the value date of the forward contract. The net profit and loss (P&L) of your transaction is US$10 mio
× (1 – Fwd/Sf)). So if the dollar rises in value, Fwd/Sf falls in value, but can only get to zero (or
infinitesimally close to zero, as USDJPY approaches infinity), at which point your profit and loss (P&L) will
be close to US$10 mio. So your upside is capped to US$ 10 mio.
If USDJPY collapses (the dollar gets weaker), then Sf will plummet infinitesimally close to zero, and (1 –
Fwd/Sf) will approach negative infinity, which means that your losses will be close to infinity. Bad trade if
there’s a chance Japan will disappear from the face of the earth!

Second paradox: Suppose you execute a forward trade where you sell US$ buy ¥ in the amount of US$1.00
for a value date one year from today (therefore we do not need to worry about interest rate compounding). At
the value date the trade will realise a profit-and-loss (P&L). Write the formula down using Sf as the final
USDJPY exchange rate. Now imagine that instead you borrow $1.00, sell it for yen, put the yen in a Japanese
bank for a year, after a year you retrieve the yen and the interest, sell them for dollars and with the proceeds
pay back your borrowed $1.00 together with its interest. Now write down the P&L of that trade, using Sf as
the final USDJPY exchange rate. Are the two formulae the same?

Solution:
First trade: at value date sell one dollar, receive an amount of yen equal to Fwd, where Fwd is the forward
date traded. Close out the position by purchasing Fwd yen and selling Fwd/Sf dollars. Net profit and loss
(P&L) of the transaction is (Fwd/Sf – 1).

Second trade: Borrow a dollar today, for one year, and in one year’s time have to return 1 × (1 + rd) to the
bank. Today, sell the borrowed dollar and receive S0 yen (where S0 is the USDJPY exchange rate at trade
date, and remember that it signifies the number of yen that can buy US$1.00). Deposit the S0 yen in a bank in
Tokyo and after one year receive back S0 × (1 + rf) yen. Sell those and receive US$ equal to S0/St × (1 + rf).
Net profit and loss (P&L) of the trade is S0/Sf x (1 + rf) – 1 × (1 + rd). This is not the same as the profit and
loss (P&L) of the first trade, above.

Reconcilliation: The profit and loss (P&L) of the second trade can be rewritten as:

(1 + rd) x [S0 × (1 + rt)/(1 + rd)/Sf – 1] = (1 + rd) × [Fwd/Sf – 1]

This looks more like the first trade but the two formulas are still not identical. The reason is that the size of
the two trades is not the same. In the first trade US$1.00 corresponds of the future value size of the trade. In
the second trade US$1.00 is what you borrow today, so that is the present value size of the trade. When
adjusted for interest rate (ie, (1 + rd)) the two formulas become the same.

The peculiar case of the renminbi market

China is perhaps the country we get the most questions on. The Chinese authorities operate exchange rate controls,
so to exchange deliverable renminbi11 for US dollars you have to provide substantial documentation. In a typical
case, a Chinese importer who needs US dollars to purchase goods that will be imported and sold in mainland China
will provide paperwork to the relevant authorities that states the above intentions and then will be allowed to buy US
dollars. This is also the case for international market participants wanting to deal deliverable CNY in the onshore
market. The onshore market trades over an exchange called the China Foreign Exchange Trade System (CFETS).
Non-Chinese corporations with real underlying export and import invoices are able to deal into this market with the
right guidance from their banking partners, a careful selection of trading times and the completion of the necessary
paperwork.

There is a reasonable amount of transparency in this market due to its exchange nature. It is possible to see both
the time and the price of each deal. Further, CFETS applies a number system as a guide to the size of the trade. For
example, a size 2 trade corresponds to US$3–4 million, while a size 6 is US$20–30 million (see Figure 4.10). So,
why do we not see more deliverable onshore Chinese yuan traded? Well, first, the majority of the FX market
participants do not have access to this deliverable market. However, the speculative community does not require the
Chinese yuan to be deliverable to take a view on China, they use NDFs. For the smaller part of the market able to
provide documentation for real export or import purposes, there are a few hurdles. CFETS is not open 24 hours and,
while they have extended their hours to take in some trading times outside Chinese time zones, a market participant
in the US will find it challenging to find a time when both markets are open for long and with good liquidity.
Therefore, to avail of the greater liquidity Asian time zones offer often involves an element of “pyjama dealing” –
that fun practice of getting up in the middle of the night to put on a trade. One further factor to consider is the
availability of expertise. Your banking partner will need to be well-versed in these trades and help to navigate the
paperwork, which is onerous. Lastly, you will need a yuan bank account in China if receiving that currency,
something which until recently was not possible offshore, meaning you had to have a relationship with a local
Chinese bank.
Open to all foreign participants, without any of the documentation requirements, is the NDF market. This used to
be the only option for anyone wanting to take a speculative position in the USDCNY exchange rate and, as
discussed above, the calculation of those rates is not necessarily based on interest rate parity. It can be purely a view
on the future direction expectations. They are private contracts between a speculator and a bank that will be settled
in the future depending on the official USDCNY exchange (of fixing) rate. In summer 2010, all this changed. A new
market emerged that saw the reduction in the amount of non-deliverable Chinese yuan trades that are dealt. Despite
China being the world’s largest trading partner (see Figure 4.11), the volumes of CNY NDFs decreased.
 We know this because Dodd–Frank regulation requires that “US persons” report their NDF trades. This provides
helpful information, to see on any given day the volumes of NDFs that are being transacted in the market. The same
reporting onus does not apply to deliverable forwards. Therefore, given the importance of China, both as a global
trade partner and as a country that the FX markets would like to speculate on, you would expect the volumes of
CNY NDFs to be placed as either number one or two as the NDF with the greatest volume. This is no longer the
case. Figure 4.12 shows that the Chinese yuan NDF was the fifth currency in terms of volume traded, behind the
Indian rupee, the Brazilian real, the Korean won and the the Taiwan new dollar.
 In the summer 2010, unique to China, a third option emerged that was responsible for this drop in yuan NDF
volume. The Chinese government allowed renminbi to be settled and transacted in Hong Kong – that is, a
deliverable offshore CNY. This is the same currency as you can hold in mainland China, although you cannot freely
transfer it from mainland to Hong Kong, or vice versa, without trade documentation. Therefore, the renminbi in
Hong Kong, usually denoted by “CNH” instead of “CNY”, trades at a different rate, and is called “USDCNH”.
This created a very interesting situation where there are literally three markets for forward transactions in the
same currency. Two of them are based, as in our EURUSD example above, on interest rate differential, because they
represent deliverable currency. The third contract is the NDFs, and it does not necessarily follow interest rate parity.
Instead, it just obeys the law of supply and demand. This means many interesting observations can be made about
the relative spreads of these three quotes by people who understand markets and their participants.
Figure 4.13 shows ①, the USDCNY exchange rate traded in mainland China and ②, the official daily USDCNY
fixing, and finally as line number 3 the USDCNH exchange rate traded in Hong Kong. Note the fixing is at times
quite different from line ①,which is where the market trades. In line ③ you see the USDCNH exchange rate, or the
Chinese renminbi, as traded in Hong Kong – same currency, but different jurisdiction and different exchange rate.
You could take your legal allowance of yuans as banknotes out of your Shanghai bank account, walk across the
border into Hong Kong and deposit the same banknotes in your bank account in Hong Kong. The difference is only
in how many US dollars the banknotes could buy you in Shanghai versus how many they could buy you in Hong
Kong. But the banknotes would be the same. If you understand the differences among these three exchange rates,
you are on your way to FX guru status!
 The code for CNH (Chinese yuan deliverable in Hong Kong) has also been adopted by all the other centres who
now have renminbi hub status, which is over 20 and continues to expand (see Figure 4.14). In each of the hubs, the
PBOC (China’s central bank) has designated a clearing bank. The role of the clearer is to ensure efficient and
smooth CNH (offshore yuan) trading in the local financial centre. These hubs provide many benefits for their market
participants. For instance, you no longer have the time zone issue, you are able to trade with your regular banking
partner, who in turn will trade and clear with that hub’s clearer bank. For banks without an overseas network, the
efficiencies are greater, reducing the need to trade with a correspondent bank. Cities that have received hub status
tend to be concentrated in North America, with Canada and then the US launching. This will bring liquidity to the
time zone. For corporations, it will be easier to settle their invoices in yuan; they will also receive a cost saving for
doing so. Many have estimated that this could be between 3–5%, but the amount can vary widely by industry and
over time, and can possibly turn negative. For asset managers, potential inclusion of China in various equity and
fixed income indexes will require CNH trading to facilitate those investments. Lastly, while most major banks have
previously been able to offer CNH trading and have settled through Hong Kong, some regional banks without this
branch network will now be able to offer RMB products and services to their clients.
 Figure 4.14 shows offshore hubs in each region and the banks quoting CNH in Hong Kong. China’s creation of
CNH and the RMB trading and clearing hubs are a major policy tool towards the internationalisation of their
currency. There was certainly a feeling among market players that the yuan could only appreciate. However, Figure
4.15, which depicts CNY between 2007 and 2017, shows that since 2010 CNY has appreciated but also depreciated.
One misconception that we often encounter is that while the currency can now fluctuate somewhat, it is still very
stable from a volatility perspective – and therefore buying options remains cheap compared to other emerging-
market (and some developed-market) currencies such as Brazil.
 Figure 4.16 shows a lower implied (and realised) volatility for China compared with other currencies. Notice
USDCNY volatility is the last row in the figure. There is often speculation as to when the yuan will actually float,
with the most frequently heard answer being in the range of five to 10 years, at which point the RMB trading and
clearing hubs will become obsolete. Until that happens, however, anyone trading in China has the potential of three
markets to choose from, depending on their circumstances, which can provide a lot of flexibility. It is possible to
examine the pricing advantages and favourable points of each market to determine which will provide the best
outcome for different requirements.

FX SWAPS
We have discussed forward transactions in a world of perfect information. Suppose a corporation enters into a
contract to buy euros to pay its supplier in one year (at a then-market rate equal to fwd1), but discovers that it made a
mistake and it needs them in nine months. What can be done to fix the error? Well, it can call the bank and tell it that
it now wants to sell those euros for value in one year’s time (at a prevailing market rate equal to fwd2), and it needs
to buy euros for value at nine months (at a market rate equal to fwd3). This would set things straight.
In our example, the original trade together with the subsequent sale of euros for one year’s time cancels out the
one-year euro purchase obligation. However it’s not that clean because there was a timing issue between the time the
first transaction was done and the second leg to unwind the one year and move it to nine months.
No euros change hands now, but in a year the bank receives from its client an amount in dollars equal to the
amount in euros times fwd1, and the bank has to pay to the client a US dollar amount equal to the euro amount times
fwd2. Net net, the bank receives a dollar cashflow equal to euros amount × (fwd1 – fwd2). If that amount is negative,
then the bank actually pays the absolute value of that amount (the bank pays the amount without the negative sign).
The number of dollars that the bank has to deliver and receive in one year’s time are equal and wash out. On top of
that, in nine months the bank pays a euro amount to the client, and receives an amount in dollars equal to the euro
amount × fwd3 from i. Lastly, the bank also receives and pays to the client the same amount in dollars for value 9
months, and those wash out as well. The operation where, at the same time, one sells (or buys) euros for one date
and buys (or sells) the same amount in euros for another date is called an FX swap. The operation whereby the client
has corrected the error in the timing, as a whole, constitutes the FX swap. So an FX swap is composed of four
cashflows. Swaps are a huge part of the market (see Chapter 2). The reason why swaps exist by themselves instead
of simply being an aggregation of forward trades is very important. If the two transactions (for one year and for nine
months) are executed separately, the bank has to cover its spot risk twice, and is at risk of losing money twice. If the
operations are put together and executed as one swap, then the bank is not at risk with respect to movements in the
spot rate. Not at risk twice, not at risk once … it is not at risk. Why is this good for the client? Because the bid/ask
spread, or the cost of transacting, is much smaller. Why is there no spot risk? If you go back to Figure 4.8, that
shows you how the bank constructed the original forward trade. If the client asks to modify it making it a nine-
months instead of a one-year trade, all the bank has to do is arrange to repay its USD borrowing three months in
advance, and to take out its deposit in a foreign land three months in advance. Not that all this is completely without
costs, as there is a bid/ask spread embedded in the forward points, but you can see that there is no need for extra spot
transactions to do it, hence there is no spot risk. As the FX needs and the timing of those needs change for
corporations, asset managers and all other FX market participants, they adjust their currency deliveries to reflect
those needs.
This makes swaps very interesting. When trading conventional forward contracts, the bank takes on a short-term
risk. In the case of a sale, the bank is at risk that the exchange rate rises suddenly before it could cover its risk by
buying euros in the interbank market. Not so in the case of a swap. A rise in the euro would hurt the bank’s cashflow
in one year’s time, but would benefit in equal and opposite measure the bank’s cashflow in nine months’ time.
Therefore, the bid/ask spread in a swap trade is much tighter as the trade itself carries much less risk for the market
maker, and therefore it can afford to make a tight price.
Figure 4.17 shows the forward market for EURUSD. Below are all the basic conventions.

Line number 1 (①) on top left “SP” is the spot quote: today is Febraury 27, 2017 (top right) and the delivery for
spot trades is March 1, 2017, at a rate of 1.0588 if the client sells euros and 1.0586 if the client buys euros.
On the left, for example, the “6M” row shows that if the client wants to sell euros for value date (or “delivery”) in
six months, then the date will be September 1, 2017, and the rate will be the spot rate plus 0.009628; if the client
wants to buy euros for the same date, the rate will be the spot rate plus 0.009714. Those two numbers are called
the forward points.
If the value date is not one of those listed, then it is called a “broken date”. An example of broken date, August 15,
2018, is on the top right. This date corresponds to 532 days (as shown). Clients who want to sell will sell at spot
plus 0.033222, while clients who want to buy will buy at the spot rate plus 0.033597.
Clients sometimes need to sell for one date and buy for another; they can do so via an FX swap, where the two rates
are calculated using the “swap points”, which are the difference between the two rates. For example, on the
bottom right we see that if a client wants to buy euros for value date September 5, 2018, and at the same time
 sell euros for value date September 19, 2018, then the swap points will be 0.000638. If a client wants to sell
euros for value date September 5, 2018, and at the same time buy euros for value date September 19, 2018, then
the swap points will be 0.001467. The actual two rates used don’t make a lot of difference, as long as the swap
points are calculated correctly – you can calculate the rate for the “near date”, or September 5, 2018, using the
table on the top right, and then the “far date” of September 19, 2018, will be calculated using the swap points.

The FX market is OTC and, for the most part, the volumes are not public. Actually, there is no central repository of
all the trades and nobody really knows which is the largest FX bank in the world. As discussed in previous chapters,
all this is changing with the new regulatory landscape. Beyond the tri-annual survey conducted by the BIS (Chapter
2), the only thing that gets close to a league table in the FX market is the annual Euromoney survey and one
conducted by Greenwich Associates. This is really peculiar in today’s technological world. Greenwich and
Euromoney send out to buy-side entities hundreds of surveys asking them to say how much volume in FX trades
they conducted in the past year and with which bank, from which they compile a global ranking. Survey respondents
can answer what they want, it does not have to be the truth as obviously there is no way Greenwich or Euromoney
can check. Therefore, banks have very often resorted to tricks to gain the proverbial “edge” and show they are the
biggest. Such tricks include asking clients to change their answers after a particular bank found out from Greenwich
that they were not leading the survey. Another approach was to hire fashion models to visit the clients and ask for
their vote in the survey.
Another very smart and totally legal trick was to try to dramatically boost the volume of your business in some
instruments that bear very little risk. For example, a short-term FX swap carries no spot risk and just a small interest
rate risk. Short-term interest rates don’t move a lot and, even if they did, their impact on the profit and loss (P&L) of
a one-week trade is very small. However, swap trades do count in the volume of the surveys, so a very smart head of
FX at a large New York bank instructed his traders to show a spread very close to zero for all those short-term FX
swap trades. The result was that his bank won a huge percentage of those trades (it quoted mid-market, while
competitors quoted bid price or ask price), and therefore boosted its volumes with clients or at least based on the
Greenwich survey definition of volume. At the same time, the bank was not losing money and was not really taking
much additional risk.
We have seen how the forward rate (that is, the all-in forward rate) is calculated and quoted. However, at times
people in the market will quote the forward rate in points. The forward rate is fwd = s + points, where “s” is the spot
rate. The “points” are also called “the forward points”. By definition, the difference between the forward rate and the
spot rate are these forward points. Why is this interesting? When transacting a forward contract in a deliverable
currency, you could break down the trade into two components: spot rate plus forward points. Instead of three or
more banks competing on the trade and you getting the best all-in forward rate, it would be better for you to receive
the best spot rate among the three and the best forward points among the three. This would be at least as good as,
and potentially better than, getting the best all-in forward rate among the three banks.
How can this be done? The client can ask the three banks to compete on a swap where they buy euros for value in
a year’s time and they sell euros for value spot (cash transaction). After competing that trade, the same client can
pick up the telephone again and call the same or different banks and ask them to buy euros for value spot, and again
assign the deal to the best of the three rates. This is a great idea to squeeze from the market the last drop of
competition, but there is one complication. Similar to the previous example where the one-year trade was pulled
back to nine months, there will be some residual dollar cashflow for value spot. The reason is that the spot rate
quoted at the time of the first auction (which technically had the client sell euros for value spot) will not necessarily
match the spot rate quoted at the time of the spot trade auction (the client buying euros for value spot). It is not a big
problem, as at the spot value date the client will have a dollar cashflow for an amount equal to euro amount × (spot1
– spot2), where spot1 is the spot rate quoted at the time of the first auction and spot2 is the spot rate quoted at the
time of the second auction. This dollar cashflow could potentially be negative for the client, so to make sure it has a
little cash-on-hand in dollars.
Many years ago, one of my reports was a young hire formerly with an Eastern European central bank. I asked him
to go retrieve the market for forward points in Kuwaiti dinar. He came back with the number, but I did not know
how to read it. Was it 0.37 units or 0.0037? My report said he did not know either because he did not want to sound
stupid asking the trader, so now he sounded stupid to his direct boss. Nice trade!

CROSS-CURRENCY BASIS
The final topic in our exposition of basic FX is not so basic or simple, but very important to understanding what
happens in the market in practice. The topic of cross-currency basis is where theory meets supply and demand, and
is one of the main arenas where currencies meet interest rates.12
This section will review the advent of cross-currency basis, and then expand on the topic by describing the
peculiar events of October 2016 and the obscure way cross-currency basis propagates to remote corners of the global
financial system. We will show how Japanese insurance companies and hyper-conservative US money market
investments are linked to the European Central Bank (ECB) monetary policy decisions and to corporate bond
issuance in foreign currencies.

FX and interest rate arbitrage

Recall what we discussed in the forward market section earlier in this chapter. Also remember that spot, forward and
the two countries’ interest rates are tied together by a simple formula: Fwd = S × (1+ rf) / (1 + rd). This is the
formula when the currency pair quotation convention is “direct”, as in USDJPY. The important thing to appreciate is
that the forward rate is calculated easily if you know spot and the two interest rates. This is known as interest rate
parity.
Now, take a look at Figure 4.18. The first column is a series of future dates. The “FX Swap” column shows the
actual market EURUSD forward quote for value on that particular date; all quotes are mid-market in this example.
The “EUR Yield” column and the “USD Yield” column show the corresponding interest rate in Europe and the US.
By interest rate parity, if we know the spot rate and the euros and US dollar interest rate, we should be able to
calculate the EURUSD forward rate. That is what appears in the “FX Swap Implied” column.
The FX Swap column in Figure 4.18 shows the market and the FX Swap Implied column shows the theory. They
are not the same! This is interesting and very troubling. The market quote in FX Swap and the calculation in FX
Swap Implied differ by the amount in the right-most column. Can the market be quoting a completely wrong
forward rate?
 The mystery is easily explained when we account for supply and demand in this particular market. It turns out that
the source of our error comes from the interest rate curves used for euros. If you want to appropriately calculate the
forward and match with the FX Swap column, you have to change the interest rate curve you are using for euros.
The original curve might look like that in Figure 4.19.
Figure 4.19 shows the Euribor interest rate curve. The more appropriate curve to use may be the so-called
EURUSD basis curve (see Figure 4.20). With this curve for the euro interest rates, the calculation of FX Swap
Implied looks different (see Figure 4.21).
 Figure 4.21 shows the calculation of FX Swap Implied after the euro interest rate curve has been changed to the
EURUSD basis curve. Note that the discrepancy, in the last column, is now negligible.
What’s going on here? Did we just move the numbers around and create a phony curve to make the numbers fit?
Does this “basis interest rate curve” really exist? The answer is yes, it does, and not only that, this curve is also used
to price cross-currency swaps.13 Since a cross-currency swap is just a collection of FX forward contracts, it stands to
reason that to eliminate possible arbitrage it must be that both the cross-currency swap and the FX forwards use the
same euro interest rate curve to price. This is really good news, and restores our faith that there is no way to cheat
the market. If you think these are details that are unimportant to the market, think again. These are some of the most
liquid markets in the world, and tiny arbitrage opportunities are hunted down and eliminated.
In the mid-1990s, two large banks merged, one of very many bank mergers at the time. The FX proprietary desk
of the combined bank was run by a certain trader, a UK national who had never been to college but was as sharp as
any PhD on the floor and possessed unparalleled street smarts. He was one of a few survivors from the mass firings
from the legacy bank. He scored very high on the cleverness scale, but not so high on the morality scale. He noticed
that the two banks were slow to merge systems, and specifically the interest rate curves they used to value FX
forwards and swaps. The one system was supposed to act as a risk transfer system to the official second system. One
system used legacy Bank 1 interest rate curve for USDJPY, while the other system, concurrently providing P&L,
used Bank 2 interest rate curves for USDJPY. As it happens, they were slightly off in a consistent way. So, this
trader would buy massive, and long-term, USDJPY swaps in one system and do the opposite in the other system. No
risk system caught this because the net position was negligible. These trades resulted in profit every day. Over a
period, he accumulated US$60 million in fictional profits. It was purported at the time that he got real seven-figure
bonuses on those fictional profits (not sure if the rumours were true) before he was caught. Remember, at that time
most of the bonuses were vested quickly and there were no clawback provisions. When the bank found out, it had to
re-state earnings, report it to the Fed, explain what happened to the Securities and Exchange Commission (SEC),
assure everyone it will never happen again and fire the trader. However, it was never able to jail him.
 Since we’re reminiscing, there is one more story that will show you how important this theory of interest rate
arbitrage is. Again in the late 1990s, in addition to my regular structuring job, I was privileged to work with two
other colleagues on a fund of funds the bank had set up to invest around US$500 million with hedge funds – for
profit, but also to encourage shops to trade their FX with our bank. We would make investment decisions
collectively, but often do the initial evaluation of the hedge fund individually to be efficient. On one of my trips to
Texas, the head of the fund of funds told me about a fund he was looking at that had a great track record and had an
FX-related strategy. He suggested that, since I was in Texas, I could go and kick the tyres, be sure they have an
office and that they are staffed appropriately. In general, we used to do a commonsense check before giving anyone
money. When I got there, it all looked fine and they told me that the way they made money was to instantaneously
trade in the interest rate and FX forward market so fast and so efficiently that they arbitraged the market and really
had no risk.
At that point, I should have walked out because I knew the formula we shared with you above (see Figure 4.8)
and I knew the market. That just did not happen back then. Anyway, I was on my way to suspending my doubt until
they took me to a conference room and told me they were linked wirelessly to the markets, and that what I was
seeing were live prices dancing on a laptop screen. I literally looked under the desk, felt the computer for wires and
found none. I could not believe that a computer could be wirelessly connected to the markets. Remember, this was
around 1997 – that’s how fast and how far we have come. When I got back, I recommended that something was
fishy about the place, and we should not invest. A few months later, the head of the group stopped me in the hallway
and said: “Great call on that investment.” He showed me the front page of the Wall Street Journal, that left-hand
skinny column that used to have interesting people stories. The fund I considered investing in was closed down by
the authorities for fraud. Of course, you cannot cheat the market, formulas actually work. I made the right decision
for the wrong reason: I missed the fraud that should have been obvious and worried about the computer without
cables! Wireless connections … who would have imagined such a thing?
Okay, back to business. All the mathematics are internally consistent and the basis interest rate curve is routinely
used to price both FX forwards and cross-currency swaps. However, what are those swaps? In a standard cross-
currency swap, you receive (pay) Euro Interbank Offered Rate (Euribor) interest rate on a euro notional and you pay
(receive) US Libor on a corresponding US dollar notional equal to the euro notional times the EURUSD spot rate at
the time of execution of the deal. At the end of the swap, you receive the euro notional and you pay a US dollar
amount equal to the euro notional times the EURUSD forward rate at the time of execution of the deal.
Simply put, a cross-currency swap is an instrument that allows a European investor to borrow some euros at the
Euribor rate (floating rate), sell them for dollars and lend the dollars at US Libor rate (also a floating rate). At the
maturity date, the investor receives back the dollar amount lent and will exchange it for the euro amount. The cross-
currency swap will take care of synthetically – creating, for the benefit of the European investor, a euro-
denominated investment out of a US-denominated loan.
In real life, this little game can be very attractive for a European asset manager who is possibly facing negative
yields in bunds (the German sovereign bonds). If yields on US$-denominated debt instruments are higher, the asset
manager would like to invest the assets in the purchase of US debt instead of German debt. The problem is that,
upon maturity of the debt, the European asset manager would have to sell the dollars and buy euros to repatriate the
proceeds, but in the meanwhile the exchange rate might have changed. In particular, the euro might have
appreciated, which would result in a smaller amount of euros repatriated. Therefore, the risk in the FX market might
eat up and completely annihilate the yield gain.
To solve this problem the European asset manager can make use of a cross-currency swap, just as we saw, to
completely eliminate the FX risk in the transaction. You can see a brief description of this in Figure 4.22. Note that
we have broken down the whole transaction into its standard constituents. This makes the explanation more boring
but has the advantage of only including very basic instruments. A straight arrow means a fixed-coupon cashflow and
a wavy arrow means a floating-coupon cashflow.

Figure 4.22 shows how a European asset manager (who is counterparty to all four trades in the picture) can
synthetically invest in a US$-denominated bond and swap all the cashflows into euro cashflows with the help of two
vanilla fixed-to-floating swaps, one in euros, the other in dollars, plus one standard float-to-float cross-currency
swap. Standard interest rate parity would dictate that the cross-currency swap be done at flat. In reality, due to
supply and demand, a “basis” needs to be added to the US dollar Libor payments. This is added in the leg called
“3M US$ Libor + Basis”. In standard lingo, the basis is a negative number and is added to the euro floating rate
payments. Here, to simplify, we make the basis sign positive and add it to the US dollar floating rate payments,
which is equivalent.
 The details of the cross-currency swap can be seen in Figure 4.23. This shows a standard cross-currency swap,
which is float-to-float. Note that there is a spread of 28.746 basis points to be added to the US dollar leg (in the
middle of the page (①)). According to standard economic theory, such a spread should be zero, but supply demand
in the financial system makes it different from zero. This spread even has a famous name: the “cross-currency
basis”.

In Figure 4.23 also note below “Market”, in the “Fwd” row, that the “EUR (vs. 3M Euribor) basis” curve is used
(curve number 92). The use of this specific euro interest rate curve makes the present value of the cross-currency
swap, inclusive of the 28.746 basis point (bp) spread, equal to zero, which is exactly where it would trade in the
market. Note the values of “Leg 1: NPV” and “Leg 2: NPV” are equal. If the euro interest rate curve is changed to,
say, the Euribor interest rate curve, then the value of the spread needs to be adjusted to zero to have the present
value of the swap equal to zero. However, this use of the Euribor interest rate curve would result in a mis-pricing of
the cross-currency swap, because in the real market the spread does not trade at zero.
Let’s review the main concept. In the market, the presence of supply and demand is such that a spread, called the
cross-currency basis, needs to be added to cross-currency swaps. Standard economic theory would dictate the
absence of spread. For the same reason, the euro basis curve needs to be used when computing FX forward prices so
as to match the market.
The next question we need to answer now is “why”. Why is the cross-currency basis present? Why is standard
economic theory failing? Supply and demand is a catch-all for many factors impacting the basis, including differing
access to the interest rate markets by different players. In the next section, we will review history and examine a
specific situation that we hope will illuminate how these stresses can occur.

The big squeeze

Figure 4.24 shows the cross-currency basis in Australian dollars, euros and Japanese yen for one-, three-, five- and
10-year tenors. The amounts shown here are in “basis points running” or, in other words, spreads that need to be
added to the US dollar leg of the float-to-float cross-currency swap. To say this in yet another way, you can receive
US$ Libor plus 80 basis points and pay ¥ Libor for 10 years (column ①) all the way to the right), and this cross-
currency swap would have a zero present value.
 It is very fascinating to realise that, before the global financial crisis of 2007–08, the cross-currency basis was
immaterial and very close to zero. It would go out of whack episodically around year-end and only in certain risk-
unwinding periods. What caused the basis to assume great importance? We need some history to continue our
investigation.

Figure 4.25 shows the one-year EURUSD cross-currency basis from 2005 to 2017. See how the basis was
immaterial up to 2007, and note that it again reached a multi-year high (in absolute value) point in the autumn of
2016. What was special about that time?
History tells us that, in the years after the dot-com bubble, appetite for risk in the US increased considerably, and
a bubble in housing prices developed. Many real estate-backed debt instruments were issued, in no small part for
foreign asset managers to buy. In some cases, they were guaranteed by residential mortgages of very low credit
quality, called “subprime”. In 2007, this bubble started to burst. For an entertaining overview of those events watch
the film The Big Short or read the book by Michael Lewis (2011). A great deal of the stress was felt by banks that
funded those speculative asset-backed debt instruments with rolling short-term uncollateralised borrowing, usually
indexed to Libor. In other words, many banks that lacked a substantial basis of deposits, such as pure investment
banks, borrowed at the one-month Libor rate and purchased 10-year US subprime mortgage-backed collateralised
debt obligations. Every month they had to renew the one-month Libor borrowing, and were exposed to the
possibility that their perceived creditworthiness might suffer and Libor might rise, perhaps precipitously. In those
days, banks were assumed to be very solid borrowers, with no questions asked.
This laissez-faire type of environment came to an end with the collapse of the mortgage-backed market in 2007
and 2008. At that time, the willingness to lend to banks evaporated and Lehman Brothers declared the largest
bankruptcy in US history. This event was inconceivable given the sterling reputation of investment banks at that
time. The demise of Lehman changed the market’s attitude to uncollateralised lending to banks, especially to non-
deposit-taking institutions, forever.

A typical bellwether of financial sector creditworthiness was (and is) the Libor–OIS spread.14 Look at the spread
in the bottom part of Figure 4.26 – note how immaterial it is before 2007, before it explodes during the financial
crisis. As mentioned, the short-term uncollateralised Libor borrowing and funding of long-term risky debt
instruments needed to be rolled at a time when Libor was materially higher. This caused non-US domiciled banks to
seek US dollar funding at all costs, creating immense opportunities for short-term interest rate desks at US banks to
profit from the desperation of other banks that needed to fund dollars. I remember those traders spending night after
night on the trading floor and sleeping in shifts on the only couch, located in the office of the global head of FX.
That was tough, I am sure, but those traders concentrated the P&L of many years in September and October 2008,
and in many cases were promoted to managing directors because of that P&L.
The top panel of Figure 4.26 shows the three-month Libor rate (in white) and the OIS rate (this is the line
indicated by the number ①). In the bottom panel you can see the spread, which was immaterial until the global
financial crisis. We can also see a spike at the time of the first Greek financial crisis rescue package, in November
2011, and another in October 2016, a time that is the focus of the present discussion.

Now, let’s take a close up of the spread in Figure 4.27. This figure shows in the top panel the spread between the
three-month Libor rate (in white) and the OIS rate (the line labelled with ①). In the bottom panel you can see the
spread. Here we show 2016, and in particular September and October, a period during which the spread reached a
multi-year high.
Here’s an interesting question. Why did two, seemingly unrelated, quantities – the Libor–OIS spread and the
cross-currency basis – widen to multi-year highs at the end of 2016? There are two explanations, of which one is
very obvious, the other not so much.
First, the rise of Libor is easy to explain due to the election of Donald Trump to the presidency of the US. Right
after the election, the dollar and US rates rallied substantially, and continued to rally over year-end. This was the
obvious factor. However, the Trump trade would not explain the peaking of the Libor–OIS spread before the US
election. For that, we need to turn to a corner of the short-term debt market in the US, namely the money market
funds arena. A money market fund is a fund that invests in short-term uncollateralised debt, usually commercial
paper, government bills or loans. The issuers can be US corporates, the US government or other financial
institutions, American or foreign.
In October 2016, a previously announced reform of some of the rules governing money market funds was
enacted. Among the changes, funds that only contained sovereign debt would be called “government” and those that
also contained corporate debt would be called “prime”. Also, funds would have to report a daily-computed net asset
value (NAV). Previously, their mark-to-market (MTM) was just parity all the time. In addition, extra fees would be
levied in case the percentage of the fund that can be liquidated within a week fell below specific thresholds. All
those changes did not really represent seismic changes, but were enough to deter investors away from prime funds.
Finally, a further deterrent was a potential requirement to re-classify the holdings of prime funds into a different
financial instruments category for accounting purposes.
 Figure 4.28 shows the assets under management at Government and Prime money market funds before and after
the reform of October 2016. As a result of the reform, funds migrated towards government funds at the expense of
prime funds, and in addition yield on prime funds rose dramatically, almost tripling.
Figure 4.29 shows the increase in yield of money market funds after the reform of October 2016. As a result of
the rise in yields, the appetite of corporations for issuing commercial paper, the typical ingredient of corporate debt
in prime funds, collapsed.
Figure 4.30 shows the drop in corporate commercial paper issuance after the enactment of the money market
funds reform of October 2016. Interestingly, corporate commercial paper was a large component of those money
market funds, but another natural category of users of those funding vehicles suffered as a consequence of the
reform. To discover who that was, we need to take a step back. By now you are thinking, “when did this book stop
being about FX? What’s all this money market stuff got to do with FX?” We’ll get there, a little patience is needed –
it will be worth it.

The Japanese and their yen

There are three main factors that make the Japanese economy different from the US economy: high savings rate, low
interest rates and a large export sector. As a consequence, financial institutions such as banks and insurance
companies in Japan have a huge appetite for higher yields.
Figure 4.31 shows in curve number ② the US Treasury yield curve and in number ① the Japanese JGB
(sovereign bonds) yield curve as of October 2016. Note that the Japanese curve has negative yields up to almost the
15-year tenor. Since the late 1990s, US sovereign yields have been higher than their Japanese counterparts. As such,
they attracted funds from Japanese asset managers eager to boost their returns. There is, however, a small
complication – they played what looked like an easy interest rate game, using a difficult FX game. There is no free
lunch on Wall Street. Boosting the yield could be done at the expense of creating currency risk on the repatriation of
the US dollar-denominated coupons and even more risk for the repatriation of the principal. The risk could be
eliminated using, you guessed it, our old friend the cross-currency swap, which fits the patterns of the Japanese asset
manager investing in US dollar cashflows like a glove.
 Figure 4.32 shows the mechanics relative to a Japanese asset manager or bank hedging the holding of US
Treasury bonds or corporate bonds or loans. The bank is funding with local deposits in this example. Due to this
structural appetite for yield in the Japanese economy, a natural and secular demand for the cross-currency basis was
created. With each new cross-currency swap traded the basis moved a little bit further, until the yield pick-up
afforded by holding US Treasury bonds had completely disappeared.
Figure 4.33 shows in line number ① the yield on Japanese 10-year JGBs. In white is the synthetic yield achieved
by Japanese investors who buy US 10-year Treasuries and utilise a cross-currency swap to synthetically turn the
investment into a Japanese yen cashflow investment. At the end of 2013, the yield pick-up was about 2%, but that
pick-up was completely annihilated by the summer of 2016, when the events of our story unfolded.
 Let’s start connecting the dots. Demand for dollar funding increased due to various factors, including the money
market reform in the US, partly due to the rise in US dollar Libor following a more hawkish stance by the Fed, and
potentially also the election of Donald Trump. At that time, Japanese asset managers tried to fund dollars using the
cross-currency swap. However, that did not work due to the natural imbalance intrinsic to the Japanese economy that
altered the demand and over time negated the yield pick-up. This demand for dollar funding could be exploited by
those lucky enough to be able to rent out their dollar funding. Now let’s turn the story around: if it becomes harder
and harder for Japanese asset managers to invest into the US, then US asset managers, being on the other side of the
deal, receive the cross-currency basis when investing abroad. They could generate yield pick-ups, as shown in
Figure 4.34, in the case of three-month funding.
 Figure 4.34 shows in line number ① the three-month negative yield on Japanese T-bills, in line number ② the
yield on three-months US T-bills and in white the yield on an investment in three-month Japanese T-bills by a US
asset manager that swaps the cashflows to US dollars using a cross-currency swap. There is a yield pick-up of about
66bp due to the fact that the US asset manager benefits from the presence of the cross-currency basis. The native
Japanese yield is negative, but if the basis is large enough this turns into a winning trade.
This structural imbalance of funding demand is not exclusive to the Japanese economy. In Europe, the
introduction of quantitative easing while the US was about to assume a less dovish rates stance generated another
supply/demand imbalance. As the ECB started to purchase euro-denominated sovereign and corporate bonds, US
corporates quickly noticed that it might be advantageous to issue bonds that could be sold to the ECB. Of course,
one of the conditions was that those bonds needed to be euro-denominated. The fact that the European financial
system is less reliant on corporate bonds than it is on bank loans amplified the price impact of the ECB demand,
compounded by a scarcity for sovereign bonds that also surfaced.
Figure 4.35 shows how a US corporation could issue euro-denominated bonds and synthetically swap them into
US dollar instruments using a cross-currency swap and two fixed-to-floating swaps. Note that in this case the US
corporation pays the cross-currency basis, and therefore needs the basis to be small enough so that the funding
advantage is not negated.
 Figure 4.36 shows a database query that was run to confirm that there was an increase in euro-denominated bonds
by non-European corporations. The amount issued quarterly increased in the run up to autumn 2016 (column
labelled “Total”). Remember this issuance, if swapped back into US dollars, tends to exacerbate the cross-currency
basis.
Note the domicile of the issuer in different columns, and that the amounts started to rise at the end of 2013 and
peaked at a 113-billion dollar-equivalent in the second quarter of 2016.
Many corporations considered taking advantage of this natural demand by the ECB, but only some of them
ultimately decided to do it. Sometimes the cross-currency basis just ate up all the advantage, at other times the credit
appetite of European asset managers for the issuing name was not as strong as the appetite in the country of
domicile, while at other times good old risk aversion created the usual “analysis paralysis” phenomenon that delayed
action until the market turned unfavourable.
Corporations also debated at times whether the proceeds of euro-denominated debt issuance should be kept in
Europe and used in the local European operations, or whether they should repatriate into dollars. In the former case,
a cross-currency swap would not be necessary.
To determine whether cross-border issuance is popular, and if it will eventually increase the cross-currency basis,
would be conjecture. We need to figure out what the credit appetite looks like in Europe versus the US, and the
corporate need for funds in Europe versus the US. To test this conjecture, we could plot the spread between US and
euro corporate credit spreads versus the cross-currency basis. If we do that, we can see that in 2014–16 the spread of
spreads was negatively correlated with the cross-currency basis. This tells us it is plausible that the non-euro
issuance of euro-denominated corporate debt might have driven the EURUSD cross-currency basis to be larger.
Figure 4.37 shows, in the top panel, US corporate spreads (in white) and Euro corporate spreads (in line number
①). In the bottom panel, the difference between those two spreads is shown in line number ② and the EURUSD
one-year cross-currency basis is shown (depicted here as a negative number, meaning that the party that pays the
euro floating rate pays less than euro Libor; this is the same as our previous charts where the party that receives US
floating receives more). In 2008–09, and then again in 2014–16, the two quantities move towards each other
spectacularly.
To this point, we have focused on corporations and on non-sovereign asset managers. In reality the tentacles of
this monster extend all the way up to sovereign wealth funds and central bank reserves. The corporate interest in
exploiting the ECB bid for euro-denominated corporate bonds demonstrates how impossible it is to separate
corporate from sovereign debt dynamics. In fact, the ECB was clearly directed towards more easing, as was the
Bank of Japan, while the Fed at the time was signalling steps to raise US Treasury yields. This caused an
unprecedented divergence in central bank policy that had not been witnessed for at least 20 years.

With US Treasury yields poised to rise (a familiar story, one that has been recited often since Ben Bernanke’s
“taper tantrum” of May 2013 and then the dramatic rise in the dollar and in FX volatility starting in the summer of
2014), and US-based sovereign bond reserves suffering losses, appetite from various non-US, US Treasury holders
started to wane in summer 2016. This caused a rise in yields and a decline in bellwether bond indexes. It is difficult
to isolate cause and effect. We suspect it played into the mindset of rising Libor and declining perceived banking
sector creditworthiness. Recall from previous comments how strong the link has been between higher Libor rates
and demand for US dollar funding. Of course, now you know that demand for US dollar funding is the major driver
of higher cross-currency basis!
 Figure 4.38 shows in line number ① the “race to zero”, or the secular declining trend in US Treasury yields since
the mid-1990s. The Bloomberg Barclays Aggregate UST index is in line number ②,and the foreign holdings of US
Treasury bonds in white. In the middle of 2016, for the first time foreigners have scaled down their holdings of US
Treasury bonds, which coincided with a large spike in US Treasury yields and a corresponding decline in the index.
All finance is connected. Many have speculated that some of the decline in foreign holdings of US Treasuries has
coincided with large emerging-market central banks having to defend their currencies by liquidating parts of their
US dollar reserves. The elephant in the room, in this case, is China, the leading foreign holder of US Treasury
bonds.

CONCLUSION
Whew! This was a monster of a chapter with a lot of information. We’ve reviewed the basic instruments in FX, how
they trade, and the theory and the mathematics behind the prices. FX basic instruments are not the most complex,
but they lay the foundation for the most liquid, least regulated market in the world. This foundation has allowed the
creation of some of the most exotic derivative instruments of all asset classes. We will examine some of those in the
third part of this book.
This chapter progressed through FX spot, forwards, NDFs, swaps, and finally an extended conversation on cross-
currency basis. As the discussion progressed, the level of complexity was ratcheted up, and we expanded the
universe of interested participants. We examined really unique situations such as China and, in the final section, we
took you on one of the widest-ranging domino effects within the global financial system. This sequence of events
touched markets that have nothing to do with each other in normal circumstances. This was the perfect story to
showcase the ubiquitous nature of FX, and more broadly the mysterious intricacies that make financial history so
intriguing.
You are now an FX expert. Mortgage your house and your firstborn, put some money in one of the online trading
portals and go for it. By the way, depending on the shop you’re dealing with and the size of your investment,
leverage can be 50:1, 100:1, or even 200:1 for smaller investments. So, contingent on how big your house is and
how hard your kid can work, you could be trading US$100 million positions versus the euro in no time. Although,
just to be sure you’ll come out ahead, maybe you should read a few more chapters first …

1 Taken from Wikipedia but the share of the market comes from the BIS study that we explored extensively in Chapter 2.
2 Global Foreign Exchange Committee (GFXC), “FX Global Code”, p 1.
 3 New York Department of Financial Services press release: “NYDFS Announces Barclays to Pay $2.4 Billion, Terminate Employees for Conspiring to
Manipulate Spot FX Trading Market”, May 20, 2015 (available at http://www.dfs.ny.gov/about/press/pr1505201.htm).
4 GFXC, “FX Global Code”, p 57.
5 This is also called “spoofing” – for example, calling a number of banks and asking for an offer price good for 100 million, and then, at the same time,
dealing with all of the banks. If the banks think 100 million is the full amount, they will quote a price such that they will reasonably cover their sale of
100 million at a breakeven rate or better. However, if instead it turns out that the total amount is more like 500 million, the market will move higher
and the market-maker trader will cover the sale at a much higher price and lose money. After all the banks are done, if the banks have covered their
sales the price will be much higher, the “spoofer” will be able to then sell at a higher price, thus closing their position at a gain.
6 GFXC, “FX Global Code”, p 17.
7 United States v Citicorp, Plea Agreement, May 20 2015 (see https://www.justice.gov/file/440486/download).
8 White-labelling means that every time a client of Small Bank asks for a price, Small Bank calls Big Bank for the same, and then potentially alters that
price in its favour and passes it on to its client. So Small Bank just passes trades back and forth between its clients, takes a margin, and passes on to
Big Bank to risk-manage.
9 Under the EU’s MiFID II unbundling regulations, the market is moving to a “paid” model for the provision of some ancillary services (see Mahmud
and Williams, 2016).
10 GFXC, “FX Global Code”, p 26.
11 USDCNY or RMB, which is the official Chinese currency, the yuan. It stands for renminbi or the people’s currency. The official symbol used is the
same as that for the Japanese yen, but in the professional markets that symbol is reserved for “JPY”.
12 We are indebted to Colum Lane at Pfizer for his insight on this fascinating subject, about which Colum is a total expert. Credit goes to him for all the
insight on this topic, all blame goes to us for any errors.
13 A cross-currency swap is an agreement to a series of FX transactions between two parties. They agree to exchange the net present value of principal
and/or interest payments of a loan in one currency for the equivalent amounts in a second currency.
14 The overnight indexed swap (OIS) is an interest rate swap where the floating payment is based on the overnight Fed Funds Rate. The difference between
Libor and the OIS rates is inversely related to creditworthiness.
 5
Trading Floor Dynamics

Modern trading floors are a thing of the past, to paraphrase the famous adage by Oscar Wilde. This rings true when
we examine how historically the sell side (banks and broker–dealers) developed a functioning sales and trading
operation. If you have ever visited a bank trading floor, you will know they have had an open floor layout for many
decades. There are no cubicles, no individual desks. It was an open plan, open outcry environment before open plan
ever became fashionable across other work environments. However, the traditional bankers, the people that make
the loans, were always in offices and their analysts in cubicles around the relationship manager’s office.
The main reason why corporate finance bankers have offices and traders have open trading floors is that bankers
need privacy. The transactions they work on operate behind information barriers (previously referred to as “Chinese
walls”), and often the individuals are required to sign a statement to “cross the wall” to participate in a transaction.
On the other hand, traders need to communicate prices to their salespeople and other traders, and these prices need
to be as efficiently accessible as possible. Therefore, the concept of an open layout trading floor was born, and had
become established by the late 1970s.
Now that we have the basics of theory and function of the FX spot, forward and cross-currency markets from
Chapter 4, we can start to understand what drives behaviour on the trading floor. This chapter will examine the
motivations and pressures that individuals experience sitting on the trading floor, typically for 10–11 hours a day,
eating lunch at their desks and practically in one another’s lap. They are trying to service their clients’ FX needs, but
the reason they are there is to generate revenue for their bank and then demand a huge (by average workers’
standards) bonus. We will examine some of the conflicts of interest inherent in the execution of FX trades, and
explore how competition may lead to a worse price for clients. Following that, we will turn our attention to the
drivers of individual behaviour for those running such trading operations, and the compensation schemes that have
made headlines over the years. We surmise as to the reasons trading floor staff are very well compensated and what
kind of behaviour that engenders. The fact is, any contemporary discussion of the FX market cannot avoid a
discussion on ethics.

MAKING AN ETHICAL PRICE, IN THEORY

Here’s a concept you may not have considered: a market-making trader in FX does not have to maintain
“inventory”. They can happily make prices in euros even when their trading book is not holding euros. This little
detail has important ethical consequences unknown, for example, to traders of sovereign bonds.
Ideally, when a client calls a bank and asks for the price to buy a corporate bond, the bank trader will already have
a quantity of the bond on their books and will quote a price. This is the ideal, and frankly it is an idealised situation.
In this case, the bond was already sitting in the inventory of the bank trader. In modern finance, even if the bond is
not held by the trader, the trader could borrow it from another bank (at a cost) and then sell it to the client, or the
trader could sell it to the client first and then worry about buying the bond in the market before the end of the trading
session. There is another possibility as well, which is that the trader would quickly buy the bond from another bank
using either the telephone or, more likely, an electronic trading platform. Only after this is done, and the trader
knows how much it cost them, would the trader make a price to the client.
This last way of doing business would be called front-running, as the initial purchase of the bond by the trader
could move the market higher. The subsequent sale of the bond to the client would most likely occur at a higher
price than the price the trader paid for the bond, and has long been against the rules. In equities, the rules are crystal
clear. According to Nasdaq, front-running is:
Entering into an equity trade, options or futures contracts with advance knowledge of a block transaction that will influence the price of the
underlying security to capitalise on the trade. This practice is expressly forbidden by the SEC. Traders are not allowed to act on non-public
information to trade ahead of customers lacking that knowledge.1
That’s in equities, but this is FX. When showing an offer (making an ask price) for a EURUSD spot transaction to a
client, the trader will often start buying euros just to gauge the liquidity in the market, while simultaneously
negotiating the transaction with the salesperson, who is on the telephone with the client. This changes the odds at the
casino. For example, the client will say they want to buy €10 million versus US dollars. The market at the time is
1.2305/1.2307. The trader then very quickly buys €2 million at 1.2307 on the Electronic Broking Services (EBS)
interbank electronic trading system, and the price moves to 1.2307/1.2309. This is a big move for just €2 million, so
the trader shows an offer price of 1.2313 to the client, because realistically they will have to trade €2 million another
four times, at the rates 1.2311, 1.2313, 1.2315 and 1.2317, so that the average rate will be 1.2313. These numbers
and amounts are exaggerated for the purpose of this example. Given the market’s liquidity, in real life transacting
€10 million versus US dollars at almost any time and under most circumstances happens seamlessly and with
absolutely no change in the market price. However, if the trade was for a few billion, that would call for a well-
planned strategy. Below are some practical examples of the spreads that exist for transacting different amounts in
euros.
Figure 5.1 shows (bottom right) the market spread for different amounts (sizes). If a client asks for a market in
€10 million notional, then the market is 1.23055 bid and 1.23071 offer. However, if the client asks for a price in €30
million then the market is 1.23041 bid and 1.23084 offer. The spread in this case is 7.9 pips wide, whereas in the
first case it was 1.6 pips wide. These are aggregated quotes from a pool of banks streaming electronic prices, so
asking just one bank will almost surely result in wider spreads.
Assume the trader shows the client 1.2313 and the client has says “done”, meaning the transaction is confirmed.
The trader will now have to purchase the remaining €8 million and hope that the average is 1.2313 or lower. It could
end up either way, but this is the risk management part of the job the trader is paid to do. Here is the rub – what if
the trader is, at the same time, able to take a position instead of just being the intermediary between the client and
the broader market? What if the trader is a market maker and a proprietary trader at the same time? Then the trader
could have claimed that their original purchase of €2 million was for their own proprietary position and had nothing
to do with the client’s request. Therefore, with a similar calculation, they would show the client an ask price of
1.2315. Once the transaction with the client is finished, and assuming that the market behaves exactly as the trader
had forecast, the market will be at 1.2317/1.2319. At that point, the trader can sell to the market the €2 million that
they were holding in their proprietary trading book at 1.2317. They bought at 1.2307 and sold at 1.2317; not a bad
day at the office.
Funding trading operations from widows and orphans
This type of behaviour is what former Federal Reserve chairman Paul Volcker was seeking to eliminate when he
proposed his now-famous Volcker rule. It was first proposed in January, 2010 as part of the Dodd–Frank Wall Street
Reform and Consumer Protection Act, it came into effect in July 2015. He and many others felt that this way of
making a buck was not ethical. Some would argue further, that banks should not take proprietary positions as it
amounts to gambling with the house funds, using money from “widows and orphans”. The reason is that bank
liquidity can originate from borrowing from other banks, but, crucially, it classically originates from retail deposits.
Banks typically would leverage any funding from deposits or other sources, and take highly leveraged proprietary
positions. This can incur much higher risk and could potentially make a bank insolvent, jeopardising its ability to
repay deposits. This argument continues by saying that the ability to save money is integral to the functioning of a
modern society, and should therefore be protected from risks generated by aggressive bets placed by proprietary
traders. The main example cited in this context is Lehman Brothers, which went bankrupt during the global financial
crisis of 2007–08. The irony is not lost that Lehman Brothers was a wholesale investment bank and did not take
retail deposits. However, regulators were concerned with the potential for “contagion” – that is, problems at one firm
spilling over to a second firm (such as a retail bank) due to exposures that the second bank might have against the
first.
Another more poignant example is that of Northern Rock, a UK bank (mortgage lender) that went bankrupt in
2007, and which constitutes the most recent example of a classic “run on a bank”. We recall the financial reporter at
a respected media outlet discussing on television the impending financial danger for those who had their mortgages
with Northern Rock. Of course, a mortgage is where the mortgage holder owes the money to the bank (not the other
way around) so if anything the mortgage holder would have gained from the bankruptcy; but hysteria was not far
behind and the writing was on the wall. Long lines of depositors formed outside branches in the last days when it
became clear to the public that depositors’ cash was in danger. This was the first time in 150 years that people in the
UK had witnessed a run on a bank (not including that scene from Mary Poppins!). The Bank of England had to
come to the rescue as lender of last resort, and Northern Rock was eventually taken into public ownership and later
sold.
The prospect of losing money saved into deposit accounts is terrifying to modern society, which is why in the US
bank deposits are generally covered by mutually financed insurance up to US$250,000 (for each account type, at
each institution).2 Numerous other countries also operate deposit insurance schemes; under some schemes, it is
possible for deposit amounts larger than a per-account threshold to be split so that each piece is less than the deposit
threshold, and hence the total is fully covered by the insurance. As an indicator of the level of risk sweeping the
financial system during the global financial crisis, many investors found that several institutions with which they had
placed deposits had either ceased trading or were being acquired during and after the crisis.
Of course, the concept that the government should protect deposits and avoid any run on the bank, at all costs,
harkens back to the Great Depression of 1929. It was key to the now largely repealed Glass–Steagall Act, the
common term for four relevant provisions of the US Banking Act of 1933. We will mention the Glass–Steagall Act
again, but suffice to say for now that it mandated the separation of commercial and investment banking, so that
deposit-taking banks could not engage in securities trading and related activities.

Risk taken must be commensurate with the potential reward

Returning to the trading example set out above, many on Wall Street still believe that the behaviour described in the
example is totally ethical. The trader (on behalf of their employer) is bearing a huge risk. In our example, the euro
behaved as expected, but that is not always the case. What if the euro fell versus the US dollar, causing the trader to
lose a lot of money? In that case the front-running part, where the trader buys 2 million for themselves before
making a price to the client, would have backfired. Additionally, the trader may underestimate the impact of buying
2 million euros. Moreover, the bank has invested capital to set up its trading operations and it is providing for its
clients’ needs. Banks need to be paid (ie, gain a return on invested capital) for providing liquidity to the world. This
thinking was obvious in many of the statements made by traders at their collusion hearings in the UK and the US.
Besides, this is effectively how business had been conducted in the FX markets since the first currencies were freely
floated in the early 1970s. Many would argue that to now stop this decades-old understanding and put these traders
in front of juries that have no context or knowledge of what the trader’s job entails may be simply unfair. The
appendix to this chapter gives the reader a front-row seat at a 2018 indictment where these topics are dealt with in a
court of law.
This is a very divisive topic. Regulation, technology and the resulting transparency have, for the most part, done
away with this type of dealing. A few years ago, regulators concluded that knowing client-flows gave those traders
an unfair advantage, and so the practice began to change. Banks were allowed to have proprietary traders but they
had to sit in a separate area, with a physical separation, called “the Chinese wall”, similar to that separating the
investment banking/advisory areas from the trading areas. Duly separated, “prop” traders were unable to hear all the
market flows and prices that were being directed to the clients of the bank. Eventually, it proved too difficult to
effectively police this information separation and the Volcker rule finally established that banks cannot have risk-
taking proprietary trading operations and must limit themselves to market making (or “client facilitation”). At the
time of writing, that was still the case. The new US administration is hinting that it will do away with Volcker and
we’re back to the future; this remains to be seen. The Volcker rule was not a new concept. As mentioned, after the
Great Depression the Glass–Steagall Act of 1933 was also intended to protect savings and deposits by separating the
deposit-taking and loan-making functions of the banking industry from the more free-wheeling, risk-taking
proprietary trading and underwriting activities. Over the years, starting in the 1960s, various pieces of this
legislation were rendered virtually void, until in 1999 the Act was repealed. It did not take long before it was needed
again.3
Since the early days of free-floating FX dealing, trading floors were places that understood that proprietary
position taking was an integral part of making prices to clients; some traders were purely proprietary traders, but
most were market makers in one trading book and also ran proprietary trades in their “prop” book. Some firms went
so far as to encourage traders to have “skin in the game” by way of a personal dealing account.
As regulation and common sense shut down the personal trading accounts, the ability for those on the trading
floor to generate profits and prove their worth continued. At many major banks, the heads of FX and rates trading
used to also run proprietary, or “strategic”, positions. Typically, they would not lose money in those positions
because they got first view of the flow. However, even with that purview, this was not always the case. Numerous
senior managers lost large sums in their prop books while the rest of the traders dealt with client deals. At one large
New York bank, the biggest fish of all was the head of FX, a man who oversaw about US$3 billion in revenues per
year. One might think that top management would concern themselves with managerial tasks; after all, one would
not expect to see the CEO of Coca-Cola to go down to the assembly line and start helping to bottle the soft drinks.
Not so in investment banking, in fact quite the opposite. In banking, those who got promoted to management were
the highest producers, but being a successful trader didn’t necessarily mean you had the skillset to be a great
manager. Therefore, the people who get promoted were very good at trading and enjoyed it, and they may not be
have been very good at managing people, and especially other traders. They did not tend to enjoy it, but accepted it
as a way to get promoted and progress in their careers. Since they still enjoyed trading, they continued to trade
through their proprietary trading books and, because they were the bosses, their risk limits were very large. And so it
was that the head of FX was often the biggest prop trader of all. Now this distraction has been largely removed, and
the individuals in these roles can perhaps be more focused on managing and supervising.

Competition does not always lead to the best price

Front-running, as we mentioned, is the behaviour described in the previous example: a trader tests the market even
before making a price to the client. The client wants to buy €10 million, the trader buys €2 million, watches the price
rise by two pips, and then quotes a price to the client. So, front-running can occur where the trader is allowed to take
proprietary positions and claim any benefit from the initial test trade was not for the client since they did not make
the client a price yet, but it was for their own proprietary position. The situation becomes more complex if the client
is on the telephone with five banks, and is trying to get the banks to compete by showing the best price to the client.
The trader now knows that if they are not selling the €10 million to the client, then some other bank will. This
changes everything. Let’s see how.
All of the five banks are competing to win the €10 million trade. Our trader will now forget about the €2 million
test trade and directly buy €10 million. If some other bank showed the client a better price, the client will trade with
that bank, which will then have to find the €10 million somewhere after they sold to the client. It is a game of
chicken, and the loser is the client. The client thinks they are smart because they put five banks in competition, and
will surely get the best possible price. What’s worse is where the client is using an electronic trading platform that
records all competing prices. In case of an audit, the client will be able to show that they dealt at the best price
among five.
The rub is that we are talking about the best price only at that moment. What might very well happen is that, at the
instant the client asks for the price, all five banks buy €10 million to hedge their potential risk. This will generate a
market shockwave impact equal to €50 million. Now the market starts to move the rate. This is particularly
exaggerated if we are talking about billions of dollars or in currencies that are not very liquid. At that point, the price
moves higher and only then do the five banks send back to the client their quotes. The client does trade at the best
price when they transacted, but at the best price in a market that just moved higher and against their interest.
 After the client is given the €10 million at a price that reflects the market, there is still the question as to what
happens to the remaining €40 million that the five banks have collectively bought. For those that “lost” the client
trade, if the price stays high then they can all sell them back to the broader market at a profit. If the price returns to
the starting point in a matter of a few seconds, then they should all be able to reverse their purchase and the average
price received should roughly equal the average price paid minus the bid/ask spread, which will be dissipated among
the market makers in the entire market. Principle 11 of the Code states: “a market participant should only pre-hedge
client orders when acting as a Principal and should do so fairly and with transparency”. The Code goes on to give an
example in Annex 1 that helps a trader to apply this in practice by allowing the trader to test the liquidity of the
market by trading a small amount first, so long as they take this amount already dealt into consideration when
quoting the client’s price. Note all the subjectively defined adjectives such as “small”, which of course are open to
interpretation. Interestingly, if the client decides not to deal on the trader’s price, then the test trade would actually
end up in the trader’s book and could result in a loss.
As the complexity of price making and execution increases, ambiguity around what constitutes front-running and
best execution also potentially intensifies. Consider the process when the client gives a limit order4 relating to the
execution of an option(s) and a spot transaction simultaneously (as will be discussed in detail in later chapters).
Suppose EURUSD spot is trading at US$1.03 per euro and the client gives the option trader an order to purchase a
combination of options if the total price (premium) is US$100,000. However, at US$1.03 per euro the price of the
option is currently US$150,000. The option trader calculates that if spot goes to US$1.04 per euro then the price of
the option combination will be US$100,000. Suppose at that point the trader has to execute a few option trades but
will also have to sell €5 million as part of the execution of the combination for the client. The option trader stands to
make a reasonable mark-up, say in this case US$50,000, if the combination is executed, as option spreads are wider
than spot (or cash) spreads. What the trader might be tempted to do at that point is sell the €5 million right away at
US$1.03. Two things can then happen.

Suppose EURUSD goes higher to US$1.04: Then the trader gets their trades executed and will have sold the
EURUSD at US$1.03 instead of US$1.04, and so will have lost money, but a known loss: (1.03–1.04) x US$5
million = US$50,000. This will be offset by the US$50,000 built into the trade structure.
Suppose EURUSD goes lower to 1.02: Now the order is unlikely to be executed. The trader will be disappointed,
but there’s a consolation prize: the short €5 million position has made the trader money because EURUSD fell,
yielding a profit of (1.03–1.02) x US$5 million = US$50,000.

Either way, the trader has hedged themselves. The Code addresses orders, and specifically states in Principle 10 that
“market participants should handle orders fairly, with transparency and in a manner consistent with the specific
considerations relevant to different order types”.

Human interaction is hard to give up

One of the main questions facing the buy side (as opposed to the sell side, the banks) is whether to trade using the
telephone, or to trade using electronic portals (using a mouse click) to execute a transaction. This question often
goes together with the question raised earlier, as to whether it is better to call one bank, ask them a price and then
execute, or to call a few banks at the same time and ask them all a price, and then trade on the best of all the prices
received. This can be arranged using different telephone lines at the same time, or by using different single-bank
portals at the same time, or using one multi-dealer platform where the comparison among different market makers is
built into the software.
Spoiler alert: the correct answer depends on the size of the transaction, the liquidity of the currency, the time of
day, the relationship between the client and the bank, and the complexity of the deal. Despite the Code, unwritten
rules about best trading practices remain, and participants continuously try to execute so as to gain a better price. In
the end, and despite electronification, the FX business remains largely based on reputation and trust. We have
visited end-users and discussed “shopping” their business electronically around a few banks simply because they
now can, at low cost. Frequently, the response is: “No need, I know bank A is really excellent and tight.” If
somebody disappoints with respect to mark-up or falls short of their reputation, the result is reduced business; clients
may threaten retaliation (less business) in cases where the prices are not good enough. A managing director or other
senior person at the bank then has to be on the telephone to apologise, and to explain the reasoning behind their
wider price.
 Let’s explore some of the considerations through examples, starting with what an electronic trading platform
looks like. Figure 5.2 is similar to Figure 5.1, but now we will focus on the different-size buckets. Look at the
bottom right, the market liquidity and spread for different-size trades. Look at the “JPY BGNE Volume Tiers” table.
The first line of that table shows that to transact in US$5 million the market is 106.804/106.851 (bid/ask), so it is 4.7
pips wide, but to transact in US$20 million size then the market is 106.800/106.855 (the third line), which is 5.5 pips
wide. In practice, a hedge fund trader would have asked multiple banks at the same time for their prices, and would
have traded at the best price. In so doing, they would also have sent a powerful signal to all but one of the banks to
tighten their prices, otherwise they will keep losing business.
Figure 5.3 shows 7 seconds of market activity in USDJPY, from 16:58:54 to 16:59:00 on February 23, 2018.
These prices are the mid-market for the USDJPY exchange rate in all sizes.
Electronic trading, based on mouse clicks, may have eliminated much of the grey area in terms of how a trade is
executed, but many buy-side traders still prefer to trade over the telephone. This is because of a belief that, on the
telephone, the urgency of the salesperson’s voice can be perceived and that it is possible to surmise the state of the
market from the background noise: is there panic or boredom, anxiety or complacency? For those who choose to
trade on the telephone, typical human dealing situations arise all the time, such as changing your mind and backing
off a trade. After a rate is agreed, there can always arise a reason for either of the counterparties to want to back off
from the trade, the simplest reason being that the client made a mistake and traded the wrong way around (bought
instead of sold). It seems that the market inevitably moves against all mistakes.
Meanwhile, the bank is trying to scramble out of the trade while minimising the cost to the client. Remember that
a dispute about a trade is unlikely to result in a lawsuit, and if the trade is large enough the client will expect the
bank to wear at least part of the loss and put it down to the cost of the relationship. Discussions ensue, telephone
calls fly, harsh language might appear here and there. At times, the disagreement might be as bad as to sever the
trading relationship. When a new salesperson starts at a desk, they are often told that we do not do business with this
or that fund as years before one of these disagreements happened. With enough time and turnover in personnel,
eventually the relationship is repaired.
In the early days of this century, FX was one of the first markets to embrace electronic trading and allow clients to
trade via mouse clicks, either on Internet portals or via proprietary applications. Traditionally, buy-side clients,
especially hedge fund traders, are very aware of the leverage they have on the banks, because they can always easily
take their business elsewhere, and they know very well that “the customer is always right”. On occasions, some
aggressive fund managers are believed to have abused that leverage. Some customers will routinely call back at a
later time and claim, for example, that a certain trade they had executed earlier was a mistake on the part of the
bank. They will claim that the direction of the trade was wrong, that the salesperson did not understand correctly and
so on, with the aim of cancelling a losing trade. Such activity always puts the bank in a difficult position.
For the most part, customers did not anticipate the extent to which trading in general would change with the
introduction of electronic trading. In the situations described above, after executing some of those losing trades via
an Internet portal, a small number of clients would still blindly continue to demand the cancellation of a trade on the
grounds that the bank person did not understand the direction. Of course, electronic trading means they had not
spoken to anybody – they had just clicked their mouse to execute the trade, and therefore there was nobody at the
other end of the telephone line they could blame.
Another client of a bank’s trading area is the bank’s own treasury department. This is analogous to a corporate
treasury, simply funding the bank’s day-to-day corporate operations and, as such, is unrelated to the financial
markets trading floor. Among other functions, this area will hedge the bank’s profits from overseas activities,
typically with FX forwards. As with any other corporate treasury, this area’s job is to shop around and get the best
price. Interestingly, most of the treasury areas dealt their FX business with other banks because they did not trust
their own trading floors to give them the best price.
Specifically, I recall one US bank’s own treasury refusing to conduct FX options business with its own trading
desk. The story goes that a few years before the treasury department had been hedging all its risk exclusively with
the internal (trading floor) desk, only to discover it was being ripped off. Evidently, one day it checked prices with
another bank’s trading desk and discovered it was being charged a mark-up. So it decided to take its business
elsewhere going forward. As a client of the other bank, it was then treated royally, being entertained by the external
bank’s sales desk. Obviously no one wins in this story, and it speaks of the misaligned incentives between individual
persons and desks on the trading floor versus those of the broader institution (incentives that we will discuss again
later). From a high-minded and smart perspective, the greater interest of the bank is to keep all parties from charging
any mark-up, and the trading floor should give the treasury the absolute best market price on all transactions.

SHADES OF GREY
Now that you have a taste for how front-running can be an issue, and we explored the most relevant issues that drive
behaviour on this topic, let’s examine some real-world examples. As we have mentioned, it is, and has long been,
common practice for an FX trader to pre-hedge a large client trade. Today, the practice is subject to certain
disclosures and regulatory limitations, and even the Global Code recognises that pre-hedging is considered normal
behaviour, and proper management of large client orders is considered a key skill that differentiates good traders
from others. Prior to the global financial crisis, traders sometimes made more money through pre-hedging than if
they had to show a price first and then covered the position.
Before denouncing this as a questionable practice, consider the last iPhone you bought. An iPhone might cost
Apple only US$200 to make, but some models sell for about US$1,000. Is this immoral? The answer is not easy.
Apple provides a service so that people can use cellular phones, Apple has invested untold amounts in the
development of the product, and at the same time it enjoys a monopoly created around the iPhone’s ease of use and
aesthetics for which some are prepared to pay a premium worth US$800. This approach protects their margin.
However, this gets more difficult to judge when one considers the behaviour of pharmaceutical companies that
exploit their patents to charge mark-ups that are a multiple of the cost of the drugs they make. For instance,
controversially, “Mylan [has] raised prices on its life-saving Epipen injector by over 400 percent over seven years”
(Smith, 2017). There are arguments on either side that we will not explore here other than to simply point out that,
before 2008, this way of doing business in FX was considered an integral part of how traders on the sell side made a
buck.
What happened in 2008? As is well known, the subprime real estate crisis precipitated the financial crisis, but it
also adversely touched millions of Americans who had nothing to do with finance and who could scarcely afford to
lose their house. This created a renewed scrutiny on that part of financial industry that had expanded from a very
specific niche of the market – subprime real estate – to encompass every area, including FX. As a consequence of
renewed public focus and popular movements such as Occupy Wall Street, the attention of the masses turned to the
bankers who enriched themselves and to the way they did it. After 2008, many practices that were considered part of
the normal functioning of the markets started to be considered illegal and were prosecuted by the authorities,
especially in the US and the UK. In numerous cases, banks agreed to settle matters with authorities, and ended up
paying huge fines (see Chapter 2).

What are you so upset about?

Over time, the public started to question the practice of imposing fines without the pursuit of accountability, for two
reasons. The first was that the banks were slow to admit wrongdoing, and therefore their downside was simply
limited to the fines paid with no consequences to their future operations. For the banks, the prospect of not settling
and going to court is not appealing, because if they lose they could face a suspension of their bank licence. Imagine
that, in a small case involving just a handful of individuals, a large multinational bank could lose its licence in a
large US state, such as Florida. That would mean a suspension of its retail bank operations, brokerage, underwriting
and trading operations in that state. Consider the case of HSBC:
The 2012 settlement with HSBC detailed how the bank violated US sanctions by conducting business for customers in Iran, Libya, Sudan,
Burma and Cuba. HSBC accounts were also used by the Sinaloa drug cartel in Mexico and Norte del Valle cartel in Colombia to launder
US$881m.5

However, the Obama administration on its own accord and with pressure from UK authorities did not prosecute,
because “if HSBC had been proven guilty of criminal action, it could have lost its banking charter in the US”. Can
you imagine how many consecutive life sentences an American citizen would have got if he were caught facilitating
all these criminal acts? Yet the bank walked away with a fine. Now it becomes clearer why people are angry.
The second reason why settlements were criticised was that they almost never involved individuals. For the
managing directors who had collected remuneration overseeing practices that were now under investigation, there
was little accountability. The perception was that they did not pay anything out their own pockets, and that their
careers continued largely undisturbed. The public demanded accountability.
After the case of the London Whale (Hurtado, 2016), clawback provisions started to be applied to compensation
language. These usually stipulate that a bonus paid to an employee could be subsequently cancelled and the
employee forced to pay it back should it be discovered, even years later, that the performance that the bonus was
supposed to reward came from illegal or forbidden business practices. One problem with such clauses is that the
money could have been spent in the meantime, hence they tend to be more realistic for unvested stock or
delayed/deferred bonuses. Even in such cases, however, the instances where they have been invoked have been
limited, so they may have been less effective than intended.
Investigators on both sides of the Atlantic have been much more aggressive in taking action against individuals,
and even forcing banks to admit guilt. Previous settlements have often been accompanied by no admission of
wrongdoing. The landscape for the prosecution or punishment of individuals is very complicated. In instances where
the case is heard before a jury, accused individuals have at times taken a risk by refusing to settle and going to court.
Part of this strategy may be to convince juries that the whole system was designed around the behaviour they stand
accused of. Accused individuals have claimed, at times successfully but also at times with catastrophic
consequences, that they had been singled out – and that in reality they were implicitly or even explicitly required by
their superiors to generate revenues based on those types of behaviours. On one side, the fact that it may have been
common practice does not stand much ground in a court of law. That approach was a gamble to raise the empathy of
the jurors, whose sympathies may be limited when their average compensation may be far less than that of the
accused, limiting the strategy’s success. In many instances, the accused were just members of teams that engaged in
the same behaviour. Were they singled out by the prosecutors and made an example of? Potentially, but does that
excuse them if they did indeed break the law?
Another legal argument that has been raised by the defence in some cases is that clients were aware of all the
mechanics related to the trading and execution of orders, and that all those details are part and parcel of the efficient
functioning of the market. At times, it has also been claimed that participants in a specific market should have
known that the mark-up was not commonly disclosed when negotiating prices, as is still the case in many other
industries. The reader will have to make up their own mind on this issue, but it can be seen as analogous to putting
one’s pet in the washing machine and then suing because the animal died. The washing machine manufacturer could
potentially argue that you should just have known that pets don’t go in the washing machine.
One case in Hartford, Connecticut, concerning bond traders is described by Dolmetsch (2017):
The cases have been a wake-up call for traders who long believed they could do most anything to sell a bond, said a chief compliance officer
at a major bank who wasn’t authorised to speak publicly about the case. Traders have become well aware of the potential for embarrassment,
professional fallout or worse, the person said. Revelations of conversations, recorded or captured in chat transcripts, reflect what was thought
to be everyday business on Wall Street. Buying, selling, bluffing, puffing, haggling over price – that’s how the market works. But in recent
years, prosecutors began to describe common tactics as fraud. Now, doing your job – fighting over every 32nd of a point in price – could
land you in prison.

To understand how sell-side traders make money from client business, let’s consider at a specific example that has
been in the news.6 The head of spot trading at HSBC in London talked to a client about a large order worth US$3.5
billion. The client needed to purchase pounds sterling versus another currency due to a pending merger deal. The
head trader and his team convinced the client to execute the deal at a published fixing rate. The most popular fixing
time would have been 4pm London time. Market consensus is that a lot of trading, and therefore liquidity, is
available at that time; consequently, it is often easier to execute large volumes at that time without moving the
market. In this particular case, because HSBC would be buying a significant amount of pounds on behalf of the
client, the market could potentially go up before the trader finished buying. This means subsequent purchases of
pounds would have to be bought at a higher price than the fixing. If this were a smaller deal, the trader would
normally be able to execute in a way that would allow them to buy without moving the price higher. But in this case
the trader moved the price substantially higher before the fixing, so they bought low and sold to their client high.
When all was set and done, the head traders were indicted on an 11-count indictment charging them with conspiracy
and wire fraud. Prosecutors claimed that they engaged in a scheme that saw them use information provided by the
client (the pending deal) to make a US$3 million profit for the bank, in addition to the US$5 million in fees on the
transaction. The US assistant attorney general claimed: “The defendants allegedly betrayed their client’s confidence,
and corruptly manipulated the foreign exchange market to benefit themselves and their bank.”
So what was the understanding between the bank and the client? From the client’s point of view, they have two
goals. First, to buy the pounds, and second, to buy the pounds at the market rate, preferably a published rate, on a
specific date and time. They can then show their stockholders they received the pounds at the true market rate, and
no worse. The 4pm London fix is a great mechanism for this (see Chapter 2 for a more detailed discussion of the
fix). HSBC therefore told the client that they would buy the pounds for them and deliver them at a rate equal to the
fix, possibly plus a pre-negotiated fee that would be quite small, perhaps in the order of 0.02%. As an aside, the
bank actually managed to convince the client that a different fixing time was better than the most popular 4pm time.
This should have raised questions from the client, but the bank successfully argued that anticipating the time would
“surprise” the market. We are not sure about the logic of this “surprise” execution, but potentially the trader might
have thought that at a less liquid time it would have been easier for them to manipulate the market price.
From the trader’s perspective, if the requirement is to buy the pounds, prudent risk management may involve
buying before the fixing period or time. Buying in a very aggressive way early, however, may move the market
higher, potentially higher than the average buying price. This may impact the fair and efficient functioning of the
market, and would therefore run counter to the Global Code. The Code did not exist on paper at the time, and at best
it may have existed in some trader’s understanding of ethical behaviour. Principles 10, 11 and 12 require that such
an order be executed fairly and (to the extent possible) without disrupting market functioning or the price discovery
process.
The head trader at HSBC did not invent any new trading method that other traders had not employed previously to
ensure that business executed around the fixing was profitable. This argument resembles the defence arguments
discussed above – specifically, that this way of dealing was common practice, and that what apparently was
accepted practice before the financial crisis was suddenly branded as illegal and conspiratorial by the authorities.
This raises the issue of retroactivity of the law.
There is, however, a further twist to the HSBC story. The bank/client relationship discussed in previous chapters
may give rise to conflicts of interest. The bank was engaged to advise the client on the best way to execute the
purchase of the pounds. Does the bank then need to work against its best short-term financial interests? From
internal communications and other evidence made public, it appears that the bankers involved in the transaction
worked in a coordinated manner to maximise the bank’s profit.7 Nevertheless, the prosecution argued that the bank
had a duty to act in the best interest of the client. To make matters worse, it has been alleged that bank employees
may have misled the client. For example, when the price of the pound started to rise shortly before the fixing started,
which is clearly a negative event from the client’s perspective, the bank told the client the reason was a foreign
investor (specifically a Russian) had started to buy, pushing the price higher. This was allegedly a misrepresentation
to cover the fact that the bank was the one buying, and documents show the bank in fact had no knowledge of any
Russians buying. So, clearly the bank had decided how to resolve this ethical dilemma of being the fox in the hen
house. Chicken for dinner!
At a high level, fixings can theoretically make FX business easier to transact, but of course the devil is in the
detail. It is clear that individuals have had strong incentives to make money for their employers, and therefore be
awarded higher bonuses. Going forward, several factors, including the adoption of the Global Code and a heavier
compliance presence helping traders to interpret the rules (FX traders with law degrees are few and far between),
should ensure that generation of mark-up is less of a grey area.
The conclusion to our story is that on April 27, 2018, the head trader was sentenced to two years in prison.
Bloomberg News reports that “Mark Johnson bowed his head and his lawyers appeared stunned as U.S. District
Judge Nicholas Garaufis in Brooklyn, New York, rejected his request to surrender at a later date. Johnson then
handed his wallet to the lawyers and removed his tie before being escorted out of the courtroom by a deputy U.S.
marshal. His lawyers asked that he be sent to the low-security federal prison in Allenwood, Pennsylvania.”8

Heads I win, tails you lose

What process does a typical FX transaction follow on the trading floor of a big bank? A salesperson may be
speaking on the telephone with a dozen clients all morning. Let’s follow what happens when, as a result of those
discussions, client ABC decides to purchase €100 million versus US dollars in the spot market. The salesperson has
no idea about the bank’s euro holdings or the spot trader’s risk profile when the phone call comes in. The euro spot
dealer, who does have that information, has been observing the market every second they are sitting at their seat.
They are computing the net position the bank has in euros and whether they want to position the bank long or short
euros. Given the size of the request, the salesperson would stand up and project the query across the floor, “euro in
100”. Notice the salesperson was not told by the client the direction of the trade. The trader gives the salesperson a
two-way (bid and ask) firm price, which is passed onto the client, who then decides whether to trade or pass. Very
often there is some negotiation and haggling. If the client agrees to deal at the price shown, then the trade will be
booked and the bank’s back office will start a long chain of events. The trade will be confirmed with the
counterparty, funds will be allocated and wired, usually two days later. It is easy to see why there would be a need to
have an office with a lot of open space. Traders, sales, middle office, analysts and managers can quickly
communicate on constantly changing market conditions, and also know what their clients are doing so that they can
prioritise their tasks appropriately.
However, modern technology means that salespeople and market makers now do not need to be in the same
physical location. Shouting across the floor has been replaced by open video lines and online chat rooms, making
the traditional trading floor something of an anachronism. Modern trading floors look and sound different from their
predecessors of the 1980s. There remains an element of tradition, but the reality is that today on trading floors much
of the pricing is done by computers. Human traders frequently just maintain the software algorithms, and possibly
take care of the largest transactions or positions. Similarly, salespeople do not need to shout as much because they
just click on their computers to execute a trade.
Let us first examine the situation of market-maker traders who manage their own trading books. They are
assigned risk parameters by management. For spot dealers, this distils down to a given notional amount of specific
currencies they can own against another currency. The net position they are allowed to carry is limited at all times
and more strictly controlled at closing time, before leaving at the end of the day. However, market makers might
need to keep a long or short position overnight. Market conditions may preclude them from completely squaring
their book every day, so a little flexibility is needed. Keep in mind that the trading book might be passed to the next
geographical trading centre at the end of the day, and to another trader. So New York will pass to, say, Tokyo, and
the trading book will be actively managed 24/5. For example, a client might buy Egyptian pounds at 5pm New York
time; it is unlikely this trader will be able to find a source for this currency pair until Europe opens up the next
morning. So, if overnight something was to happen in Egypt that caused the pound to move significantly, this trader
is at risk that they may be unable to hedge.
As we’ve discussed previously, regulations made under the Dodd–Frank Act have decreed a separation in banks
between the market making and the speculative or proprietary trading functions. Before Dodd–Frank, a market-
maker trader might see most of the bank’s clients buying euros in the morning, and it would be an easy strategy to
go long for their own speculative trading position. This came to an end after the global financial crisis – proprietary
positions at banks are now a thing of the past. Today, risk and position limits are much tighter, and at times a market
maker will be required by compliance to justify formally why they could not square a particular position (eliminate
risk) or why they took a new position at a time that their trading book was flat (had no pre-existing positions).
Compliance could even ask why the trader hedged a position in a particular way. Why did they use a particular
instrument instead of another? For example, it is quite common to hedge USDBRL (US dollar versus Brazilian real)
positions using USDMXN (dollar versus Mexican peso) as the latter is more liquid. Today such hedging might
generate a call from compliance and the trader might have some explaining to do.
Prior to the global financial crisis, heads of FX desks at numerous banks were at pains to emphasise the
proprietary trading capabilities of their businesses in convincing buy-side clients (corporations, hedge funds, asset
managers) that it would be in their best interests to consider the bank as a partner, not simply as a liquidity provider.
The intention was to trade together with their clients. The bank might share its market information, gleaned from its
role as market maker, with the client – and might go so far as to take the same speculative position as the client.
However, in the wake of the Dodd–Frank Act a new business model emerged that saw a renewal in senior ranks.
Some senior-level departures were hastened by the hurdles of making money when one could no longer speculate on
client positions, but also by some high-profile investigations on benchmark rate rigging.
This can serve to highlight another characteristic of the individuals that took large proprietary FX positions at the
banks, namely that they had the “nerve”, or risk appetite, to be able to run such large, risky FX positions. A major
reason for this was that they were in possession of “flow” information – that is, traders at large institutions were
often able to deduce which way the market as a whole was positioned, based on the client volumes flowing through
the trading book. They stood to make profits for the bank (and, by extension, potentially an improved bonus for
themselves) if they were consistently correct. If they were consistently wrong, however, they may get no bonus and
in the worst-case scenario may lose their job. In some circumstances, this latter outcome became a badge of honour:
some “let-go” traders were taken on elsewhere, hired simply because they had the risk appetite to take large
speculative positions. This cultural peculiarity resembles what happens in the world of Silicon Valley Internet
startups, where having worked at many failed efforts also constitutes a badge of honor. In both jobs just one home-
run performance will pay for years of long shot chances.
Importantly, speculative trading and market making are two distinct skills, although at times the lines between the
two became blurred. It was not unusual for a bank to promote a successful market maker out of that activity and into
speculative trading because they had been a profitable market maker. However, it was not uncommon that, within a
matter of months, they would be losing money and struggling to keep their jobs because they were no longer trading
around customer flow.
The aura around strategic traders carried over to the buy side, particularly to hedge funds. Buy-side firms continue
to recruit frequently from the sell side. Some successful hedge funds, however, are not looking solely for trading
talent alone when hiring from the sell side, but also for experience of the techniques that the sell side employed to
make money out of the buy side. Hence, joining a bank spot trading desk and then moving to trade at a hedge fund
was an ideal career progression. As with many things, however, chance played a large role. For example, one day
the global head of an FX business at a large US-based FX bank was visiting the London trading floor and chatting
with the spot traders. He saw the head trader looking at a 10-inch stack of résumés from candidates looking to be
hired by his desk. He split the stack in half and threw one half in the trash. The global head asked the head spot
trader if he was mad – to which, with perfect English aplomb, the head spot trader replied: “I always do that; I don’t
want to hire somebody who is not lucky.”
Market-making activity is compensated for by the amount of money that accumulates in the trading book.
Theoretically, the trader is supposed to earn the spread between the bid and the offer. When they are allowed to
speculate, additional profits can be made. For such opportunities to be profitable, however, not just in FX but all
markets, the markets must move; hence, volatility is critical to market making.
HOW THE SELL SIDE IS COMPENSATED
No discussion of how the sell side is compensated would be complete without looking at bonuses. For front-office
employees working on a trading floor, total annual compensation may be composed of both a salary and a bonus. A
bonus is frequently composed of a cash portion, which is paid typically early in the new year, and various deferred
portions such as cash or unvested shares of the employing bank. The idea of the bonus is to reflect the performance
of the individual. They are frequently described as “discretionary”, with their value being decided by management.
In the best of situations, a bank may determine an individual’s bonus using a formula based on mark-up that was
extracted from client deals. See the prosecutor’s comment in the indictment later in the appendix: “Over 80% of
defendant’s 2011 compensation resulted from bonuses for performance.” A bonus is not to be confused with
commission, which is usually paid to retail brokers, and is almost always calculated using a set formula.
In essence, the bonus expresses how much management wants to retain an employee for one more year relative to
their peers (at the same bank or other banks). If a trader’s bonus is relatively too low, they might decide to quit and
go to work for another bank. Before 2008, this was reasonably easy to do as there was a lot of demand in the labour
market. One might see colleagues quitting in March, ready to join a competitor which guaranteed them a bonus
increase for the following year, before they turned around and came back to their desks the next day because their
original bank had made them a counter-offer of an even bigger bonus guarantee. The existence of the bonus creates
a strategic game between management and front-office talent. If a trader stayed with a bank for a long time, they
would have been unlikely to have been paid as well as someone new who had been enticed to join the bank with an
attractive bonus.
Traditionally, there have been two moves per year in this process. The first is the bank awarding a bonus in, for
example, February. If the bank’s management want an employee to stay, they will pay that employee such that they
are reasonably happy and less motivated to speak with a headhunter looking to move the trader to another bank.
Switching implies risks such as that one’s clients will not follow (if the employee is a salesperson) or that the new
bank will not have as much customer flow through the trading book as at one’s current firm. On a more sensitive
note, moving jobs might also be risky because you could never be sure that in your new bank you would be allowed
to employ whatever practices you had used to make money at your old bank. There is also the risk that internal
politics pre-dating the trader’s arrival will make it impossible for them to develop the business. There are many
anecdotes concerning managers trying to fight a turf war by hiring entire teams from competitors. When those teams
arrive for their first day on the new job, they might find out that a rival manager already has a team in place that
does exactly the same function.
The second move in the process usually takes place between March and June, and this is the employee’s move.
The employee effectively has to decide each year whether to stay or to switch employers. Given the risks in
switching, normally a reasonable raise in pay would be expected. In addition, it was customary for a new employer
to guarantee the bonus for at least the first year. For example, in 2005 the market for FX traders was tight to the
extent that it became almost customary to guarantee the bonus for the first two years. Particularly aggressive French
and UK banks were seen bidding-up the bonus guarantee, but many people then experienced a dramatic drop in the
third and subsequent years that effectively left them worse off than if they had stayed at their old bank. Another
typical reason to switch jobs was that one felt there was nowhere to go in their current firm. Usually this was the
result of internal politics, or sometimes business drying up.
An example of this was the decline and ultimate demise of derivative structuring desks, a process that started in
2011 and ran for about four years. As mentioned, structuring desks were staffed by very quantitative types, usually
with a curious and intellectual bent. At the time, structurers were the most analytical and most specialised group on
the trading floor. However, in times of crisis and financial contraction, intelligence is less valuable than
opportunism, street smart, connections and occasionally simple treachery.
For example, on one bank’s hedge fund sales desk, just as one colleague, let’s call him John, left his desk to go to
the bathroom, his big client called. The salesman sitting next to him, let’s call him Steve, picked up the line:
Hey, this is Steve, how can I help? No, I think John is out today. But here’s what happening in the market, and sure, let me work your order.
By the way (Mr Client), in the future just call and ask for me instead of John.

A long list of nasty examples could follow. Like the one of the sales manager who stayed late every night to replay
the tapes of his reports’ phone conversations and try to undermine them all.
In any case, clients started to retreat from complex deals due to opaque pricing, high credit charges and lack of
liquidity, and the structuring game has never fully recovered. Structurers themselves could likely forecast their fate
before management, as many firms ultimately closed their structuring desks.
During and after the crisis, criticism of bonuses was all the rage in the news, mostly because total compensation
for some traders was very high relative to the general public, and in the context of the financial devastation that hit
the high street, thanks to Wall Street. However, there was a fundamental misunderstanding by the public at large,
which believed that to reduce total compensation the bonus had to go. In reality, this is poor algebra. The split
between salary and bonus has little to do with the fact that total compensation was perhaps too high. Both could just
be cut in half and total compensation would therefore be halved; there was no need to eliminate the bonus or to treat
it as a scandal.
After 2008, many banks reduced the bonus and increased the salary, following pressure from the public, the
regulators and the press. It may be unclear what benefit this caving in to the witch hunt has created. In the UK, a law
was proposed to eliminate bonuses entirely, but of course banks could very easily have granted the same amount of
total compensation, just composed it entirely of salary. Meanwhile, in the US a law was proposed that would have
levied a tax rate of 90% on bonuses; this also missed the point. In the good years, banks were prepared to pay their
best-producing employees US$5 million if those employees were able to produce US$100 million dollars in trading
revenues for the bank; that is pure commercial logic. Reducing the bonus by law would have been completely
ineffectual: the main issue is that total compensation was high. The problem in increasing fixed pay to compensate
for cutting bonuses is that it is more difficult to change base compensation based on annual budgets and results. If I
pay you £30k in salary and £70k in bonus but the public opinion is pushing me to eliminate the bonus, then I can
just pay you £100k in salary. Now the public is happy, and you are happy because the bonus was discretionary, so I
could have reduced it if you misbehaved. Now you have no incentive to behave.
The scandals that ensued were not a consequence of a change in behaviour on the part of the employees, but the
realisation that all these revenues they produced for the banks were at the expense of the bank’s clients. In a
booming economy this is not a big deal, but after the financial crisis the public had no tolerance for economic
suffering caused by “a few crazy traders on Wall Street or in the City”. In the 1970s and 1980s, a job on a trading
floor was really a very niche profession that only few considered as a career. After the level of compensation rose
steadily for a decade, things changed and the brightest graduates from all the top universities started to plan careers
on trading floors. As a society, we are not prepared to see our best and brightest in trading rooms (we still prefer
lawyers and doctors), and a select few were caught up in scandals when they could have been using their
engineering degrees for something else. However, in open market capitalist economies, it is very hard to cap
compensation in any one profession. All the pressure governments put on banks just caused the best candidates to
move to the buy side, where they can do the same job but this time they are not at a bank. In other words, abuses
should be eliminated by more enforcement and scrutiny from regulatory authorities. Capping compensation will just
cause people wired in this way to move to another entity, another country, or possibly a related business that offers
comparable earning potential.

Why was the money so good?

Finally, why is it that compensation levels in some roles in financial and capital markets were so high? Why don’t
we have star farmhands that make similar compensation? Why aren’t electrical engineers working for a company in
Middle America being awarded comparable bonuses? Quite simply, such individuals could not generate anywhere
near the same revenues in those fields, and so would not be able to demand the same levels of compensation.
Similarly, they found the mathematics and financial modelling required in banking to be child’s play after finishing
engineering school. Also, the bulk of the millions of bank employees are not compensated anywhere near the levels
of the relative few on a trading floor. For example, according to Glassdoor9 the average back-office salary in a bank
(the people who process all these trades) in New York City is less than US$36,000 per annum. So, why are there
also people making significantly more? Here are some observations.
It seems that there are several factors at play. First, that’s where the money is. Not in the Willie Sutton context,
but because banks themselves frequently generate large profits. For example, in the 2000s US financial industry
profits reached as high as 40% of US business profit (Smith, 2016). Second, trading businesses had an outsized
impact on the total profitability: fixed income, currencies and commodities (FICC) profits are typically key numbers
that can be material towards a bank meeting its analysts’ estimates. Therefore, a small group of people on a trading
floor, in an institution of hundreds of thousands of employees, can impact on everyone’s life by their success or
failure. It’s not surprising that they take on significant importance. Also note that it would not have been possible to
distribute this oversized contribution over a larger but more moderately compensated number of employees. This is
because there can only be one market-maker EURUSD trader, and only one salesperson that the big hedge fund will
call every day to discuss EURUSD.
This importance is amplified by the rare ability to assess the value of each trade, and therefore an individual
trader’s contribution, more precisely than most other jobs. That is, a bank can determine the value of each one of its
trading employees pretty much down to the penny. Imagine a newspaper that can almost exactly measure the value
of each of its reporters in dollars and cents, or a theatre company each of its actors, or a government office each of
its clerks, or a car company each of its assembly line workers. Of course, there are other jobs where performance
outcomes can be measured precisely – for example, the performance of a marathoner, or how many rivets an
assembly line worker can tighten in a minute, or the number of words per minute for a typist. However, banks can
do it for a larger swathe of employees, and in real time. This ability, and the commensurate compensation, has been
crucial over the years to the incredible success of banks to attract talent away from other industries. Their ability to
pay well and the perceived fairness of the system attracted the very same talented individuals that were attracted by
professional sports – people who do not want to all be paid the same, people who like to compete and who like to
have their results visible on a scoreboard.
Third, there was a sense of meritocracy – a feeling that everyone was on a more-or-less equal footing. If you
made money, it did not matter whether you kissed up to the boss or not, you got paid. This is something which many
smart and talented young graduates found attractive. However, as in many areas of life, the total truth is not that
crisp. This sense of meritocracy was a great draw to a career in financial markets, but the difficulty for many was to
get through the door. For the most part, especially in FX and especially in the 1980s and 1990s, the job was akin to a
boys’ club of white male jocks, where testosterone levels were compared and contrasted. Years later, at a diversity
internal meeting held at a major bank, which we shall call Money Loans, one of the traders complained about the
existence of the “Women in Finance” club. He questioned why there was no “Men in Finance” club. One female
senior manager shut his mouth by retorting: “There is indeed: it is called Money Loans!”
It was still easier for relative outsiders to get jobs on trading floors than prestigious jobs on the investment
banking side of the firm. That is, most old-school spot and forward traders had little interest in the new-fangled,
mathematics-infested, derivative products that needed computers and risk systems to calculate the right price. The
sophistication that followed FX structuring opened the doors to a new brand of employee who were as comfortable
discussing statistical concepts as others were talking about the Yankees or Chelsea.
The fourth reason for high compensation has to do with the personal risk/reward calculation. Many a trader or
salesperson in FX was called into a conference room, alone or collectively, and informed that their services were no
longer needed (for an extremely accurate enactment of this procedure, watch the first scene of J. C. Chandor’s
Margin Call, 2011). Typically, there was a security guard nearby and the now-former employees were asked to
“please go straight to the elevators, do not go back to your desk”. These are not people that have done something
wrong or illegal: “It’s just that we’ve changed our strategy, thank you, you were great, good luck.” When job
security is absent from one’s life, one behaves in a certain way. Like a squirrel, one collects as many nuts as
possible, as quickly as possible. One former strategist, working at a large New York bank at the time, is now a
portfolio manager and has pointed out that it has always been the case in history that those who take the biggest risks
can reap the largest rewards. In previous ages, it might have been merchants, or perhaps warlords, but today it is
financiers. That strikes many people as fair. The thing that sticks in the public’s craw is that, beyond a trader’s
personal risk, the financial risk taken is off society’s dime. That is, as discussed above, being able to privatise gains
and socialise losses.

Luck and the pecking order

We’ve previously stated that, particularly in spot FX, the most successful dealers did not come from Ivy League
schools or Oxbridge, nor did they spend a lifetime learning higher-order mathematics. They typically were scrappy,
street smart, fast-thinking young people that happened to catch a break. Consider, for example, my dear friend Phil’s
story. Phil was a short-order cook for his family’s diner in the Wall Street area of Manhattan. Working in
restaurants, and in the kitchen in particular, especially of a family business, requires particular mental and physical
skills. A short-order cook deals with many people screaming 10 things, each simultaneously. The food must be
prepared in the appropriate order so it comes out together. He must answer questions relating to pricing and special
orders, and be clear, concise and in command of the situation. When the head FX trader of a major shop walked into
the diner for breakfast one day, his jaw dropped watching this young guy control the entire place while his hands
and mind were moving at lightning speed. When things quieted down the trader called Phil over and asked him if he
wanted to be an FX spot dealer. Phil probably had no idea what a trading floor was, let alone what a spot dealer did.
The conclusion to this story is that Phil became a legendary FX trader on the Street, before starting one of the most
successful proprietary FX trading firms.
It was tough for non-cookie-cutter hopefuls to get a chance on the trading floor – you needed that break. I got my
chance to trade options on a trading floor because I was able to program the Black–Scholes option pricing formula
in this fancy new tool called a Lotus 123 spreadsheet. At the time, I did not know that there was a trading floor in
the building, nor what they did there. Further, while it was admittedly tough for men, it was doubly difficult for
women of any background, or for minorities.
 Beyond the luck of finding one’s way to the trading floor, the mandate a trader received greatly impacted their
profitability and earning power. First, some assets are easier to trade than others. The trader who gets to be the
EURUSD (euro versus US dollar) market maker will always generate a larger P&L than the NZDUSD trader, unless
you are at a New Zealand bank. Management decides the trading book assignments. Profitability becomes easier or
harder depending on the instrument, the size of one’s bank, the type of client flow and one’s specific function.
Structurers, for example, were tasked with slicing and dicing complex financial risk to create innovative products for
clients. In the general industry decline that followed the financial crisis, consider a situation where management
arbitrarily tripled structuring’s annual revenue production goals, and expected profitability to significantly increase
from specific clients. At year-end, one may find that despite having done a great job and earning a lot of money for
the bank, one’s overall compensation was less than expected because the definition of what constitutes success (in
this case, unreachable goals) was set by someone else.
The ability to measure the performance of a single employee could be a double-edged sword. On one side, it
injects an aura of set rules, meritocracy and fairness, but on the other it has a deep impact on the way individuals
make decisions. In solitary pursuits such as individual sports, it makes a lot of sense to measure the performance of
the individual. Although individual performance is lauded and romanticised, the truth is that success is only
achieved through teamwork. The most contentious relationship that pulls at this is the conflict of interest between
salespeople and traders.
Consider the following simple example. A salesperson talks to a client and then asks a trader for the ask price on
€10 million versus US dollars. Suppose that the mid-market price is 1.0500 and the trader quotes 1.0495/1.0505,
1.0505 being the bank’s offer price. The salesperson wants to add a degree of mark-up. Sales quote the client a
1.0515 offer, and close the trade. Further, suppose that we are in ideal market conditions and the trader is able to
cover the position at mid-market, buying the €10 million from another bank at 1.0500. So the bank made (1.0515 –
1.0500) x US$10 million = US$15,000. How should management evaluate the performance of both the trader and
the salesperson? One idea could be to attribute the spread (0.0005) to the trader and the mark-up (0.0010) to sales.
Such a simple system has been tried and found wanting in many respects. Spoiler alert: nobody has found the
perfect system. Here are some of the problems that can arise. If the client is a very competitive hedge fund that
makes five banks compete for every deal, then any amount of mark-up is likely to be too much, and the trade would
almost certainly have gone to the competition unless there was zero markup. Sales should have only quoted 1.0505,
resulting in 0.0005 for the trader and zero for the sales. If you are a salesperson and you only cover hedge funds,
then you will make zero. A possible solution that has been tried involves giving sales a percentage of the spread (eg,
25%) – so in this case the salesperson will be credited with 0.25% x 0.0005 = 0.000125 and the trader will be
credited with the rest (ie, 0.000375).
Another problem arises when the opposite happens. For a different client type, the salesperson might apply a
larger mark-up. In such a case, the disparity of compensation may have motivated the trader to enlarge their spread
to ensure equal trader/sales compensation. However, if the salesperson gets it wrong, the price shown to the client
will be too wide and the client will trade with another bank. The business will be missed and everyone makes zero.
The trader is incentivised to second-guess the salesperson’s judgement. This is not very efficient, as the salesperson,
not the trader, is in the best position to determine the appropriate mark-up for that particular client based on a
number of factors (including credit) that the trader may not necessarily be party to.
In other instances, banks have employed a system called “sales credits”, sometimes referred to as “funny money”.
In other words, in our previous example both sales and trading would be attributed the entire 0.0015 and the conflict
of interest would be eliminated. At the end of the year, the sales credits will be attributed a value. This is not a very
good solution, and it is easy to see why. If this was the only trade done all year, the bank would have US$15,000 in
profit in the trading book and would allocate the bonus arising from that to be split between the salesperson and the
trader. The problem is that the trader and the salesperson will then have to split the bonus money anyway, and we
are back to square one. So, the introduction of sales credits just delayed the problems until the moment of truth at the
end of the year when the pot has to be split. In essence, there is no perfect way to split the revenues. This is a
consequence of the fact that despite the individualistic nature of a trading floor job, there is always a component of
teamwork.
Here’s a less obvious example of how a lack of teamwork can play out. At one of the banks I worked at, a risk
advisory director had a productive, ongoing dialogue with a large corporate client, a large industrial company, a
former Dow Jones Industrials member. One might ask, what does a risk advisory director do? They are experts in
measuring the financial risk that corporations face, through methods such as scenario analysis and value-at-risk
(VaR) calculations (these will be discussed further in Chapter 11 and then in Part III). Before “unbundling” rules set
out in the Dodd–Frank Act and MiFID,10 corporations often relied on banks for this work, with the promise of larger
business volumes in the future. If the volumes from the client did in fact increase, the risk advisory director got their
contribution recognised – usually via an amount of “funny money” equal to the revenues on the increased business
volume.
On one occasion, our director was asked by a client about risk calculations using “fat-tailed” probability
distributions. This is a very typical type of calculation if you are a structurer or an option trader. The fat-tail
distribution evaluates the risk on each and every currency according to option prices, thus reflecting the fact that
remote risk events tend to happen more frequently than classic bell-shaped probability distributions imply.
Our director did not know any of the above, but very shrewdly refrained from asking any of his colleagues in the
fear that the distribution of “funny money” in case of increased volume would have to be split with them. He
proceeded to Wikipedia, where he came across the notion of t-student distribution, which is a type of distribution
that does have fat tails. He then used such a distribution for his calculation, and happily went back to the client
stating that he had incorporated fat tails into his assessment.
Little did he know that fat-tail distributions only make sense if specific to the currency of the client’s exposure,
and not via a generic t-student distribution fished out of Wikipedia. In that sense, the fat tails you can see in the
expected probability distribution for USDBRL are completely different from those you can see for EURUSD, and
all those differences can be observed in the prices of vanilla options. So much so that option traders maintain very
sophisticated quantitative systems, very often made in-house by the in-house programmers and quants. Our director
did not know that structurers or traders could have given him all the data and all the distributions at the click of a
button. They would have been correctly tailored to the actual exposures the client was facing, not to a generic risk as
explained in Wikipedia. There are typically a lot of smart people on trading floors. The most successful of these
leverage their colleagues to the fullest.

Salesperson’s mark-up considerations after the Code

Earlier in this chapter, we discussed how a market maker and/or a proprietary trader thinks about turning a profit, on
one hand, and keeping their job on the other. Here, we will discuss how the salesperson, who is between the trader
and the client, is able to earn a living.
Principle 14 of the Code states that the “mark-up applied to client transactions … should be fair and reasonable”.
It will be noted that the Code does not say there should not be any mark-up. So, even in the new regulatory world, it
is expected that sales are allowed to charge “spread”. The lingo is your enemy here. Usually “spread” refers to the
smallest possible difference between ask and bid prices. Usually this is referring to a price made to the best clients.
Here, “spread” is meant as “markup”, which is the difference between the ask price made to the client and the ask
price made by the trader. Ideally, the trader will make the lowest ask possible that does not make them lose money.
On the other hand, the salesperson might use the largest possible markup that does not make them lose the trade, or
worse lose the client. Mark-up is allowed to reflect a variety of factors, which can include “risks taken, costs
incurred and services rendered”. However, a primary consideration for FX is now transparency. While all clients
knew previously that mark-up was taken (banks generally being corporate entities that exist to make a profit), what
has changed is that the Code promotes this transparency by encouraging market participants to publish “a set of
disclosures regarding their FX business” that addresses mark-up specifically.
Explicitly, firms must, via documentation and disclosures, make it clear that the “final transaction price may be
inclusive of markup” and “help clients understand the determination of mark-up”, which may include a breakdown
of the factors that contribute to the mark-up.
The Code does not, however, place all of the disclosure burden on the salespeople, but also sets out firm-wide
guidelines that should be established so as to assist the salespeople in determining that prices charged to clients are
indeed “fair and reasonable considering applicable market conditions and internal risk management practices and
policies”.11 The Code is also very specific around “not misrepresenting any aspect of the mark-up to the client”.
This will come as a welcome relief to the vast majority of salespeople who are law-abiding, and who always knew
that this was a long-term game. As we have seen previously, it is much easier for prosecutor to get a conviction if
they can prove that the salesperson or the trader lied to the client. Being evasive when talking about markup is a
different legal ballgame.
The rules, though ostensibly voluntary, are clearer and management should now be on board with a sustainable
mark-up policy. Some banks have even gone as far as to take the lead and disclose to the client exactly the mark-up
that is being taken on each trade. There are many in the market who would not be unhappy to see OTC markets go
to a more exchange-traded model. That would involve agreeing, at the outset, a brokerage scale, and awarding more
business to those who do a good job executing the trades. The attraction of this model would be that it allows the
sales force to concentrate more on proactively servicing clients and selling the bank’s products rather than walking
the fine line between marking-up too little or too much, as they would have a clear charging formula that already
accounts for the myriad factors which management may impose regarding return on capital employed, minimum
credit spreads, appropriate liquidity charges, cost of overheads and staff, and so on.

CONCLUSION
In reading the previous chapter, one might have concluded that FX is a relatively straightforward instrument, one
that uses simple mathematics to arrive at an orderly way to expedite client orders via bank trading floors, which then
distribute the risk to those who are willing to carry it. That is the direction of travel, as more and more work is being
performed by computers. However, as we saw in this chapter, the truth has been rather different and frequently
outcomes have not occurred in an orderly manner. Humans were, and are, involved – so human relations and human
interactions complicate the process. It’s also what makes FX one of the most fun markets to work in, and the people
some of the most colourful. As a trader, making a price requires tremendous concentration and awareness of one’s
immediate environment, including the market, one’s clients and one’s colleagues. In the final analysis, the price a
trader makes invites questions around the application of mark-up and generation of trading revenue: questions that
are considered at the time of price making and execution, and also at year-end when the bonus pool is being
allocated. How these pressures are handled can reveal a lot about a person.
Most people, at some point in their lives, will have the opportunity to exercise authority over others. On the
trading floor it happens everyday, in very apparent ways. Clients can be contemptible to the salesperson, the trader
can be contemptible to the broker or the salesperson, the front office can demean the back office, the manager can be
contemptible when allocating credit and bonuses. The competitive nature and zero-sum mentality that was pervasive
on a trading floor fed and rewarded this attitude and surely was the cause of many of these cases now being
adjudicated. However, we also remember fondly many of the situations and the character of colleagues when they
had the opportunity to be contemptible with no consequences, but were instead generous of spirit and supportive.

APPENDIX: THE INDICTMENT

Within Chapter 5 we’ve given you a high level description of the HSBC case and raised issues that should concern
us all. Following are excerpts from an indictment filed in January, 2018. We want to include this indictment as is,
with no commentary, because the behaviour and issues come to life in a very palpable way. The fact that it’s about
FX options could not be more relevant. The trading behaviour has elements of what we have already discussed but
also elements of what we will discuss in the forthcoming chapters, like volatility and delta hedging. The language
used by the defendant and the strategy used to extract extra profits are excellent examples of the worst kinds of
activities that used to occur. The transcription of the phone calls to the client and the subsequent conversations on
Bloomberg IB are fascinating. Additionally the cold, calculating statements of the prosecutor are reflective of what
outsiders to this world see and feel. Finally this is a great example of how volatility effects options and how option-
delta hedging impacts spot and forward markets.
Although this is part of the public record, we’ve chosen to redact the name of the defendant, the company and the
bank. We do not wish to inflict any further pain or embarrassment on the parties concerned, we just want the reader
to get a front seat at what used to be a typical FX trading floor at a bank.

BANK, Inc.
2. Beginning in 2008, the defendant DEFENDANT, a/k/a. DEFENDANT, Jr., a citizen of the United States and
resident of New York, New York, was a foreign exchange (“FX”) trader at the New York branch of BANK.
In 2011, DEFENDANT was the head of BANK’s FX trading in New York, and supervised other FX traders
at BANK. Over 80% of DEFENDANT’s 2011 compensation resulted from bonuses for performance.
3. Beginning in 2007, co.Conspirator I (“CC-I”), a citizen of the United Kingdom and resident of London,
England, was an FX trader at the London branch of BANK. In 20 I I, CC-1 was a senior FX trader at BANK
in London, and supervised other FX traders at BANK. Over 80% of CC-I’s 2011 compensation resulted from
bonuses for performance.
4. The CLIENT Company (“CLIENT”) was a publicly traded technology services company. Person 1 and Person
2, whose identities are known to the Grand Jury, were CLIENT employees, and were responsible for
managing foreign currency transactions on behalf of CLIENT. Both Person 1 and Person 2 worked at
CLIENT’s primary place of business, which was located in Palo Alto, California.

II. Relevant Definitions

OMMITED

III. The Scheme to Defraud

 13. From at least in or about August 2011 and continuing until at least in or about October 2011, the defendant
DEFENDANT and CC-I, together with others known and unknown to the Grand Jury, devised and
executed a scheme to defraud CLIENT of money and property.
14. The purpose of the scheme to defraud was to enrich BANK, to benefit the members of the conspiracy, and to
conceal the existence of the scheme.
15. In furtherance of the scheme to defraud CLIENT and to effect the objectives thereof, DEFENDANT and his
co-conspirators used the following methods and means, among others:
a. Misappropriating CLIENT’s confidential information (namely, knowledge that CLIENT intended to sell
a large quantity of cable options in September and October 2011) in violation of a duty of trust and
confidence that BANK, DEFENDANT, CC-I, and other agents of BANK owed to CLIENT by using
CLIENT’s confidential information, to CLIENT’s detriment, to engage in trading activity calculated to
reduce the value of the options that CLIENT planned to sell to BANK;
b. Making, and causing to be made, false, fraudulent, and misleading material misrepresentations,
omissions, and half-truths to CLIENT regarding the role that BANK and its agents played in devaluing
CLIENT’s options and the trading positions that BANK had taken between September 27, 2011 and
October 4, 2011; and
c. Depriving CLIENT of its right to control its money and property by (i) concealing their plan to depress
the value of CLIENT’s options; (ii) making false statements about the manner in which BANK traded
FX between September 27, 2011 and October 4, 2011; (iii) advising CLIENT to delay the sale of its
options so as to provide the members of the conspiracy with additional time to manipulate the FX
market in BANK’s favor; (iv) providing CLIENT with false, deceptive, and intentionally misleading
explanations as to why CLIENT’s options declined in value just before BANK acquired them; and (v)
falsely promising CLIENT they would maintain the confidentiality of its information.

A. CLIENT Acquires Acquisition Target

16. In or about 2011, CLIENT negotiated to acquire Acquisition Target Corporation PLC (“Acquisition Target”),
which was an entity engaged in computer software development and distribution. Acquisition Target
maintained dual headquarters in San Francisco, California, and Cambridge, United Kingdom. CLIENT
engaged BANK to provide it with financial advisory services in connection with the contemplated
acquisition of Acquisition Target.
17. United Kingdom government regulations require that a foreign entity seeking to acquire a British company
have access to sufficient pounds to complete the transaction. Accordingly, before publicly announcing its
intent to acquire Acquisition Target, CLIENT had to ensure that it had ready access to several billion
pounds.
18. In or about August 2011, representatives of CLIENT consulted with representatives of BANK concerning
methods to satisfy this requirement, including through different types of FX transactions.
19. There are several ways that a United States entity can gain access to pounds. An entity can exchange dollars
for pounds by purchasing pounds at a specific exchange rate in an FX spot transaction. Alternatively, an
entity can purchase cable options, which offer the right to exchange dollars for pounds at a pre-determined,
fixed cost. Since FX options do not require the entity to actually purchase pounds, but instead give the
entity the ability to do so if it chooses, FX options allow an entity to maintain maximum flexibility.
20. In consultation with BANK, CLIENT determined that it would purchase approximately £6 billion worth of
cable options from BANK. These options would provide CLIENT with the right to exchange dollars for £6
billion at a pre-determined, fixed cost.
21. In August 2011, CLIENT engaged BANK to assist it in purchasing £6 billion worth of cable options in
advance of the public announcement of CLIENT’s planned acquisition of Acquisition Target. BANK sold
CLIENT £6 billion worth of cable options with an expiration date in February 2012.
22. BANK received compensation for its involvement in CLIENT’s August 2011 purchase of cable options.
23. On or about August 18, 2011, CLIENT publicly announced its intention to acquire Acquisition Target for
approximately $10.3 billion.
24. In or about September 2011, CLIENT determined it no longer needed the options, and would not use them as
part of the acquisition of Acquisition Target. CLIENT decided to “unwind” the cable options by selling
them back in the FX market in several increments, or “tranches.” CLIENT’s representatives engaged in
discussions with BANK about selling the cable options back to BANK.
B. BANK’s Obligations to CLIENT

25. BANK and its agents, including DEFENDANT and CC-1, owed CLIENT a duty of trust, confidence,
honesty, and disclosure.
26. BANK and its employees undertook a duty to keep confidential information provided by CLIENT. On or
about August 17, 2011, BANK and CLIENT executed a confidentiality agreement governing CLIENT’s
acquisition of Acquisition Target, in which BANK undertook to keep CLIENT’s information confidential
and “only use the Confidential Information with respect to providing financial advisory services to the
Company.” BANK and its employees represented to CLIENT that they would maintain the confidentiality
of information regarding CLIENT’s plan to unwind the options because public dissemination of this
information could result in trading by other market participants, and cause CLIENT’s options to decline in
value.
27. Accordingly, on or about September 27, 2011, DEFENDANT promised representatives of CLIENT that “this
will be kept very quiet” and stated that a breach of Client’s expectation of confidentiality would be “a fire-
able offense.”
28. In the course of the discussions concerning whether CLIENT would award the unwind to BANK, BANK
employees made representations to CLIENT about acting in CLIENT’s “best interest.” Specifically, on or
about September 27, 2011, DEFENDANT advised representatives of CLIENT that it would be in their
“best interest” to trade on specific days and promised that he and BANK “[would] go to the mat” to obtain
a favorable price for CLIENT on the unwind. During this same conversation, Person 1 advised
DEFENDANT that CLIENT was “asking for a pretty trust-type exit because I think you’ll treat us fair,” to
which DEFENDANT responded “yep, and I’ll try to step up as much as I possibly can for that.”

C. Execution of the Scheme to Defraud

29. On September 8, 2011, DEFENDANT spoke with Person 1, and discussed the mechanics and pricing of
CLIENT’s planned unwind of the cable options.
30. At or around 7:35 a.m. Pacific Time on or about September 27, 2011, DEFENDANT engaged in a telephone
conversation with Person 1 and Person 2. DEFENDANT told CLIENT’s representatives that the conditions
in the market were not ideal for CLIENT to immediately unwind £2 billion worth of cable options.
31. During that telephone conversation, DEFENDANT informed CLIENT’s representatives that BANK FX
traders were “not touching the market” and were “not doing anything,” when in fact DEFENDANT
intended for BANK to engage in FX trading calculated to depress the value of CLIENT’s cable options.
32. Having promised CLIENT that BANK FX traders were “not touching the market,” at or around 8:25 a.m.
Pacific Time on or about September 27, 2011, DEFENDANT and CC-1 held a telephone conversation
during which DEFENDANT declared that “we need to figure out what to do with this information” (i.e.,
the information conveyed to BANK by CLIENT several minutes earlier). CC-1 noted that “it would be nice
to short, short ahead of” the planned unwind. DEFENDANT and CC-1 agreed to sell options, so as to
affect the value of cable option volatility and depress the value of Client’s cable options.
33. During the conversation referenced in paragraph 32 above, DEFENDANT derided Person 1 as “the kiddie”
who is “playing a little poker here” and remarked to CC-1 that Person l’s suggestion that CLIENT was
prepared to trade that day was “fucking scary.”
34. On or about September 27 and 28, 2011, DEFENDANT and CC-1‘s subordinate traders sold FX options to
ensure that BANK maintained a “short” position with respect to cable, i.e., that BANK had negative net
ownership of cable options.
35. At or around 6:30 a.m. Pacific Time on or about September 28, 2011, DEFENDANT engaged in a telephone
conversation with Person 1 and Person 2 during which DEFENDANT observed that volatility for cable
options had fallen. During this conversation DEFENDANT attributed the fall in volatility to a lack of
“resolution in Europe” and a “stock market rally”, but omitted and concealed the fact that BANK had been
placing trades calculated to depress the price of cable option volatility.
36. On or about September 28, 2011, at DEFENDANT’s direction, BANK FX traders executed the first tranche
of the unwind by purchasing £2 billion in cable options from CLIENT.
37. On or about September 29, 2011, in anticipation of the second tranche of the unwind, CC1 instructed BANK
FX traders in an electronic chat to sell cable options in a way that would further depress the price of cable
option volatility. Specifically, at or around 4:20 a.m. Pacific Time, CC-1 told BANK FX traders that
“tomorrow m[or]ning, sell a[g]gressively ahead of [the] next tranche.’’ During this chat exchange, CC-2,
 who was an FX options trader employed by BANK and whose identity is known to the Grand Jury,
suggested to other BANK traders, including CC-1, to “HAMMER THE MKT [market] LOWER.”
38. From on or about September 29, 2011 through on or about September 30, 2011, BANK FX traders sold large
amounts of cable options, reducing BANK’s inventory of cable options to a “short” position and depressing
the price of cable option volatility.
39. During a telephone call at or around 6:14 a.m. Pacific Time on or about September 29, 2011, CC-1 advised
DEFENDANT that he intended to continue selling cable options in front of the next tranche: “I was going
to call out tomorrow morning and basically bash the shit out of this again.” DEFENDANT then informed
CC-1 that he cautioned another BANK trader to be discreet in order to avoid the attention of senior
executives at BANK: “if it gets back to CLIENT by some loose lipped market monger [] that we’re selling
cable off of them or we’re getting out of a six yard option (i.e., a £6 billion option) over the course of a
week it will go straight to [the head of BANK’s United States operations] and your ass will be in a fucking
frying pan in November.”
40. During an electronic chat at or around 6: 15 a.m. Pacific Time on or about September 29, 2011,
DEFENDANT advised Person 1 that BANK was “long” on cable options when he then and there well
knew that BANK was in fact “short” on cable options (i.e., BANK had taken a net negative position).
During this chat, DEFENDANT attributed the decline in volatility to an “equity rally” but did not disclose
that BANK had been trading in a manner calculated to depress volatility.
41. During a telephone call beginning at or around 6:43 a.m. Pacific Time on or about September 29, 2011,
DEFENDANT instructed CC-1 to continue placing trades calculated to depress volatility: “you know you
should like offer this fucking shit down,” to which CC-1 responded “okay fine.”
42. During a telephone call beginning at or around 7:19 a.m. Pacific Time on or about September 29, 2011, CC-1
advised DEFENDANT that “tomorrow morning we were going to call out and just spank the market like
good and proper. Right, so we’d be [better on] the second clip (i.e., tranche).”
43. During an electronic chat at or around 7:28 a.m. Pacific Time on or about September 29, 2011,
DEFENDANT advised Person 1 to wait to execute the next tranche of the unwind, which had the effect of
providing BANK with additional time to manipulate the cable options market.
44. During an electronic chat at or around 7:38 a.m. Pacific Time on or about September 29, 2011, CC-1 advised
other BANK traders that “we need to sell a[g]gressively between now and tomorrow … tom[orrow]
mngwe call out and sell shed-loads” (sic).
45. On or about September 30, 2011, CC-1 and BANK FX traders, at DEFENDANT’s direction, sold FX
options, thereby depressing the price of volatility and diminishing the value of Client’s options.
46. During a telephone call at or around 3:41 a.m. Pacific Time on or about September 30, 2011, DEFENDANT
and CC-1 agreed to lie to representatives of Victim Company by stating that the decline in the price of
volatility had been caused by other banks’ activity in the market, in order to conceal the scheme.
DEFENDANT cautioned CC-1 that if asked why the price of volatility had declined, “you and me need to
independently stick to the story of… we haven’t been offering the cable curve, other banks have.”
47. During a telephone call at or around 7:29 a.m. Pacific Time on or about September 30, 2011, Person 1
observed that the price of volatility had fallen and asked CC-1, “so, what did you guys do this morning that
like caused it to dive like it did?” CC-1 represented to Person 1 that the decline in volatility was caused by
other banks’ activity in the market but concealed the fact that BANK had been trading in a manner
calculated to depress volatility.
48. On this same call, in response to a question from Person 2, CC-1 stated that BANK had sold “four-fifths” of
CLIENT’s first tranche. CC-1 failed to mention that BANK had sold other options, and that BANK’ selling
had the effect of depressing the price of volatility.
49. During a telephone call at or around 8:52 a.m. Pacific Time on or about September 30, 2011, DEFENDANT
and CC-1 recognized that the market had moved in a way that would harm CLIENT. DEFENDANT told
CC-1 that “I have to tell [Person1] in a very nice way that instead of letting us fuck you for three or four
million dollars, you perfectly supervised the market at large fucking you in the ass for 25, so
congratulations.” Asked by CC-1 to “say that again,” DEFENDANT reiterated that “instead of allowing us
to charge you four million dollars, you perfectly supervised the market at large fucking you for 25.”
50. During a telephone call at or around 10:36 a.m. Pacific Time on or about September 30, 2011,
DEFENDANT told Person 1 that the stock market and other banks’ activities were likely the causes of the
decline in the price of volatility. DEFENDANT did not disclose the material fact that BANK FX traders
had manipulated the price of volatility to CLIENT’s detriment.
51. On or about October 3, 2011, BANK bought CLIENT’s remaining £4 billion worth of cable options.
 52. During a telephone call at or around 6:20 a.m. Pacific Time on or about October 3, 2011, DEFENDANT
spoke with CLIENT’s representatives by telephone. DEFENDANT told Person 1 that DEFENDANT had
“literally done everything on [his] end to tighten [the spread] to as tight as what everybody would be
possibly comfortable with [at BANK].” DEFENDANT did not disclose the material fact that BANK FX
traders had manipulated the price of volatility to CLIENT’s detriment.
53. On or about October 4, 2011, after BANK completed its purchase of the second tranche from CLIENT,
Person 1 asked DEFENDANT how BANK was able to execute the transaction without “the market
reacting in a panic” like it had before. Rather than disclosing that the price of volatility had not declined
further because BANK traders, having completed their purchases of options from CLIENT, stopped
manipulating the market, DEFENDANT stated that the unwind of the second tranche did not disrupt the
market because BANK could “lob out little prices through our franchise and through the brokers and not
have to run out and ‘hit bids’ and get aggressive with the market” (sic).
54. The scheme led CLIENT to lose millions of dollars in the value of the cable options it had originally
purchased and enabled BANK to make millions of dollars by acquiring the options from CLIENT at a
discounted and favorable price.

COUNT ONE: (18 U.S.C. § 1349-Conspiracy to Commit Wire Fraud)

55. The factual allegations in Paragraphs 1 through 54 are re-alleged and incorporated by reference.
56. From in or about August 2011 through in or about October 2011, in the Northern District of California and
elsewhere, the defendant, DEFENDANT, CC-1, CC-2, and others known and unknown to the Grand Jury,
did knowingly conspire to devise and intend to devise a scheme and artifice to defraud as to a material
matter, and to obtain money and property by means of materially false and fraudulent pretenses,
representations, and promises, and by omissions and concealment of material facts with a duty to disclose,
and, for the purpose of executing such scheme and artifice and attempting to do so, to transmit, and cause
to be transmitted, by means of wire communication in interstate and foreign commerce, certain writings,
signs, signals, pictures, and sounds, in violation of Title 18, United States Code, Section 1343.

All in violation of Title 18, United States Code, Section 1343.

COUNTS TWO THROUGH SEVEN: (18 U.S.C. §§ 1343 & 2-Wire Fraud and Aiding and Abetting
OMMITED

57. The factual allegations in Paragraphs 1 through 54 are re-alleged and incorporated by reference.
58. From in or about August 2011 through in or about October 2011, in the Northern District of California the
defendant did knowingly, and with intent to defraud, devise and intend to devise, and willfully participate
in, a scheme and artifice to defraud and to obtain money and property by means of materially false and
fraudulent pretenses, representations, and promises, and by omissions and concealment of material facts
with a duty to disclose.

Each count a separate offense, in violation of Title 18, United States Code, Sections 1343 and 2.

1 See http://www.nasdaq.com/investing/glossary/f/front-running
2 See https://www.fdic.gov/deposit/deposits
3 For a quick summary on this, see Onaran (2017).
4 According to the SEC, “A limit order is an order to buy or sell a stock at a specific price or better. A buy limit order can only be executed at the limit
price or lower, and a sell limit order can only be executed at the limit price or higher. A limit order is not guaranteed to execute.” (See
https://www.sec.gov/fast-answers/answerslimithtm.html.)
5 See http://www.bbc.com/news/business-36768140
6 See https://www.justice.gov/opa/pr/global-head-hsbc-s-foreign-exchange-cash-trading-desks-arrested-orchestrating-multimillion
7 Indictment of Mark Johnson and Stuart Scott, U.S. v. Johnson et al., case no. 16-CR-457, US District Court for the Eastern District of New York.
8 https://www.bloomberg.com/news/articles/2018-04-26/ex-hsbc-currency-trader-is-sentenced-to-two-years-in-prison
9 https://www.glassdoor.com/Salaries/new-york-city-back-office-salary-SRCH_IL.0,13_IM615_KO14,25.html
10 The European Union’s Markets in Financial Instruments Directive, 2004/39/EC (in force November 1, 2007).
11 Global Code, Principal 14.
 6
FX Options: An Intuitive Approach

In Chapters 4 and 5, we explored the theory and mathematics behind the most basic FX instruments, the
environment and dynamics that drive the people involved in FX trading. We hopefully provided a sense of how all
this is changing. For the next four chapters, we will go to the heart of the book, and look at FX options. By
definition, this involves understanding the behaviour of instruments that are mathematically derived from spot.
Therefore, we will need to use more mathematics and you will see why it was important to understand the FX spot
and forward market beforehand.
In this chapter, we present the practical reasons why options are an awesome hedging and investment alternative,
and then go through some basic rules and market practices. As we have seen, there are some very smart people in
these markets and that the devil is in the details. The rules and how they are derived and applied often make the
difference between profit and loss, a successful hedge or looking for a new job.

THE HEDGER’S PARADOX

Suppose that the treasurer of a corporation is dollar-functional1 but has euro revenues. Assume that those revenues
are being realised in one year, and they decide not to hedge the FX risk. It is possible that, over the following year,
the euro appreciates and the choice not to hedge was a prescient one. It is also possible that the euro will depreciate,
and in such a scenario they would have a lot of regret. The regret arises because:

the dollar value of the euro-denominated revenues has fallen, and this could impact operating margins; and
depending on the severity of the depreciation, it can jeopardise their standing in the company and, in extreme cases,
even mean that they sold the goods or service at below cost price.

Let’s examine these scenarios in more detail. Assume the spot rate is US$1.08/€ and the one-year forward rate is
US$1.10/€. The treasurer decides to hedge, and so they proceed to enter into a forward contract with their bank to
sell a €10 million amount for value date one year from now at US$1.10/€. The amount in euros corresponds to the
amount of the future euro revenues. Therefore, in one year these European receipts should result in US$11 million in
revenue when converted to US dollars. Assume the euro falls today from US$1.08/€ to US$1.07/€. Also assume the
interest rates in the US and Europe are constant for our scenario, so that the difference between spot and forward
points is constant.2 As the value (the daily mark-to-market) of their exposure falls, the settlement on the forward
contract will compensate. In other words, the company loses US$.01/€ in the expected dollar revenue from the
European sales but will gain US$.01/€ from the contract they entered into with the bank. That is, the underlying
exposure and the hedge are completely inversely correlated.
In the scenario where the euro rises today, say from US$1.08/€ to US$1.09/€ and the one-year forward goes from
US$1.10/€ to US$1.11/€, the European revenues will have a larger value in dollars. However, whatever gains they
make on the potential conversion of the revenues (€10 million at US$1.11/€ is US$11.1 million) they will have to
give up to the bank, who will ask for the exchange of €10 million and give them only US$11 million as per their
forward contract. They can end up regretting that they hedged when in fact the euro rose. Now this sounds a bit
unrealistic. Why can’t they just be happy with the guarantee of US$11 million?
Most companies enter into this strategy and hedge forward. They are generally satisfied to lock in their revenue
and ignore what happens subsequently. However, that can be a bit frustrating. A few practical considerations come
into play. All the other stakeholders, their boss for one, tend to be great Monday morning quarterbacks, and may not
be happy to have given up the opportunity for more dollar revenue. Consider that the euro has had moves of more
than US$0.10/€ a year in the last few years. This means your revenues could have been almost 10% higher. That’s a
big deal. Also consider what happens if you hedge and your competitors do not. They can lower their price by 10%
and be just as profitable. There will be more on this in later chapters, but for now remember that locking in a hedge
with a forward hedge locks in the amounts you are exchanging.
Figure 6.1 describes visually what hedging with a forward looks like. The black line depicts the potential to have
more US dollars at year-end if the euro rises but also to have fewer US dollars if the euro falls. So, take a point on
the graph say, 1.05 on the x-axis. The €10 million will be converted to US$10.5 million and the P&L in US dollars
(in the right-hand side vertical axis) will be a negative US$500,000. The forward contract, which is the light grey
line, behaves exactly the opposite. As the euro rises, the contract loses money. As the euro falls to, say, 1.05, the
contract is worth US$500,000, which would make up for the loss on the underlying. When they enter into a forward
contract, it gains and loses exactly in the opposite way to the underlying position. That’s what you want! In the end,
you will end up with your originally anticipated US dollar revenue of US$11 million.
Why can we not all be rational and accept that regret can happen, and refrain from second-guessing hedging
decisions? The truth is that human nature is irresistibly attracted to narratives. When, in the early 2000s, Southwest
Airlines was famously the only airline to have hedged its fuel cost in a rising-price environment, it made a killing as
all its competitors were then facing higher fuel costs. It was hailed a genius by the markets and the press. After the
global financial crisis of 2007–08 and the consequent dramatic drop in oil and jet fuel prices, however, Southwest’s
hedging contracts didn’t look so great anymore. It had locked in its fuel costs at many US dollars per gallon higher
than the current market. Suddenly, Southwest was criticised in the press. What is the difference between being a
market genius and a fool? It can be very slim indeed. The disproportionate importance of hedging decisions forces
human nature, in hindsight, to attribute a narrative that labels people as geniuses or fools after the fact.

Here is the hedger’s paradox. Both hedging and not hedging can lead to regret. We need a clear vision of the
future – we need a crystal ball! Rejoice: crystal balls exist, but they are not free and they are called options. An FX
option, which acts as a crystal ball for our corporation in the above example, is called a “one-year at-the-money
forward euro put dollar call”. It is a euro put because a put gives one the right to sell – in this case, they will “sell”
their euro revenue to convert them into US dollars. It is called “at-the-money” (ATM) or “at-the-money forward”
(ATMF) as it might allow you to hedge at the rate available today for a one-year forward that is at the market rate, or
the forward rate.
Let us review the sequence of events by which the tremendous power of the crystal ball is used for the benefit of
the corporation. Today, the treasurer purchases the crystal ball (ie, the option) from a bank for some amount called
the premium. The premium is calculated using the most famous formula in all of finance, the Black–Scholes
formula. Then they stick the option in a drawer, and they don’t look at it for one year.
 After one year, the treasurer looks at the EURUSD exchange rate because the moment to play Monday morning
quarterback has come. If the euro had risen from a year earlier, then the crystal ball would have told them not to
hedge, so they would not have sold their euros to the bank. They happily exchange their euros at the prevailing rate,
which is higher than US$1.10/€, and get more dollars than the US$11 million which the forward contract would
have imposed. No regret!
In the opposite scenario where the euro had fallen from a year earlier, the client would have the bank know that
originally, on advice from the crystal ball, they had correctly forecasted that the euro would fall, and therefore they
had hedged their revenues using a forward contract at the then prevailing forward rate. Again, no regret! What a
beautiful thing – we have eliminated the possibility of regret. The next section will examine the application of this
methodology to life.

A CRYSTAL BALL AND SOME PRACTICAL CONSIDERATIONS

By the way, when speaking about FX options, it is important to be very clear on what the “put” is on and what the
“call” is on. Some of our readers may be familiar with equity options, and you would typically say “a call on XYZ
corporation”. A call is the right to buy, so it is understood that if you are buying the right to buy XYZ stock at a
price, you will give US dollars in exchange. In currencies, that is not so obvious. I may choose to buy a euro call (I
want the right to buy euros), but in exchange I may not give you US dollars but UK pounds. That option would be a
euro call/pound put. There are hundreds of potential combinations, and one would be surprised about how many of
those are liquid and trade extensively. It is important to understand that these are not two options. A euro put/dollar
call are two faces of a coin, not two coins (so much anxiety generated by lingo).
The time horizon of one year in our example corresponds to the moment in time when the client will have to sell
the euros, but options exist with any time horizon, and such time horizons are called the “tenor”. For example,
people will say that our option has a tenor of one year. Any specific date can be the date that the euros are sold, with
this date being called the “settlement date” or “value date”, of the option. Therefore, our option has a tenor of one
year and a settlement date of February 26, 2021. Spot in EURUSD settles in two days, so two days before the
settlement date is February 24, 2021. On that day, at 10am New York time, the client will have to inform the bank
about what they had seen in the crystal ball one year earlier. In our example, February 24, 2021, would be called the
“expiry date” of the option. Sometimes people decide that, instead of 10am New York time, the decision has to be
rendered at 3pm Tokyo time – and that is just as good, and very convenient if you live in Japan. The time of the
decision, be it New York or Tokyo, is called the “cut”. The cut convention depends on the currency. For example,
USDMXN (dollar Mexico) options expire at 12.30pm New York time.
One important detail – the option we have been talking about gave us the better choice between a forward contract
hedge and no hedge. There are other choices in our crystal ball selection that might give you a different rate than the
forward rate. Imagine that EURUSD spot is at US$1.08/€ and the one-year forward rate is US$1.10/€, but we want
an option that would potentially allow us to sell euros at a rate of US$1.08/€. Of course, US$1.08/€ is not as good as
US$1.10/€, but there are many cases in which US$1.08/€ would be a very good rate. For example, in a case where
the spot at the one-year date ends up at parity (US$1.00/€), the ability to sell at US$1.08/€ would be very valuable.
Therefore, this particular option should still have a price, albeit a smaller one than an option whose strike is equal to
US$1.10/€, which is the ATMF rate. The particular rate at which you would be able to sell is called the “strike” of
the option.
These are the key parameters defined so far:

put or call – the right to sell (or buy) the base currency;
strike – the rate at which the currencies will be exchanged;
notional – the amount of currency that will be exchanged;
expiry date – the day on which the right ends and the option can be exercised;
tenor – the time difference between today and the expiry date;
premium – the price of the option today; and
Black–Scholes – the name of the formula used to calculate the premium.

At this point, we can start thinking about how we will be able to determine the price of this crystal ball. We begin
with a few easy considerations. First, the lower the strike the lower the price of the euro put US dollar call option. In
comparing the two options from a theoretical perspective, the ability to sell euros at US$1.10/€ must be more
valuable than the ability to sell euros at US$1.08/€.
Second, the tenor of our option is important. If the option’s tenor is two years instead of one year, then the cost
(the premium) will be higher. The reason is that it is more difficult to predict the future two years out. Another way
to look at it is to realise that a two-year insurance policy must cost more than a one-year insurance policy. Finally,
Figure 6.2 may help in understanding that, in two years, there is a higher probability that certain scenarios may
occur and therefore the risk and the cost should be higher. Although this is not a rigorous proof, imagine that within
one year an imaginary exchange rate can move from a starting point of, say, 100 in the same way as in Figure 6.2.

This means that with equal probability the currency will be 101, 100 or 99. Respectively, the value of the option
at expiry will be 1, zero and zero (in the second case, the option is exercised with gain zero or not exercised, still
with gain zero; in the third case, the option is not exercised, so again its value at the expiry date is zero). If we
multiply the possible expiration gains with the respective probabilities, we get 33% × 1 + 33% × 0.00 + 33% × 0.00
= 0.33, which is the value of the option today. This procedure gives the wrong price for the option, but it will be
helpful just to illustrate the argument.
If the option is now two years in tenor, then our picture changes to that of Figure 6.3. Here, we see the value at the
end of the first year, and since the option’s life is not over we continue for a second year, with three possible
scenarios for each of the scenarios at the end of year one. Logically, there is the probability of two consecutive up
moves, from 100 to 101 and then from 101 to 102. The first move has a probability of 33%, then the second move
from 101 to 102 another 33%, both together totalling 11%. The expected value is 11% × (102–100).
 The interesting thing happens for 101. There are two ways to get there. First, if the rate moves to 101 in the first
year and then stays there, or the rate stays at 100 in the first year and then moves up in the second year. Therefore,
the probability to get to 101 is 22%. We run all the numbers and then sum up the values to get the price of the option
today, and find out that the total price is 0.44, which is bigger than 0.33, the price of the one-year option. This is
why the longer the tenor, the higher the price of the option. Don’t stress about some of the calculations of the value
if it’s not clear, we will revisit this many times in the coming pages.
The third factor to keep in mind when pricing the option is the amount of euros that it is good for, also known as
the notional. Obviously, an analogous option good in €20 million instead of €10 million will have double the cost.
The reason being that an option on €20mio is absolutely the same thing as owning two options on €10mio each.
Note that you will always either exercise or not exercise both. There is no scenario in which you would exercise one
but not its twin. Clearly, the price of the option is linearly dependent on the amount, which is, as we mentioned,
called the notional of the option.
Now for some practical considerations. On the expiry date and time of the option, suppose that the strike price
was US$1.10/€ and the notional was €10 million. Also suppose that the spot rate at that exact time of expiration is
US$1.05/€. Remember the corporation has these revenues from Europe hitting the income statement in two days’
time, at the settlement date. As it stands, this is not good news because they were hoping for these euros to be worth
US$11 million, but right now it seems they will actually be worth only US$10.5 million. All is not lost, however, as
by exercising the euro put option the company has the right to sell the euros at a rate of US$1.08/€. In other words,
there is no need to call a bank and ask for a quote – instead, the treasurer can just contact the bank that sold them the
option at the expiry date and time, and tell it they are selling €10 million at US$1.08/€. The bank will tell them that
it is sending US$10.8 million.
This means that the option protected them from losing an additional US$300,000. How much is this benefit
worth, at the time of exercise? It surely is worth something to be able to sell euros at US$1.08/€ when the market is
at US$1.05/€, but how many US dollars is this benefit worth? After expiration, we know that this benefit was worth
US$300,000, but how much was it worth at the trade date? Stated differently, we know that if the corporation had
spent US$300,000 for the upfront premium, our breakeven point would have been an ending spot rate of US$1.05/€.
Unfortunately, we do not have this certainty when entering into the option. As we will see, calculating this premium
is where most of the effort in option pricing goes, and it as much art as science to determine if the price is right.

OPTION EXPIRATION, CALCULATING P&L

Before we get knee-deep into the option premium, let’s get more practice in calculating the value of an option at
expiration, and creating some visuals around this concept that can help us draw broad conclusions quickly in the
future. The following exercises will provide an opportunity to calculate a specific point on the spectrum of possible
outcomes, while Figure 6.4 will visualise the outcomes for the entire spectrum. We take the opportunity with these
exercises to make some additional points about how options are quoted, and to highlight other option characteristics.
Therefore, if you are not very comfortable with options, we suggest you go through these exercises.

Exercise 6.1: At expiration, calculate in dollars the value of a US$1.08/€ euro put/dollar call, on a notional
of €10 million, the spot level at expiry is US$1.05/€. This was originally a 1 year option and you paid
US$300,000.
Traders refer to the premium in particularly nuanced ways based on what is most convenient. More on
this later. Notice that at expiration, your decision to exercise or not is not dependent on the premium you paid
nor the original term of the option. All you consider is that you can sell euros at US$1.08/€ and the best you
can do in the market right now is US$1.05/€, so you exercise.

Solution 6.1: You call the bank and say, “I exercise my euro put.” The bank is now expecting €10 million
euros in their European bank account and they deposited US$10.8 million in your US dollar account. In
market parlance, you are short €10 million euros and long US$10.8 million. Then you need to fulfil your
euro obligation. If you have revenues coming into the account that’s great, otherwise you need to go to the
market and buy €10 million at the prevailing rate of US$1.05/€, and spend US$10.5 million dollars. In
aggregate then our euro position is now zero and in our dollar account we had US$10.8 million and have
spent US$10.5 million, for a net US$300,000.
 In formulas: US$ Profit = € Notional × (Spot at Expiry – Strike).

What’s missing from this formula? The premium you paid! Therefore the complete3 formula is:

US$ Profit = € Notional × (Spot at Expiry – Strike) – US$ Premium Paid.

Let’s examine the graph of the profit profile at expiration (see Figure 6.4). The y-axis represents the US dollar
profit, and the x-axis represents the spot rate at expiry. At US$1.05/€, as discussed above, you were left with
US$300,000 profit after exercise, but when you consider that you spent US$300,000 to buy the option you are at
your breakeven point (where the zero profit line crosses the number 1 (①) option value line). If spot at expiration is
at any point above US$1.08/€, you would not exercise the option and your maximum loss is US$300,000, which is
the loss of the premium you paid at the trade date. At any level below US$1.05/€, you are net better off with the
option than if you had not purchased the option. Let’s look at an ending spot of US$1.00/€. US dollar profit = €10
million × (1.08–1.00) – US$300,000 = US$500,000. For all points below US$1.08/€, your profit keeps going up
infinitely (not really, zero will be a hard stop4).

Exercise 6.2: Calculate in dollars, the value of a 115-dollar-call yen put on a notional ¥100 million. At
expiry the spot rate is ¥120/US$. For the record you paid US$15,000 to buy this option.

Solution 6.2: The idea is exactly the same but we mixed up the type of option and the quoting convention.
Would you exercise? You have the right to call (to take possession) of dollars and give 115 yen for every
dollar. The market right now will pay you 120 yen for every dollar. Of course you will exercise and take
possession of all the yen your contact allows at 115 and then turn around and sell it at 120 yen per dollar and
pocket the 5 yen per dollar. So you exercise the option by selling 100 million yen and buying US$869,656.22
(¥100 million/115). Then close out our JPY position by buying back 100 million yen and selling
US$833,333,33 (100 million/120 dollars). This will leave in your account US$36,322.89. Therefore in
aggregate the profit is US$21,322.89.

The formula is:

US$ Profit = JPY Notional × (1/Strike – 1/Spot at Expiry) – US$ Premium

Notice we used 1/strike and 1/spot at expiry for algebraic reasons. More on that later.

PANEL 6.1 VALUE OF A USD CALL AT EXPIRY

We want to conclude with a summary list.

N is the notional, which can be in local currency or dollars, K the strike and S the spot at expiry.
We will use EURUSD and USDJPY to represent the two market conventions to quote spot. The value of
the option at expiry, in case of a USD call is:

1) Local N, USDJPY, USD P&L: N × (1/K – 1/S)

2) Local N, USDJPY, JPY P&L: N × (S/K – 1)
3) Local N, EURUSD, USD P&L: N × (S – K)
4) Local N, EURUSD, EUR P&L: N × (K/S – 1)
5) USD N, USDJPY, USD P&L: N × (1 – K/S)
6) USD N, USDJPY, JPY P&L: N × (S – K)
7) USD N, EURUSD, USD P&L: N × (1 – S/K)
8) USD N, EURUSD, EUR P&L: N × (1/S – 1/K)

Let’s examine the graph of the profit profile at expiration (see Figure 6.5). The y-axis represents the US dollar profit
and the x-axis represents the spot rate at expiry. At ¥120/US$, as discussed above you were left with US$21,322.89
after accounting for the premium. Your breakeven point is near ¥117/US$. If spot at expiration is at any point below
¥115/US$, you would not exercise the option and your maximum loss is US$15,000. You would exercise the option
at all levels above ¥115/US$, but at any level above ¥117/US$ you are net better off with the option than if you had
not purchased the option. Between ¥115/US$ and ¥117/US$, you would still want to exercise but the money you
make by exercising is not enough to cover the entirety of the premium you spent; but you would still exercise to
minimise your losses..

Assume something dramatic has happened in Japan and the exchange rate has changed accordingly. The closing
spot rate for our option is now ¥145/US$. In case you think that is unrealistic, think again. That is the fun and danger
with currencies!

Profit = ¥100 million × (1/115 – 1/140) – US$15,000 = US$164,910.04.

Not bad, right? You’re able to get more than 10 times your investment. You are now beginning to see the power of
options. The most you can lose is US$15,000 and your profit can be unlimited, in theory. Here’s something to think
about, how probable is it that USDJPY will be ¥145/US$ in a year?
 Did you notice the shape of option payouts at expiration? These are called “hockey stick” charts for obvious
reasons. Also note the following when we create the x-axis or spot at expiration – when the underlying currency
rises as spot moves to the right, then a call on that currency will always look like the US dollar call of Exercise 6.2.
The US dollar rises to the right. A put will look like the hockey stick of Exercise 6.1. The euro rises to the right and
falls to the left in that chart. This is for purchased options. For sold options, it’s the mirror image with rotation
around the x-axis. Figure 6.6 shows a classic set of hockey sticks since no options book would be complete without
them.

WHAT IS BETTER, AN OPTION OR A FORWARD?

Let’s start to get a sense of how to compare and contrast these FX instruments with a concrete exercise.
Suppose you are comparing the mark-to-market (MTM) of a forward to sell €10 million at US$1.06/€ versus a
€10 million, euro put/dollar call struck at US$1.06/€. Assume you paid US$0.03/€ (US$300,000) for the option.
Which is worth more at expiration.
This exercise gives us the opportunity to explore two topics. First, we can start to think about how different
instruments compare, and second we can start to explore the different priorities and perspectives market participants
have. If EURUSD at expiration is lower than US$1.06/€, then both instruments have a similar result. Think of it this
way: the owner of both instruments can sell euros at 1.06. The owner of the forward must sell at US$1.06/€ and the
owner of the option has the option, no pun intended. What’s the difference? The premium you paid at the very start.
Now, what happens if the expiring spot is above US$1.06/€? With the forward, you must sell at US$1.06/€ even
though you don’t want to. The owner of the option has forfeited the premium paid, but that is the end of their
obligation. Another way to think about this can be that, if you own an option, you can never lose more than the
premium. If you own a forward, theoretically you can have unlimited losses.
What does a visual of the solution look like? We can approach the solution in a couple of ways. First, one can
look at the comparison from the perspective of a hedger. What is the US dollar amount a company will realise from
its €10 million in revenue if the treasurer hedged with a forward versus an option? Also, note that in this scenario the
treasurer inherited the underlying position. This is the case with many FX participants, such as corporations and
many asset managers (there will be much more on the ramifications of this when we discuss each participant’s
objectives). By the way, very often the letter K is used to represent strike. I assume it comes from baseball.
As Figure 6.7 shows, with a forward you will convert €10 million to US$10.6 million, and the ending spot level
does not matter. However, with the option you will convert the euros to US dollars at US$1.06/€ – although, to be
precise, we must account for the fact that it cost US$300,000 to buy the option, leaving only US$10.3 million for all
levels below US$1.06/€. Above US$1.06/€, the treasurer has the luxury of selling the euros at the market and getting
more dollars than US$10.6 million.
Note that the shape of the resulting position of the option looks like a call on the euro, when in fact we know we
bought a put! Crazy, but true. When you hedge an underlying position with a put, the resulting payout profile looks
like a call. This is a huge concept to digest because it speaks to a fundamental truth – when you buy a hedge, you
want the hedge to be worthless at expiration. You do not want your insurance to have to pay out, and you always
want your underlying position to be successful. Think of it this way – you never want to call up your homeowners
insurance and celebrate the fact that your house burned down just so you can collect a big payout from the insurance
company. This may sound straightforward, but many a treasurer has been berated by their boss or the board because
they spent all that money on option premiums, and they have nothing to show for it. One of my colleagues used to
say that being disappointed because the hedge is “out of the money” is similar to when you buy life insurance and
then hope that someone gets to cash it in (for more ways to combine options with the underlying positions to
transform the total result, see the building blocks in Figure 6.9).

The second approach is to look at the comparison from the perspective of a speculator. The speculator will not
want to end up with the notional euros or dollars in their bank account – all they will have is the net profit or loss.
Therefore, the question they will ask is, “Which strategy will leave me with a greater profit?” For all levels below
US$1.06/€, the profit is the same except for the premium. Above US$1.06/€, and the losses are limited to the
premium. Note that the maximum loss with the option is the premium and also note how the shape of the option
results look like a bought euro put (see Figure 6.8). The essence is exactly the same as that of the hedger; the
differences concern the underlying position. However, the way it is communicated is important, as we will see in
later chapters.
 It is important to understand that, at any point in time before the expiry, there is a multitude of future scenarios
that coexist in the future and that the option is a superposition of all the values that the option assumes in each of
those scenarios. This is not dissimilar from the way quantum mechanics treats the coexistence of all the possible
positions of an electron until an experiment is carried out to determine its actual position. More on this later.

SOME OPTION FUNDAMENTALS

This section lists some basic realities about options and market conventions. We will simply outline the most
relevant, and talk more extensively about the quoting conventions as those have driven option users and
professionals nutty trying to figure out what the market is saying.
The building blocks
In the previous discussion, there was an interesting transformation. When the underlying position was added to the
purchased euro put, the resulting position was essentially a euro call. Well, this happens with all combinations of
options. Figure 6.9 lists a range of these for your reference, but special attention should be given to the very last,
which combines a put and a call and we end up with the underlying. Combining options in this way has many
applications, which we will explore in later chapters. For a range of definitions and practical considerations on
currency options, see Figure 6.10.

Quoting option prices

Let’s examine the six ways to quote an option price. This will demonstrate how needlessly complicated market
conventions can be.
Suppose that you have priced a US$1.06/€ euro call/US$ put with a notional of €10 million.

1. The price is €10,000; this is the first way to express the premium, and is called the euro price.
2. Spot is trading at US$1.03/€, so the dollar premium of the option is €10k × US$1.03/€ = US$10.3k; this is the
second price convention and is called the dollar price. Note that we are going back to basic algebra. The euro
signs cancel out and you are left with the US dollars (this is the same thing as the dimensional equations used
by physicists).
3. The third way to quote the option price is as a percentage of euro notional, which is calculated as €10k/€10
million = 1% of the euro.
4. The fourth way is called a percentage of US dollar notional, and is calculated as [€10k × 1.03]/[€10 million ×
US$1.06/€] = 0.9716% of the US dollar amount. Note that, to calculate the US dollar notional on the option,
we took the euro notional and we multiplied not by the spot rate but by the strike rate.

Let’s extract as much information as possible from what we have. For example, it would be nice to know where
would I break even after I pay this premium (see the above discussion about breaking even the P&L on the option).
Let’s express the breakeven level as a function of the strike of the option. This is where the exercise of the option
exactly recoups the premium spent in US dollars but no more than that. If the final spot level is S, then you would
exercise the option buying €10 million and spending €10 million × US$1.06/€, dollars. Then you would sell back to
the market the euros and pocket €10 million × S dollars. As S is larger than the strike, the net dollar amount is
positive. We want this amount to be equal to the US dollar premium, which is €10k × US$1.03/€, so the equation is:
 €10k × US$1.03/€ = €10 million × (S – US$1.06/€)

The solution of this equation is S = US$0.0103/€ + US$1.06/€.

5. We say that the premium of the option in US dollar pips equals 103; this is the fifth way to express the
premium. One pip equals US$0.0001/€ in exchange rate units. It is human nature to be interested in the
breakeven when we buy something.
6. The sixth and final way to quote the option premium is the same as the fifth way, except that you pretend the
market convention is flipped and that you are quoting EURUSD as USDEUR, so the exchange rate represents
the number of euros that buys US$1.00. This is almost never used, as nobody uses the wrong convention.

You may wonder how these conventions have come to be. The answer is easy. Prices are quoted in the way that the
least number of calculations can be used. The market starts by the way the notional is typically communicated. For
example, when discussing USDJPY amounts the market quotes US dollar notionals. When quoting GBPUSD,
EURUSD the market quotes notionals in UK pounds or euros. Given that the notionals are quoted this way and most
participants want to end up with a US dollar premium, then the maths gets easy if we quote the premium in US
dollar pips when quoting euro (as we saw above), € notional × US$/€ = US$ premium by multiplying. If the notional
is in US dollars versus yen, then we use a percentage of the US dollars, US$% of US$ = US$ premium. However, it
is good to be aware of all the possibilities. Many systems and some professional participants use non-conventional
approaches.

CONCLUSION
This chapter has got some of the basic boring but necessary conventions out of the way. In the process, we explored
how options compare with forwards – specifically, how to calculate P&L at expiration for options versus what we
learned about forwards in previous chapters. Very importantly, we introduced the different way that options are used
and perceived by hedgers as opposed to speculators. Where a hedger looks at the entire exposure, the underlying
plus the hedge, to see if the whole operation was a success, a speculator looks to the option alone. This latter
evaluation more closely resembles the way you would look at an investment in a stock (with a stop–loss) or a lottery
ticket.
We used visuals extensively for a couple of reasons. We all perceive and understand concepts differently, and
some of us need a visual before we can fully grasp the maths. Additionally, we presented basic illustrations of what
option strategies look like so that we can build on these with more complex shapes and mathematics in future
discussions.
Chapter 6 was also the gateway to understanding option valuation at expiration, conventions and practical
considerations. If you are familiar with equity options, this chapter should have given you a way to translate the
ending P&L calculations, as well as the different terminology in FX versus other asset classes.

1 Functional currency refers to the main currency used by a business. It is the monetary unit of account of the principal economic environment in which
an economic entity operates.
2 This is an approximation but will be fine for the purpose of our argument. In actuality, if interest rates in the US and Europe stay the same then the ratio
between the forward all-in rate and the spot rate stays constant. The forward points, the difference between the forward all-in rate and spot, does not
actually stay constant.
3 For argument’s sake we do not discount to present value here, but notice at trade date the premium is present value and the exercise money is future
value.
4 Paradoxically, if EURUSD goes to zero then you have a profit of USD 10.8mio.
 7
Famous Formulas, Fame and Fortune

Having grasped basic option terminology and the straightforward way we can calculate option value at expiration,
we are ready to start exploring option valuations before expiration. This topic is key because calculating option
values at any point in time before expiration allows us to understand what constitutes a fair price, and opens a
window into the FX market’s mind. What does the market expect to happen in the future, how has it reacted in the
past, what levels are important to watch and on what dates?
In the next couple of chapters, we begin our trek to discover the most celebrated formula in all of finance, the
Black–Scholes formula. This formula was discovered and applied in the early 1970s by Fischer Black and Myron
Scholes, whose work in turn was based on previous studies by, among others, Edward O. Thorp. Thorp tried to
make money out of being one of only three people on the planet to know the correct price of options, and went on to
become very successful operating a hedge fund. Black and Scholes also attempted to make money but failed, and so
in 1973 they capitulated and accepted academic fame instead by way of publishing the formula to the entire world
(Poundstone, 2005). At the heart of their formula is getting to the right option premium. We will take two
approaches to explain how you can think about option prices, and to understand how other factors impact the change
in the option premium between inception and expiration.
In this chapter, we will begin our exploration with an appreciation of the role statistics plays in options, and then
leverage that knowledge into real-world adjustments. What are the factors that affect premiums and which ones
matter? The first approach is to look at a fun example using dice, without involving currency. The beauty of this
example is that it allows us to talk about expected value in terms of the fundamental statistics. The second approach
moves us into the currency world, slowly taking us into a bit of mathematics and theory. Here is the punchline so
you can more easily spot the answer:
An option premium represents the expected value of the payout at maturity discounted to the present.

To begin to understand what an option is worth before expiration, we want to get a handle on what the option will be
worth at expiration. We have seen from all our exercises that we only need to figure out where spot will be in the
future and – bingo! – we can then calculate the value. Yeah, good luck with that. If that were possible, there would
be no need to read the rest of the book. A more realistic approach might be to try to get a handle of the possible spot
outcomes, and then we can put some parameters around where spot might be.

ON THE HUNT FOR THE FORMULA WITH NYC HUSTLERS

To intuitively understand the formula, let us begin from something we are all familiar with: rolling dice. Suppose
you are in New York City and excited to visit Wall Street where all the magic happens (ok, not so much anymore, as
the magic now happens electronically …). You come up from the subway and you see a small group of people
gathered around a folding table. There’s a lot of shouting and commotion. The person at the centre of the activity
urges the crowd to step up and make a bet (this may be more out of the New York of the 1980s rather than today,
but let’s go down memory lane).1
The man behind the table looks at you and says: “Hey, Mac, you wanna make US$100? I’ll give you US$100 if
you roll a six with this die.” You pick up the die to roll, but he grabs your hand and demands US$20 before he’ll let
you play. Not a bad deal right? 5:1 odds are pretty good. Being the deliberative person you are, you ask him why
US$20 and try to reason if that is a good deal. How much money should you pay for this amazing opportunity? You
should start by asking what your chances of winning are – that is, getting a six?

probability of rolling a six is one out of six, or one-sixth = 16.67%; and

from your school statistics class, you know that 16.67% * US$100 = US$16.67.
Therefore, the US$20 was too expensive – if he got many people to roll that die, he would earn on average US$3.33
(US$20–16.67).
Let’s expand this method to that of rolling two dice 1,000 times. Can you guess what rolls you will end up with?
It turns out we get 30 rolls of two, 55 rolls of three, and 30 rolls of 12, and so on … (see Table 7.1).
Let’s use this methodology to calculate the probability of getting one of these rolls. For instance, a two rolls 30
times out of 1,000 or 30/1,000 = 3%, a six rolls 140/1,000 = 14%. Figure 7.1 shows the entire spectrum of
possibilities.
Looking at Figure 7.1, let’s highlight a few things.

In every probability distribution there is a point, called the median, such that 50% of the distribution lies to its left
and 50% to its right. In our case, it is the number seven.2 In this case, the median also happens to be the average
or mean of the possible outcomes (remember 1 is not a possible outcome); in FX option theory the forward rate
is the mean. By the way, the forward rate is not the most likely future outcome in empirical tests, but we use it
in the same way we use the number seven in the rolls, as it is the best we have and because we can actually
hedge there. OK, now we are way over our skis, so we will come back to the concept.
Does the visual of the distribution look familiar? It should, as it looks like the normal distribution you learned in
school. This is a perfect distribution, with all outcomes around the mode (seven in our example, or the forward
rate in the FX world) being the mirror image to the left and to the right.

Take the language and the numbers to connect the outcomes to our option premium quest. Let’s bring this concept
closer to options.

PRICING SIMPLE OPTIONALITY

The same NYC hustler, realising that he’s dealing with a sophisticated FX dealer, now says he will pay US$10 a
point for any roll over seven on two dice. You just have to hand over US$10 to play – again this seems like a great
deal. You pay US$10 and you can make US$50 if you roll a 12. Let’s think this through and use all the new tools we
have discovered.
Specifically:

what is the probability of rolling over seven?

what is the expected value of this game? and
what piece of data do we need?

Let’s use the decoder we learned above:

Sum up all the outcomes to arrive at total expected value. This is pretty cool! We can turn around and, with little
regard for our well-being, tell our friend that he’s overcharging by US$0.30.
Now, let’s really get a handle on this. The hustler is willing to pay US$10 for every point over an eight roll. How
much is that worth?

In the above two examples, the hustler was selling you a call option on the dice roll with strike prices at seven and
eight, respectively. Notional amount: US$10.
In summary:

if Result < Strike price, then forfeit price of the game;

if Result > Strike price, the payout is US$10 × (Result – Strike price);
price of game = Premium of option = Expected value of the game.

What would a put look like? Suppose the hustler said that he will pay you US$10 for every point below seven.
Before we go any further, take a guess at what the expected value is, keeping in mind that, in this case, what is
above the mean is exactly the same as what is below the mean.
Let’s extend our analogy to the FX world. Suppose I asked you to price a UK sterling call struck at US$1.25/£ on a
notional of £1 million, assuming no premium for now and the same distribution?

The calculation for the payout is what we discussed earlier. Profit = UK sterling notional × (Spot at expiry – Strike).
For an ending spot of US$1.3000/£, £1 million × (1.3000 – 1.2500) = US$50,000. Sum up the potential ending
outcomes multiplied by their probabilities, and you arrive at US$9,700.
Congratulations, you’re ready to trade! Just kidding, keep your money in your pocket for a while longer.

REAL-WORLD COMPLICATIONS
For our analogy, the two critical parts of the formula – payout and probability – remained constant. After all, unless
the dice are loaded they will always have the same probability distribution if rolled often enough, and they will also
always have seven as the mean. This is not the case with currencies.

Shifting of the forward

In fact, that central (unbiased) outcome can shift in the blink of an eye – actually, a lot faster than that as currency
rates are updated in milliseconds. The mean3 is the forward rate; however, as we learned in Chapter 4 the forward
rate is composed of the spot rate +/– the points, which are based on the interest rate differentials. OK, so the central
point can shift. What does that imply for our option?
Suppose you bought the above UK sterling call US$ put for US$9,700. You come back the next day and notice
that the forward rate is no longer 1.2500 but 1.2700. What is the value of your option?
Happy days – you have the right to get £1 million in exchange for US$1.25 million dollars, and the current central
exchange rate for the future is US$1.27. After applying the probability for the possible outcomes, your option has
more than doubled in value – you can now make reservations for the best restaurant in town!
What happens if the next day the forward rate drops to US$1.2200/£?

Oops, back to the cheaper corner food stalls for lunch.

Changing distribution
As if the shifting forward rate were not enough, the distribution can change at any time. Let’s examine two potential
changes from the wonderfully symmetric scenario in the previous examples.
Suppose the potential set of outcomes narrows dramatically, meaning only a few outcomes are possible. In the
real world, this can be due to a variety of political and economic conditions. Think of the extreme case where a
central bank decides that their currency will be fixed to all other currencies.
The above scenario is one where everyone is convinced that the only possible outcomes at expiration can range
between 1.22 and 1.28. Note how the price of the option dropped from US$9,700 to US$5,500. Figure 7.2 shows the
payout and the distribution superimposed on each other. This will be a very important consideration in our later
chapters. The grey line is a bizarre looking bell curve, and this is due to our particular choice of probabilities.
As a quick aside for later, if you continue the concept of multiplying each payout by its probability, you can
extend that concept to Figure 7.2, and then you would multiply the grey line by the black line. This would give you a
third line, not shown here. The undiscounted (or “future value”) of the option would then be equal to the area limited
by the third line and the x-axis (just in case you are looking for a fun fact that you can use to impress your family
and friends!).

Suppose there is tremendous uncertainty in the political and economic spheres in the UK. The possibility of the
exchange rate fluctuating increases. In the example below, the probability of the rate hitting US$1.2000/£ went from
3% in the original example to 6%. The greater the set of outcomes, the greater the price for protection against
currency moves.

As we close out this section, remember that option premiums before expiration revolve around the forward rate, and
that the forward rate as well as the probability of future outcomes is in constant flux.

CHOOSING THE RIGHT CRYSTAL BALL

The above sections demonstrated that option premiums are the expected value from a set of potential outcomes.
These outcomes are impacted on by the spectrum of potential payouts and by the probability of any given outcome
occurring. In this section, we will explore the most dynamic and least transparent components of these: the
probability distribution. The section will rely a bit more on higher-level maths – however, if that’s not your cup of
tea, try following along because it will lead us to the famous Black–Scholes formula and some very practical
elements of the market. In the end, most of our readers will want to be able to use options to hedge or to speculate.
This book is not intended for someone who wants to script option instruments in Python or if you have a Bloomberg
terminal in DLIB <go>. It is intended to provide an understanding of what levers make option prices change and
what impacts on those levers. Think of it like driving a car – you can be an excellent driver without knowing how to
build an internal combustion engine.
Before we begin, let’s discuss a few more option conventions and assumptions that may have become obvious to
you from our discussion. Starting with the fact that, if you are hedging, buying an option is like buying insurance.
You pay a premium and if something bad happens (rates move against you), then someone promises to make you
whole. Below are some best-practice considerations to keep in mind when you buy an insurance policy on your FX
risk.

The best insurance is that which is not used (ie, an option that is not exercised provides the best outcome).
Remember that you always want your underlying position to make money.
Do not cash in your insurance when the house is on fire (ie, when an option hedge is profitable, the underlying
position is losing). As you saw in the examples above, if the forward shifts in your favour and/or the distribution
expands, your option may increase significantly in value. The temptation may be to sell the option and cash in
on the profits. However, if you’ve purchased the correct hedge, by definition your underlying position must be
losing money. If you sell your hedge, you risk the market continuing to move against you but now you have no
protection.
When an option is used as a hedge, the premium is not the most you can lose. This is because, as a hedger, you need
to think about opportunity cost. You can always hedge at the forward, but if you choose a strike away from the
forward you are taking a bigger deductible, to continue with the insurance analogy. We will revisit the above
considerations when we talk about hedging as a corporation or an asset manager.

In the examples cited in the previous section, we used a forward rate of US$1.25/£. If, in fact, we buy an option
whose strike is US$1.25/£, then this option is called an at-the-money (ATM) or at-the-money forward (ATMF)
option. If we know the forward is US$1.25/£ but we choose to strike a UK sterling call at something higher, then
that is an out-of-the-money (OTM) option. Similarly, if we choose to strike the call option where we can buy UK
sterling better than the forward, then this option is known as an in-the-money (ITM) option. At this juncture, we will
not demonstrate how choosing the strike impacts the outcome, but there will be many examples in our discussion of
option implementation later in the book. For now, let’s state the following.

ITM: This is either a call option where the asset price is greater than the strike price, or a put option where the asset
price is less than the strike price.
ATMF: This is an option in which the strike price equals the forward price of the underlying asset.4
OTM: This is either a call option where the asset price is less than the strike price, or a put option where the asset
price is greater than the strike price.
Let’s revisit the approach from Chapter 6 and consider the different ways a speculator versus a hedger would look at
where to strike the option. A speculator has a more straightforward analysis – they pay a premium and if at
expiration the closing spot is above the breakeven point (strike + premium), then they are happy. If the plan is to buy
an option and hold it to expiration, then the strike choice is a reflection of the confidence they have in their
projections. Assume, in the above example, you believe the pound will rise, and you are very confident we are
headed to US$1.40/£, and most certainly will end up above US$1.25/£. Then, the forward gives you the highest
profit at every point above US$1.25/£. If you are confident but not 100% sure, or maybe worried that if you are
wrong it may collapse, then an ITM option is best. You pay a lot of premium upfront because the strike is better than
the forward, but if it rises your breakeven is the closest to current levels. Should the collapse happen you have a
definitive stop at the large premium you paid. At the other end of the spectrum, suppose you are just not that
confident – you suspect it will rise, and frankly it may take off and rise a lot, but there is a not insignificant
probability it will go down. So you pay a little premium, to be in the game, and then you let it ride.
 Figures 7.4–7.6 summarise the fact that, as a speculator that will hold this position to the end, you are at the
greatest risk and reward with a forward, and the least risk and reward with an OTM option.
As an aside (and we will revert to this later, when discussing the euro–Swiss “de-peg” of January 2015), if you
are just holding a long position on the spot and you placed a stop-loss order below the current level, you can never
really be sure that your maximum loss will be the amount you lose when you sell your position at the stop-loss level.
The reason is that the execution of your stop-loss order might not happen at the order level you have chosen due to
“slippage” at the time of execution. Slippage means that, at the time the spot fell down to your order level, nobody
wanted to buy it at that level and your order was executed at a lower level instead, increasing your losses. By how
much? Who can be sure? In the famous example of the euro–Swiss de-peg, it was multiple percentage points.
However, if you just bought a call option the worst that can happen is that the option ends up OTM and it is not
exercised, in which case you only have lost the option premium. This is why options can be very attractive to hedge
funds.
What about the hedger? Suppose you are a corporation that has a payable in the UK for the same timeframe and
the same amount, and like the speculator you are convinced to various degrees that UK sterling will rise. Therefore,
you will either buy the pounds forward or buy a pound call/US dollar put. Do you remember the previous chapter
discussion about how a call used to hedge turns the entire position into a put, and a put bought to hedge an
underlying position turns into a call? This changes everything. Recall that when you have an exposure that you do
not hedge, it’s like the speculator who bought a forward. The position can have unlimited profit and unlimited loss.
Given that you value your job and cannot explain to your board massive losses, nor will they give you a seven figure
bonus for massive profits, you choose to hedge the underlying position with a forward. This eliminates the potential
for profit or loss. You know exactly how many dollars you need to meet that payable in the future. Sounds like the
right thing to do. However, if you did that in June 2015 and bought UK sterling at around US$1.60/£, how happy
will your colleagues and boss be in June 2016 when the pound is US$1.30/£? You spent ~30% more on this expense
than you had to. Some may question your judgement. Remember our discussion about regret in Chapter 6?
 Therefore, you decide to buy a crystal ball. But which one? In fact, if you choose to spend a lot of money upfront
and buy an ITM option, unlike the speculator you are actually taking less risk than if you spend less on premium.
The reason is that if things go against your underlying, the pound rises, you bought pounds at a rate better than the
forward. The forward is the yardstick.
As an aside, when measuring whether a hedger has done well or not, it is difficult to decide what constitutes the
best time and rate to hedge. These exposures and revenues are an ongoing occurrence, typically without a predefined
time and date of inception. If we use the forward as the yardstick at an arbitrarily chosen point in time, any slippage
from that forward must be considered an opportunity cost. Therefore, your total potential cost as a hedger is the
distance from the forward, in points plus the premium you spent. We will explore these concepts when we talk about
corporations, but for now look at the details in Figures 7.7–7.9.

CONCLUSION
This chapter has focused on moving us down the road to understanding option pricing before expiration. To
accomplish that we made some intuitive connections between statistics and option pricing, and then began to realise
what one gets for the money. In the process, we explored how two very important components in option pricing are
constantly moving. The first is the probability distribution, or the expectation of where future FX rates can go. The
second is the realisation that the very anchor of our option pricing, the forward rate, moves constantly.
Finally, we explored some basic terminology around option strikes and considerations about choosing the right
price. This is an endless topic that takes on many dimensions depending on what you are trying to accomplish. In
Part III, we will explore some of the possibilities as they concern hedgers, but also examine the many ways
combining different strikes allows for the expression of many risk and reward profiles for speculators. The important
consideration to take away is that even choosing the right option is not that straightforward (pun intended).

1 I am forever grateful to Fred Stambaugh for sharing this approach with me many years ago.
2 Seven is also the mode and the mean of this distribution of outcomes. The mean is the average of all the outcomes, while the mode is the most frequent
value, or the peak of the probability distribution. The mode is the outcome that has the highest probability of happening. For a generic distribution, the
mean, the median and the mode need not be the same outcome.
3 To a level of approximation sufficient for our argument here.
4 In fact, there are various market conventions for the meaning of “at-the-money”. Usually, for G10 currencies and tenors below two years, this means
that the strike equals the forward rate. In all other cases, it tends to means that the strike equals the strike of a straddle (the purchase of a put option
and of a call option with the same strike) whose delta is zero. We will explain what this means in detail.
 8
Getting to the Formula and the Correct Probability
Distribution

This chapter will finally bring us to the derivation of the Black–Scholes formula. To get there, we employ a less
intuitive and more formal approach to understanding how option probabilities and returns on FX can be calculated.
This chapter will explore the heart of option pricing, volatility, and tie together a few disparate concepts about the
premium.
We will try to resolve the issue of how to come up with the correct probability distribution for the future, and use
that distribution to arrive at the right premium. Specifically, we discover the central limit theorem and get a glimpse
into the future. We will begin to understand that assumptions about how a currency moves have a significant impact
on the ultimate price of the option. Then, we look at historical performance to see if that can help predict the future.
Unfortunately, we find history has a limited capacity to divine the future, but if we start with history the market can
then add its subjective assessment to arrive at a probability density distribution and a price. Realising how key the
standard deviation is to the price of the option, we delve deeper into understanding volatility, as well as taking a
look at the different volatilities available to price options.
The chapter closes with a critical evaluation of all the conclusions arrived at, and tries to poke holes into the very
essence of what we worked hard to achieve, an option pricing methodology. We also examine ways to understand
what the option premium is telling us about the tenor and the expected movements of a currency. Finally, we search
for ways to see if the formula is correct and, in the process, discover a crucial activity in FX option market making
and portfolio management: delta hedging.

A QUEST FOR THE CORRECT FUTURE DISTRIBUTION

The central limit theorem
Now that you have an intuitive sense of how to price an option, let’s buttress that with some mathematics, as
intuition can carry you just so far. A few years ago, I found myself disagreeing with experienced dealers about the
final resulting position from a forward deal that was transacted. The traders felt insulted, and somewhat indignant
that I had questioned them. The trading floor has changed considerably since the mid-2000s, and FX spot and
forward trading desks were macho places to work in. Making money consistently involved being in the right place at
the right time, but also a lot of posturing in the market and much intuition as to who was trading what and in what
size (amount). So when I was called to task, I showed the maths and the logic that proved I was indeed right.
The trader’s comeback was that their results made intuitive sense and mine did not. They were not trained
mathematicians, so they did not try to work out a logical argument for their intuition. Guesswork can be the biggest
and most dangerous temptation. In finance, it is very important to become as sure as you possibly can by using all
the tools available. But how can one become more sure? There is no easy answer to that, but we need to go deeper
into finding ways to get to the correct option premium and the Black–Scholes formula. The point being, intuition
alone can be a very dangerous master. You must seek a methodology to prove to yourself that you really understand
what is going on. This can be mathematics or something different, but intuition on its own is not enough. In this
section, we will arrive at the answer by discovering the distribution (the probabilities) with another methodology.
We must start with statistics again, but this time we take a slightly different approach and once again start with a
game.
What is the entry price for the following game? Flip a coin, and if you get heads you pay US$1 but if you get tails
you receive US$1?
Solution: Zero. How did we figure that out? There is a 50% probability that you pay –US$1 plus a 50%
probability that you receive US$1. Therefore, EV = 50% × –1 + 50% × 1 = 0 is the “expected value” of the game.
 So far so good, but unless I can get a loaded coin there is not much benefit in playing this game.
OK, one more. What is the cost to you to enter into this new game: you flip a coin, and if you get heads you pay
US$1 but if you get tails you receive US$2?
Solution: 50 cents. How did we figure that out? There is a 50% probability that you pay –US$1 plus a 50%
probability that you receive US$2 dollar. Therefore, EV = 50% × –1 + 50% × 2 = US$0.50.
This is getting more profitable for you with every example, so let’s keep going. What is the entry price for the
following game? You throw a die, and receive US$1 multiplied by the result of the die throw?
Solution: Many people might think the answer is US$3 because three is half of six, or because it sounds about
right. But not you – because you’ve read the previous chapter you have a much more sophisticated response.

EV = 1/6 × 1 + 1/6 × 2 + 1/6 × 3 + 1/6 × 4 + 1/6 + 5 + 1/6 × 6 = US$3.5

For now, let us assume the interest rate in euros and US dollars is zero; ignore the premium paid for now, and
consider a euro call/US dollar put with a strike price of US$1.06/€ and a tenor of one year. Suppose the spot rate is
US$1.03/€.
Figure 8.1 is a screenshot from an option pricing page from the Bloomberg terminal (OVML). This is what a
professional trader would use to make a real price. Note all the fields we talked about in the earlier chapters. Many
large banks also use their own proprietary software, but in the end the key inputs required are the same. The input
levels may be the same, but that is where this becomes an art rather than a science.

We know that in one year, if the euro is above US$1.06/€, the owner will call up the seller at 10am EST and ask
to buy euros at US$1.06/€. Now, consider what happens at some point before expiration. Suppose there is a 40%
chance that at expiration this option will be exercised. At that time the price should be:

Option price = OP = 40% × Money made exercising + 60% × Money made not exercising

This is too easy! Substitute, “Money made exercising” with “money I receive minus money I pay”:

OP = 40% × (Money I receive minus money I pay) + 60% × zero

If I don’t exercise, nothing happens, so that is a big zero. In the case of exercise, the money I pay is the strike price,
so US$1.06/€. But what do I receive? That is a bit more difficult as the euro could be at US$1.0610/€ or at
US$1.2000/€. We are not able to calculate it at this point. Our strategy will be to estimate where we expect the euro
to be if we exercise the option.
That is, EURUSD is higher than US$1.06/€:

OP = 40% × {Expected EURUSD if EURUSD > 1.06} – 40% × 1.06 + 60% × 0

Now, don’t think that this is in any way an approximation; you are staring at the Black–Scholes formula right now,
the most famous formula in all of finance. Its concept is that simple.
What follows is a more detailed discussion that will clarify the formula and explain how the 40% is calculated
(the probability to end up above the strike), and also how the famous “Expected EURUSD if EURUSD > 1.06” is
calculated.
Both answers come from the knowledge of the probability distribution of yearly returns for EURUSD.1 We will
make a series of reasonable assumptions about the nature of this distribution, and it will then be apparent that this
distribution only depends on one parameter: “volatility”. Knowing the volatility reveals the full probability
distribution curve, from which both the 40% and the “Expected EURUSD if EURUSD > 1.06” are shown to be
functions of the strike with respect to the standard deviation.

It is no exaggeration to say that the only true parameter one needs to calculate the price of the option is where the
strike sits with respect to the volatility parameter. So, in essence there is just one input to the Black–Scholes
formula. Well, in essence …
All right, here we go again with the maths. To calculate our famous “Expected EURUSD if EURUSD > 1.06” we
need a method, which we will now start detailing. Remember, we want to figure out the probability distribution of
the EURUSD exchange rate at the expiry of the option. Start with basic steps. It would be very reasonable to assume
that in the next second EURUSD can either go up by 0.0001 or go down by 0.0001. In this case, the distribution
density function of EURUSD in the next one second looks like that in Figure 8.2.
Now the process repeats for the second second, and in this way we have four possibilities for the path of the
EURUSD exchange rate from time zero to time two seconds:

up, then up;

up, then down;
down, then up; and
down, then down.

These are explained by Figure 8.3.

 These can also be visualised by the “probability tree” in Figure 8.4, where we show the final exchange rate
followed by the probability to arrive at that level at expiry.
Continue by considering what happens after three seconds and three moves (see Figure 8.5).
Increase this process to an infinite number of seconds and the result is a perfect curve like the one we saw when
we rolled the dice. Mathematicians have been studying this curve, called a Gaussian, or normal distribution, for
centuries. This fact about the number of one-second moves going to infinity and creating a smooth bell curve is a
celebrated statistics result known as the central limit theorem (see Figure 8.6).
At the centre, the curve has its convexity pointing down, and at the extreme left and extreme right the convexity
points up, but there are two points, X + 1σ and X – 1σ, where the convexity flips from pointing down to pointing up.
In this case, σ2 will be called the standard deviation3 of the distribution.
 What does this curve really tell us? After making an arbitrary assumption about the way euros can move in the
next one second, we arrived at a distribution, but is that something we can comfortably start using to price options in
the real world? What if we are wrong in our assumptions?

History as a guide to the future

How about learning from history? To gain a better insight about the Gaussian curve, let’s refer to past EURUSD
prices. Suppose we look at various points in time in the past, and in each case compute the one-year return of the
exchange rate. This gives us a very good historical base to estimate the return that we can have during the life of our
option. To then estimate where EURUSD will end up one year from today, we overlap the historical returns to the
current starting point, which is the current forward rate.
To make things clear, suppose on December 31, 2003, EURUSD was at 1.2600 and on December 31, 2004, it was
at 1.3550. This represents a logarithmic return of 7.27% = ln(1.3550/1.26). Based on this historical example, we
would not think that in a year’s time EURUSD will be at 1.3550 from its current starting point of 1.03. This would
be a jump of 0.3250, as opposed to the previous year’s jump of 0.0950, so way too large. However, we may think,
instead, that EURUSD might again have a logarithmic return of 7.27%. Then, in a year from today, at 1.03 × e7.27%
= 1.1077, since 1.03 is the starting point today. You can see that returns drive our assessment of future probabilities
while absolute exchange rate levels do not.
To make systematic this example, take a bunch of yearly returns and compile a histogram. Figure 8.7 shows daily
returns for EURUSD from July 2016 to July 2017 (see top left for data ranges) and for comparison, a perfect
Gaussian curve (denoted by the ①), drawn using the same mean and standard deviation, listed in upper left, as the
historical distribution. The historical distribution is very irregular. We used daily returns because it gives us an
advantage. If we use calendar year returns then there would be fewer than 50. Remember, the free-floating FX
market was born in the 1970s. If we superimpose the returns – calculating, for example, returns for the periods
January 1, 2010–January 1, 2011; January 2, 2010–January 2, 2011; January 3, 2010–January 3, 2011; and so on –
then we get a lot of returns, but they are linked to each other (the technical term is “autocorrelated”), meaning that if
the first return is 15% then the second return is likely to also be in the neighbourhood of 15%, and that is a problem.
Taking daily returns has its advantages.

Once the histogram in Figure 8.7 is calculated, we can also calculate the average return,4 and the standard
deviation. For assets with strong mean-reversion the average returns should be close to zero. For assets that tend to
exhibit long lasting trends the average return might substantially deviate from zero, depending on the historical
period under consideration. Our assumption here will be that the distribution of the future EURUSD exchange rate
return will be Gaussian (ie, a normal distribution) with the same standard deviation as the one we just calculated in
the histogram. In this idealised world, also assume that the mean of the distribution of future returns is zero.
Signifying that the expected return with respect to the forward rate is zero. All this represents very reasonable
assumptions that will help tremendously in what follows when calculating the option value. Keep in mind that the
Gaussian is the distribution of the returns on the exchange rate, not of the exchange rates. This will be a source of
complication in the future, but it is the price you pay.
Before we get too comfortable using history to predict the future, let’s expand on a philosophical constraint that
applies to finance. History is a good place to start to get a sense of how much EURUSD can move over a year. We
can even construct the histogram in Figure 8.7. It clearly tells us that such a histogram represents the probability
distribution of returns over a given period. However, when turning to the future, it’s a guess as to the probability
distribution of just one return: the return from today to the expiry date. We can never know if the probability
distribution that we attributed to the future event was correct. How so, you ask, after all the future eventually reveals
itself and therefore we will know if we were right. Not exactly. Suppose the distribution of returns we project has a
mean of zero and a standard deviation of 10%. In a year’s time, the actual return is 12%. Does that tell us anything
about whether or not our assumed distribution was the correct one? No, it does not. A physicist can run an
experiment a million times under exactly the same conditions and construct a histogram similar to Figure 8.7, and
then compare the 0 mean to the histogram. The closer those two are, the more correct the assessment.
In finance, we cannot run such an experiment. Naïvely, you would record the yearly return of EURUSD over
many different periods and again build a histogram just like the one in Figure 8.7. Actually, that would be the wrong
approach as every time the option market prices an option it must take into account the current state of the whole
financial system. Therefore, a prediction of the yearly return starting on January 1, 2000, will be intrinsically
different from a prediction of the yearly return starting January 1, 2010. Consider the dramatic example of GBPUSD
(UK sterling, also called cable, as mentioned earlier). A prediction of the monthly return on cable starting on June 1,
2016, will be intrinsically different from a prediction of the monthly return on cable starting June 1, 2006. For one
thing, the former prediction has to include the Brexit referendum on whether the UK would leave the European
Union (June 23, 2016). The point is that the historical yearly return from June 1, 2010, is not going to help you
estimate the yearly return from June 1, 2016, as the situation of the financial system was completely different.
Here’s a more illustrative example. Suppose that to price an option you have to estimate the monthly return
distribution of EURUSD starting on November 1, 2016. In essence, you have to estimate the probability that Hillary
Clinton or Donald Trump would become the president of the US on November 8, 2016. Are you going to use the
return on EURUSD corresponding to all the past presidential elections? There is no way that would be helpful, given
that there are so few data points and that each US presidential election has been completely idiosyncratic. In other
words, we don’t have the luxury of the physicist and cannot run experiments under identical conditions. Every
situation and moment is in some way unique.
What then is the meaning of probability distribution in this example? In a physicist’s set-up, with the luxury of
being able to completely reproduce the conditions of the experiment, the probability distribution is just the
histogram of all the experimental results put together. Not so for the probability distribution about events that cannot
be repeated. In the example of the US election, confidence in a prediction based on the probability distribution takes
on a completely different meaning. In other words, if the probability distribution for the US election says that there
is a 40% chance that Hillary Clinton will be elected, this means that you are 40% confident that she will win the
election and 60% confident that Donald Trump will win the election. Nobody in the world will ever know what true
physicist’s probabilities were for that election. Just the concept of using a probability distribution implies the
assumption that the US election was an event driven by randomness, which is debatable. To summarise, we have
seen that probability can signify two different things. For a physicist running the same experiment probability is the
number of times the same experiment will have a particular result. For the trader that is trying to predict the future
probability means the degree of confidence that a particular event will happen.

Probability can only be understood in ranges

History gave us tepid assistance. However, that remains all we’ve got at this point. With the limitations of historical
distributions firmly understood, let’s use them in a cautious way to calculate the probability of that certain rate
occurring in the future. Theory says the mean must equal the forward rate of EURUSD, with a value date equal to
the settlement date5 of the option that is being priced. This is because it does not cost anything to enter into a
forward contract, and therefore to maintain mathematical consistency the forward rate must take the role of the
expected value of EURUSD in a year’s time.6
Now for a surprising fact. The probability that EURUSD will be exactly equal to the forward rate in a year’s time
is mathematically zero. To convince yourself that for any specific number the probability of that rate occurring is
zero, think of the following. There is an infinite set of numbers that the rate could equal, so if for each number there
were a finite probability, no matter how small, that their sum would greatly exceed 100% – it would actually be
infinite. We should reason that there exists a positive number “a” such that the set of exchange rates which have
probability greater than “a” of happening is infinite. This would imply that the sum of these probabilities is infinite,
not 100%. This reasoning is very bad maths, but it serves the purpose of intuiting why one number, one probability,
does not make sense. Stated differently, the probability of EURUSD being inside a specific range in a year’s time is
greater than zero only if such a range is wider than a single number.
 Therefore, the probability density function (ie, the Gaussian curve) does not show the probability that EURUSD
will equal any specific rate shown on the x-axis of the graph. To get probability information from that curve, you
have to consider intervals, and the probability for EURUSD to be within such and such an interval will be the area
limited by the x-axis, the curve and the boundaries of the interval. For example, the probability that the rate will be
between X and X+1σ is 34.1% from our knowledge of statistics – see Figure 8.8.
The entire area between the curve and the x-axis is equal to 1.00, or 100%. The area delimited by the Gaussian
curve and the x-axis between μ and 1σ is equal to 0.341, or 34.1%. The area to the right of 1σ, above the x-axis and
below the Gaussian curve, is therefore equal to 15.9%. Also note that to the right the curve extends to infinity
without ever touching the x-axis, and the same to the left. Still, the total of the area below the curve and above the x-
axis is finite and equals 1.00, or 100%. This latter property will not be true for lognormal distributions, see following
pages.

Exercise 8.1: How do we calculate the area to the right of 1σ?

Solution 8.1: The graph is symmetric, so left of μ is 50%, and then we take away the 34.1% the result is
15.9%.

Given our theoretical understanding of the curve, let’s apply it to the example of the EURUSD distribution with a
time horizon of one year. We know that the forward EURUSD rate is equal to μ, and assume for the time being that
the volatility (or standard deviation) used to price the option (the implied volatility) is equal to 10%, so σ = 10%, or
σ = 0.10. In this example, we simply decided to start with the only thing we have, the historical σ, and see where it
leads.
Using all the facts known about the Gaussian curve, the probability for EURUSD to be between 1.03 and 1.13
(which is approximately a 10% move, a move equal to σ) in a year’s time is 34.1%. Perhaps more to the point, we
can say that the probability for EURUSD to be above 1.13 in a year’s time is 15.9%, or that the probability for
EURUSD to be above 1.06 in a year’s time is 40%. Keep in mind the 40% is made up to provide us with an
example. But wait, this is exactly the probability that our 1.06 euro call/US dollar put will be exercised. This is one
of the numbers we need to price the option.
This is great news. By analysing past EURUSD moves (and assuming the calculation of the volatility returned
10%) we have managed to build the corresponding Gaussian curve. This is comparable to rolling the dice 1,000
times and extrapolating from past experiences to predict the future. The beauty of the situation is that Gaussian
curves are very well understood. Again, there are some crucial assumptions that we take on faith, such as our
assumption that the distribution of the future EURUSD would be the same as the distribution of the past EURUSD.
Therefore, each time a trader must price an option, they will have to estimate how closely the past will repeat itself
and will actually adjust the distribution (and therefore the price) to take into account their sense of how the future
will unfold. For example, the past history used to calculate the distribution may include events that are unlikely to be
repeated during the life of the option. Think of a political or actual earthquake, an event that happens infrequently
such as an election (eg, the Brexit and US presidential election mentioned above), or a fiscal or monetary policy that
may be starting or ending. This analysis has practical implications that traders must consider before quoting a price.

Figure 8.9 is from the Bloomberg page FXFM. It shows on the left-hand side a list of scenarios. For example,
there is a 73.2% probability that EURUSD will be inside the 0.9685–1.1972 range in a year’s time. On the right-
hand side is a probability density curve (or, more exactly, a probability density distribution) derived from the prices
of EURUSD options with a one-year expiry. You can see that such a curve is not as regular as the mathematical
Gaussian curve. You may have guessed this, or wondered how it is that –1σ, which is 34.1% of the area under curve,
plus +1σ, another 34.1%, should be 68.2%. Why is the Bloomberg page saying it’s 73.2%? In fact, the market is
saying that the probability density derived from option prices is not perfectly normal (not perfectly Gaussian), and
the market as a whole believes there is a greater probability that we will be in the range 0.9685–1.1972, than if the
probability density of EURUSD were a perfect Gaussian distribution.
This is what happens in real life. Traders do what we did here – they take a look at history and they take a look at
the maths and then amend the distribution, which amends the price, to reflect their perspective. In Figure 8.9, we
snuck one past you. The resulting probability density distribution is not just one trader’s perspective, but rather the
collective wisdom of the entire market! It is generated from a complete laying out of every possible strike for every
possible date (within reason) for that currency pair. Pretty cool stuff.
In pricing options, the standard deviation of the probability distribution of returns on the exchange rate is also
known as volatility. As we saw above, we used history and then the subjective evaluation of how that should be
adjusted for the future. Therefore, there are two types of volatility. One type (historical, or realised volatility) is
calculated from past moves of the exchange rate and, given its origin, this number talks to us about the past. The
second type of volatility (implied volatility) is based on future expectations, and is in fact the one used to price
options. When a trader makes a price on an FX option, they must first think of how volatile the exchange rate will be
in the future. Of course, nobody knows that, but the trader has to have an opinion. That opinion is called implied
volatility, and it is used inside the Black–Scholes formula to calculate the premium of the option. On the other hand,
if I just know the price (premium) of an option that just traded in the market, with that information I can reverse-
engineer the formula and deduce what implied volatility must have been used. This is easy to do. Try a volatility,
compute the price of the option and if my price is too high I then try again with a lower volatility and so on and so
forth. This is the brute force approach.
The prices of options that are quoted today in the option market reflect adjustments made to include real-world
expectations. So, there is a lot of art along with the science. To give you one glaring example, ATM one-year
options on GBPUSD were quoted with a volatility of 10.03% on June 23, 2016, the day before the Brexit
referendum. The historic volatility for the last year for GBPUSD was only 8.4%. After the results of Brexit were
announced, and the UK was destined to exit the European Union, option traders stopped caring so much about the
past and realised that the future will be more volatile – or at the very least different than the past. Therefore, past
distributions were of questionable use. The market, which is the sum of all prices shown by all traders, jumped from
10.03% to 12.55%.

Figure 8.10 (from the Bloomberg page VOLC) shows two kinds of volatilities. The white line is the volatility
measurement called implied volatility. This is the volatility that is derived (implied) from existing option prices as
they reflect future market expectations. Line ① is historical (or realised) volatility. This is a reflection of what has
actually happened in the past. Clearly, market-makers have ignored history from November 2015 on, basically
saying “history is not a good guide for the craziness that is coming”. More on volatility later.

The real distribution of exchange rates revealed

The previous discussion has not been as mathematically rigorous as the subject deserves. It was important to discuss
concepts and provide a palpable sense of what takes place in the market – but it is no longer the path of least
resistance; it is now time to climb to higher mathematical ground.
We have a confession. We told you that the yearly returns, or the daily returns for that matter, will have a
probability distribution function that is shaped like a Gaussian curve. This is actually a fair assumption. Then we
showed this Gaussian curve with the possible exchange rates on the x-axis. Herein lies our sin. In fact, if returns are
shaped like a Gaussian curve, then the exchange rates probability distribution will not be shaped like a Gaussian at
all.
This is the only way things can be. Think of it this way – there is a small chance that the return will be a very
large negative number, and this is fine. However, the exchange rate will never be a negative number because a euro
will always be an asset and never a liability, so the exchange rate will never go negative. Nonetheless, all Gaussian
curves will go negative at some point. Therefore, the exchange rate probability distribution cannot be a Gaussian.
This is not really a huge problem, it only means that we need to reserve Gaussians for returns and to now fess up
with what the real distribution of the exchange rate looks like. It is called lognormal instead of Gaussian or normal.
As a matter of fact, we kind of had this backward from the start – we never really explained why we thought FX
returns had a normal distribution. You believed us because we showed you the central limit theorem. We could have
started by observing that FX rates had a lognormal distribution and then it can be mathematically proven (not by us
here) that if the rates are lognormal, the returns of such a series of numbers are normally distributed.
First of all, see the equation below for the probability distribution of the returns:

The way you want to read this formula is: “the probability density of the return x is equal to all that mathematical
mumbo jumbo” (“mumbo jumbo” being a well-understood mathematical construct). Note the μ for the mean of the
distribution and the σ for the standard deviation of the distribution. As mentioned, as x varies, the equation above
will draw the Gaussian bell curve. Also note that for x very large and negative ρ(x) will tend to zero, and for x very
large and positive ρ(x) will tend to zero as well. For x equal to μ, then ρ(x) will equal

The mean (the average, or the centre of gravity of the bell shape), the median (the value such that 50% of the returns
are higher and 50% of the returns are lower) and the mode (the highest point of the curve) are all equal to μ, which
can be calculated as (rd – rf) – σ2/2 where rd is the domestic interest rate, rf is the foreign currency interest rate and σ
is the volatility.. This return is linked to the interest rate differential between the two currencies. Very simple and
logical. Note that the mystical appearance of π in the formula is indeed a little piece of magic for those seeing this
for the first time.
 Figure 8.11 shows the probability density for the exchange rate return in case the volatility is 35% and the
expected return is zero. The figure does not depend on the starting exchange rate. Also note that 35% = 0.35 is
where the curve flips from convexity facing upwards to convexity facing downwards, and analogously for –0.35.
Also note that the peak of the curve is between 1 and 1.25. As we noted before this number has virtually no
meaning: it is not the probability of the return being equal to μ.
The probability distribution for the exchange rate is then a consequence of the formula for the probability
distribution for the return on the exchange rate. Let us call y the exchange rate, and μ again the expectation of the
exchange rate return (usually assumed to be zero, but in reality equal to (rd – rf) – σ2/2). The following is the
equation that describes the probability distribution for the exchange rate:

Figure 8.12 shows the lognormal probability density for the exchange rate in case the starting exchange rate is 1.00
and the volatility is 35%. This level of (annualised) volatility would be too high for most FX markets, but it makes
the figure more readable and also makes it more obvious that the curve is not symmetric.
In the case of this curve, the mean is equal to

where S0 is the starting spot exchange rate level. Given the definition of μ, the exponent is equal to (rd – rf) – σ2/2 +
σ2/2 = (rd – rf). Therefore, the mean is exactly equal to the forward rate. Again, this has to be this way because a
forward contract is zero cost and so it must have a P&L expectation of zero. The median is S0 · exp(μ), which is easy
enough, and the mode is

Those are just well-known facts about normal and lognormal distributions that come from statistics. We are just
using them to investigate the price of FX options.
Let’s go back to our perfect return distribution where we have a Gaussian with σ = 10%. What if you don’t have a
strong opinion on the future, and you wish the past to be the guide to your Gaussian curve? As we have seen, one
approach is to just assume that in the next second EURUSD can move up by 0.0001 or down by 0.0001. But, how
can we estimate the all-important 0.0001 number? How do we know it is not 0.0002, for example?
Probability theory comes to our assistance here, giving us an explicit recipe to calculate σ, which is the volatility,
given the past history of spot rates. Specifically, we will use the daily fixings for EURUSD, going back a sufficient
amount of time, that we estimate is enough to describe the behaviour of the currency pair. As we stressed before,
your choice of this time interval in the past is very important and will impact the option price greatly. We are
assuming that the past probability distribution of returns is equal to the probability distribution of future returns.
Remember that what investors really care about is returns, not the absolute levels. Therefore, if we take the daily
fixings we can turn them into returns using the following formula:

Similarly, we would prefer our Gaussian curve to represent returns on the x-axis and not the absolute level of the
EURUSD exchange rate in a year. Once we have the daily returns, the standard deviation of this series of returns ri’s
is calculated with the following formula (R is the average of all the daily returns calculated above):
Here, the number 255 is there to remind us that our calculation of volatility is over a year (it is called “annualised
volatility”), whereas our daily returns are over a day, so we need to rescale by the number of business days in a year.
Which means we assume Saturdays, Sundays and holidays have no volatility; a big assumption.

CIRCLING IN ON THE FAMOUS FORMULA

Armed with all this new information, let’s return to our original quest. Remember we were trying to understand the
probability that spot will end above 1.06. See the option structure in Figure 8.13 to refresh your memory.
Place the option strike, 1.06, on the x-axis in the lognormal curve chart, after which we will be able to calculate
the probability that the rate will be higher than the strike at expiry date. By the way, here’s an interesting by-product:
if we know the probability of the rate being higher, we also know the probability that the option will be exercised. In
our case, σ will be close to 10%, so X + 1σ is roughly 1.13, and 1.06, our strike, will be between X and X + 1σ.
Once 1.06 is placed on the x-axis of the lognormal curve, we can then calculate the probability that EURUSD will
be above 1.06 in one year’s time as the area above the x-axis, below the lognormal curve, and to the right of 1.06.
Let’s call this all-important number, the probability of being above 1.06, call it 1 – N(1.06).7 Now to answer
questions such as “what does it look like?”, “how do you compute it?” and “is it a formula?”. Here is the surprise –
it cannot be written as a classic formula, composed of √, ∑ or any other fancy mathematical symbol. As it represents
an area, it can be written as:

where p(x) describes the Gaussian curve; oops, we mean the lognormal curve. It was proven many years ago that
there is no simple formula and no simpler form to compute it. The integral symbol (∫) means to measure the area,
which is synonymous with probability. Keep in mind the entire area below the Gaussian curve measures 100%. So if
we want to calculate this number we have to use numerical methods. There is nothing very difficult with that,
computers are very good at calculating this number to any desired level of precision.
 Exercise 8.2: What is 1– N(1.03)?

Solution 8.2: 50%, as 1.03 is the centre of the distribution, so half the area is to its left and half to its right.

Exercise 8.3: What is 1 – N(0.93)?

Solution 8.3: We know that σ is about 10%, so about 0.10. Therefore, 0.93 is about X – σ, so the probability
to be above that is about 50% + 34.1% = 84.1%. Note that, appropriately, this number is higher than 50%.

The next thing we want to do is to rescale the Gaussian curve to the case where σ = 1. This is done so that
standard calculations on the probabilities can be carried out on a very standard curve. The calculations and the
formulas are just easier if you assume σ = 1, and in such a particular case we will write to specifically indicate that σ
= 1. The only thing that really needs rescaling is the 1.06. In our standard normal distribution X = 0 and σ = 1, so
1.06 will more or less lie at the point 0 + (1.06 – 1.03)/σ.
This is not exactly right, because then a return of 1% (expressed as 1.03 × exp(1%)) and a loss of 1% (expressed
as 1.03 × exp(–1%)) would not be the same distance from X. They would be at the same distance if we used a
different formula to compute returns, getting rid of the exponential returns. For example using 1.03 × 0.99 and 1.03
× 1.01 would give you two returns which are equidistant from the starting point. Therefore, we need to correct the
formula to this one: 0 + log(1.06/1.03)/σ. So we are trying to calculate 1 – N(log(1.06/1.03)/σ) = N(log(1.03/1.06)/
σ). You can convince yourself of the equality using the properties of the logarithm and of N. This is still a bit too
simplistic, since the market-implied expectation of EURUSD in one year is not 1.03 but rather should be the forward
rate, S × exp(r × t) – remember that we assumed the interest rate in euros to be equal to zero. Here r is the annualised
US interest rate and t is the time to maturity expressed in years. So our formula replaces 1.06, again taking the
logarithm, becomes:

Here, S stands for 1.03, the spot rate, and K stands for 1.06, the strike rate. Our next step is to account for a very
obvious fact in life, a very unfortunate fact, which is called the “volatility drag”. If you invest in an asset and the
price goes up 10% and then down 10%, or down 10% and then up 10% you might expect to be back at the starting
point. But that is not the case. In both cases you end up with less money than you started with. This seems unfair but
needs to be accounted for, so we will modify our formula so as to make it a little bit harder for the exchange rate to
reach 1.06. This is precisely why the definition of μ includes the subtraction of the volatility term, see above.

One problem is left – the probability that EURUSD will be above 1.06 at expiry increases as the tenor (the amount
of time between the trade date and expiry date) increases. One way to intuit such things is to go to extremes. If the
expiry is in three minutes and spot is now 1.03, then you can say that the probability that EURUSD ends up above
1.06 in three minutes is very tiny. If the tenor is 10 years, then the probability that EURUSD ends up above 1.06 in
10 years’ time is about 50%. In fact, you see that our formula depends on time; with greater time the probability
increases.
This turns out to be good, but not exact, because we are forgetting that the exchange rate does not just march
upwards every day and does not march downwards every day. In most situations, you will have some up days and
some down days, so there are quite a number of scenarios where the exchange rate ends up somewhere in the middle
of your distribution of expectations. With our current dependency on time, if in one second the rate can move up or
down by 0.0001, then in two seconds it can move up 0.0002 or down 0.0002. In reality, we know that in two
seconds it can move up 0.0002, or down 0.0002, or return to the starting point! To account for that, we need to
adjust the dependency on time in the following way:

Now, we can calculate precisely the probability that EURUSD ends up above 1.06 by substituting 1.06 with a new
number, d2. The probability then becomes the area under the curve and to the left of:

That number is indicated as N(d2). In our example, the d2 number will be about 0.3.
This brings us close to discovering the final official version of the formula. Stay strong. The second quantity that
is missing is the {Expected EURUSD level, under the assumption EURUSD > 1.06}
That number is the size of the area under the curve and to the left of:

and multiplied by the forward rate. The official notation for that number is N(d1), and after we multiply by the
forward rate it will become Fwd × N(d1). Here’s an important nuance: N(d1) is the expectation of the level of
EURUSD rate, subject to the assumption that EURUSD ends up higher than 1.06, multiplied by the probability that
ends up higher than 1.06. It is not the probability of the EURUSD to be higher than 1.06 (which is N(d2)), and it is
not just the expectation of EURUSD (which is the forward rate). You can think of it as a conditional expectation –
the expectation under the condition that EURUSD will be higher than 1.06.8
Now, putting back all the pieces together, we can write the price, P, of the option as:

And this is the Black–Scholes formula … Sometimes it’s that easy to win a Nobel prize!
Some of you may suspect foul play. OK, we made assumptions. This is not the full Black–Scholes formula. We
cut a couple of corners. The formula above is the expected value as per the expiry date, not as of today. To correct,
we need to present value the price. Assume r is the interest rate in the US. Use continuous compounding over a time
horizon t (which is one year in our example) to arrive at:

Here, Fwd is the forward rate and rd and rf are, respectively, the domestic (US) and foreign (euro) interest rates.
Actually, we are not using the foreign interest rate here, as it is embedded into the forward calculation. We can make
it explicit, because Fwd = Spot · e(rd −rf)·t and then you get another way to express the same formula:

Finally, we have arrived at the Black–Scholes option pricing formula. For completeness, you might want to know
the Black-Scholes formula for put options. You can just repeat the whole argument again, and you would arrive at
the Black–Scholes formula for put options:

Here’s an odd question. Is this a difficult or an easy formula? The answer is that it is an easy formula, because it
only depends on the relationship between the strike (K) and the implied volatility (σ). The process to a price is the
following. We know all there is to know about a normal distribution with mean zero and standard deviation equal to
σ. The only thing to do is to calculate the two following numbers.

How many standard deviations does the exchange rate have to travel to hit the strike K (this represents the cost of
paying the strike)? That number is roughly expressed by d2.
How many standard deviations is the exchange rate expected to have travelled under the hypothesis that we exercise
 the option? Here’s a fun project to do to impress your friends – that number can also be found by cutting a piece
of wood in the shape of the part of the probability distribution to the right of the strike, and balancing it on a
peg. The point where the peg touches the piece of plywood is d1.

Figure 8.15 shows the piece of the lognormal distribution that lies to the right of the strike. The peg (the arrow) is
placed such that the piece of the distribution, if imagined made of plywood, would exactly balance on the arrow (the
peg) so that it would not tilt to the right or to the left. The point where the peg touches the plywood is d1.

Then everything is easy as we know everything about the probability distribution, and so can easily calculate the
areas of the different pieces of plywood using our N(·) function.
One last remark. The way we have explained the Black–Scholes formula makes a lot of sense and is very
intuitive, but we have not demonstrated that it is correct. In other words, the fact that our explanation makes sense is
no substitute for a formal mathematical proof.9 The proof was derived by Black and Scholes, and, as we will see in
the following pages, is based on showing that the owner of the option can make sure of extracting from the option a
profit equal to the price paid. This basically says that the owner of the option can make sure to break even, if they
have paid the Black–Scholes price for the option and the volatility assumed actually materialises.
There is another funny geometrical way to compute the price of an option, if you like to play with plywood. In
Figure 8.16 you can see how it works.
The solid line is the lognormal probability distribution. The double-dashed line is the payout of the call option:
zero to the left of the strike, and then steadily climbing at an angle of 45 degrees to the right of the strike. The dotted
line is the product of the probability distribution by the payout of the option. The price of the call option is the area
between the dotted line and the x-axis in the chart. The idea is that you multiply each possible future option payout
level by its own probability of happening. Just a fun Sunday afternoon project for the kids.

VOLATILITY BRIEFLY EXPLAINED

Now that we have the formula, we can direct our attention to one of the ingredients of the formula that we have not
really explained, σ, or volatility and its progeny, variance. σ appears on most of the formulas above and in the chart
of the Gaussian curve. It is a key parameter, denoted by the Greek letter sigma, and is called “volatility”. In finance,
volatility is the standard deviation of returns. Variance is volatility squared, or σ2. As we said, at the centre of the
curve, the curve has its convexity pointing down. At the wings of the curve (extreme left and extreme right), the
convexity is pointing up. The exact point where the convexity transitions from pointing up to pointing down is
marked on the chart with 1σ (it should be X + 1σ really, but in the chart it only says 1σ). The distance between X
and X + 1σ is obviously equal to σ, and it is sometimes erroneously referred to as “one vega”. As we saw in the
formulas in the previous section, the value of σ is very important because it will determine the value of the option.
Let’s return to where we left off before all the mathematics. How does one calculate what the value of σ should
be? Recall the example of EURUSD moving up or down 0.0001 each second. The quantity of 0.0001 is arbitrary.
You may not have realised it then, but that was a big and presumptive step for us to take. It turns out that value is
very important, and in fact determines the shape of the curve. Had the example used been 0.0002, the central limit
theorem will give us a very different bell curve for EURUSD – specifically, one that is more spread out towards the
left and the right, and whose peak is lower (the actual number is more like 0.000021). Using 0.0002 implies that it is
more difficult to predict where EURUSD will be in one year’s time. Go to extremes to gain intuition. If that interval,
and consequently volatility, gets really close to zero, then EURUSD does not move at all. Then, it is very easy to
predict where it will end up in one year. It will end up at the forward rate, not the current spot. Therefore, that choice
of 0.0001 generated a particular bell curve and, as a consequence, a particular point, where the convexity changes
and where 1σ is.

Realised volatility
Got it? If we know how much the exchange rate moves per second, we can calculate σ, the volatility, and once we
know σ we then know the shape of the probability distribution. The probability distribution, in turn, tells us how
much an option is worth – and then we know if we should buy it. So how to arrive at that number? Let’s start very
practically. Look at how much the EURUSD rate has moved in the last second, and assume that in the next second
the move will be similar. Brilliant, but just consider that the last second might not be enough to give us a true
picture. After all, it may be Saturday afternoon and the FX markets are closed, or it may be the second after Brexit is
announced and it’s crazy in the markets. One second is simply not enough data to rely on. In fact, traders look back
over a stretch of time that they think makes sense depending on the task at hand. Very often, when pricing an option,
traders will look back at a stretch of time equal to the tenor of the option they are pricing. One can try to take an
average of the amount that the EURUSD moves each second, say over the last year, but the actual calculation is
expressed by the following formula:

where N is the number of seconds in a year. Some of you of a certain age are thinking: How do you measure a year?
In daylight – in sunsets, in midnights? In cups of coffee perhaps? Oh, I know: it’s 525,600 minutes, times 60 of
course, since there are 60 seconds per minute. You’d be wrong. Instead, we are interested in the number of seconds
during which the EURUSD market is actively trading in a year. x– is the average return from one second to the next,
and xi is the return on EURUSD from second number i to second number i+1.
For options whose strike is the forward rate (ATMF options), doubling the volatility roughly doubles the premium
– or, in other words, the premium is roughly directly proportional to the volatility for ATMF options. Volatility
effects out-of-the-money (OTM) or in-the-money (ITM) options very differently from the ATMF options. More on
that in the coming pages (get excited!).
The volatility calculations described above give “realised volatility”, or “historical volatility”. This is obvious,
right, as the calculations are based on the past behaviour of EURUSD, its historical movements and what was
actually realised.

Implied volatility
What happens if the trader is expecting next year to be much more volatile? We’re back in the same mess, having to
guess what the interval of movement is. Granted we have one more piece of information, history, but you know that
historical performance is not necessarily an indicator of future performance (as discussed earlier in this chapter).
Let’s build on what we have. If the expectation is for more uncertainty, then the volatility number to be used when
pricing an option should definitely be higher than the realised number. We need to use a forward-looking estimate of
future volatility to price options. This forward-looking estimate is called implied volatility. Implied volatility doesn’t
necessarily refer to one’s sense of what they imply; it refers to the volatility number we reverse-engineer from
observable prices. Let us explain this concept in detail.
Note that, in the Black–Scholes formula, all the inputs except one are known: notional, strike, tenor and interest
rates. The only missing input is the implied volatility. This means that, after observing an option price on an
exchange or if getting a price from someone in the market, one can easily back out what implied volatility is. It’s the
only missing number. Therefore, if there is an active market for options and prices are observable, then the implied
volatility is known for those specific options, but also for all other options because we can create a surface from the
information we have. OK, we’re now getting way out in front of our skis …
Suppose that one year € calls with a strike of 1.06 have been quoted a 100 times today in the market and we
know, from the quoted prices, that the implied volatility is 10%. How should a trader at a bank respond if a client
asks for the price of a three-month 1.07 euro call? They can use all the same inputs in the formula, including the
same volatility of 10%, and just change the strike from 1.06 to 1.07 and quote the client, right?
Not right. There is an important difference. Knowing the price of a one-year, 1.06 euro call does not allow one to
calculate the price of a three-month 1.07 euro call (see Figure 8.17).
By now it is also hopefully clear that volatility levels are interchangeable with premiums when describing the
price of an option. Remember, we know all the other option parameters. As a matter of fact, we can ignore the other
factors when making the price and focus on what is important: volatility. Think of it like this. When US Treasury
bonds trade, the quote is in terms of the yield not the price. It makes all the calculations much easier than quoting
bonds in price, even though at the end of the day if you sell a bond you will receive back a dollar amount and not a
yield. The same thing applies to options where you quote the volatility and then you fill in all the other details such
as the calculation of the forward rate.

There’s a lot to be said about implied versus historical volatility, and the dance they engage in around each other
and across different terms. We will discuss this more fully when we talk about how volatility traders can use these
differences to make money. OK, let’s take a breather from the option premium analysis and check on our progress.

IS THE BLACK–SCHOLES FORMULA CORRECT?

Congratulations – we’ve gone on a great adventure and discovered the theory and a formula to price options. But
how do we know all this is correct? Apart from our abstract argument to derive the formula, which we just went
through, is there another way to confirm the formula? Spoiler alert: we know the formula is not totally correct, but
we all agree to use it so it becomes the standard.
Ideally, we want to devise an experiment to verify the formula. Here’s a straightforward thing to do. Go back in
time and pretend that we bought all sort of options at a price calculated using the formula. Then fast-forward the
time series and calculate how much all these options would have been worth at expiry. If the formula is correct, then
on average all these trades will break even. I bet you thought we were going to say that: “if the formula is correct,
the results would prove buying options made money”. Not so. This is a very important concept.10
As a speculator, if we agree on all the parameters of an option and believe the formula correctly prices the option
then there is no reason to buy or sell an option. The premium will get you to a breakeven only. No need to spin your
wheels. So, then why would a speculator buy or sell an option? Because two people can look at the same world with
all the same information and make wildly different conclusions about where we are going from here. This is where
the art versus the science comes in. As a hedger, on the other hand, there are very different motivations. Hedgers
don’t necessarily care where spot will be at expiration. They just want that insurance against some other objective.
Maybe they are concerned about the total revenue in US dollars from selling their widgets in Malaysia. If that
revenue estimate turns out to be realised or exceeded, they don’t really care about where USDMYR is in a year
(more on this when we talk about corporations).
Back to figuring out if the formula is correct. Can we carry out the experiment and check all the past option
prices? No, we can’t. The problem is that purchasing all the options at a price equal to the price given by the Black–
Scholes formula raises the thorny problem of what implied volatility to use in the formula in the past. We know
what the volatility actually was for every option (realised volatility), but we discovered that after the option expired.
It would be great to go back to calendar year 2003, imagine buying an option at the realised volatility during
calendar year 2003 and then check how much money you would have made, or lost. But that experiment does not
make much sense because at the beginning of 2003 you would not have known the realised volatility during 2003.
You would have known the realised volatility during 2002, but you would have to wait one year before the realised
2003 volatility would be available to you. Therefore, the formula cannot be checked this way. One other thing we
could do is to use in the formula the implied volatility prevailing in the market at the time that we are pretending to
have bought the option. This makes a lot of sense because then the price that we calculate matches the market price
that you would have paid for the option at that time in the past. Suppose that we now run such a lengthy number of
calculations and the result is that, on average, we would have made a lot of money by purchasing all these options.
This only proves that in the past the market has underpriced options – ie, that the implied volatility quoted in the
market in the past was too low and that, in aggregate, the market expectation about future volatility of the EURUSD
exchange rate undershot. So, again it is no proof that the formula is correct.

In fact, many people take this backtesting approach to see if it makes sense to enter into a strategy. Figure 8.18
shows a historical study where a one-year ATM euro call/US dollar put option has been hypothetically bought every
month starting on September 4, 2007, and ending on December 1, 2015. The white line on the top panel shows the
aggregate P&L of all these option trades. In the lower panel, the white line is the spot rate and the lines denoted by
① show the strikes of the different options bought. Each line ① starts at its own trade date and terminates at its own
expiry date. Comparing the line ① with the spot line shows that not many of those options have been exercised.
They were exercised only in periods when the line in the top panel rises. This is because the euro was going down
during the years we chose to examine. Of course, if the euro was rising in a consistent manner many of the options
would have been exercised. Therefore, you can see how quickly an investigation about past implied volatility levels
turns into a realisation of past spot level trends. In other words, we could argue that the result of the backtesting
exercise leads us to believe options make you lose money. However, the result was driven by the falling trend in the
exchange rate during those years. That is hardly an argument for implied volatility to have been overpriced or
underpriced.
There is one more argument to be tried. Assume that the implied volatility at the time the option is bought is equal
to what will turn out to be the realised volatility during the life of the option. This last number cannot be calculated
at the time of purchasing the option, but it can be calculated at the time the option expires. Also assume that we
know the volatility that the market will exhibit during the life of the option (realised volatility), and not just the
average volatility but the instantaneous volatility at every instant of the life of the option. Then there is a recipe that
can be used, a mathematical process, to determine a sequence of forward trades. This sequence of trades, together
with the option, will generate a P&L that will be mathematically certain to equal the price of the option as calculated
using the average of the realised volatility during the life of the option.
This will finally confirm that the Black–Scholes formula is correct. The reason is that a person who bought the
option would be able to extract from such an option a P&L equal to the price paid for the option, thus proving that
the option was truly worth that price. We are assuming that the implied volatility used to price the option equals the
average future realised volatility.

Delta dawning
So what is this recipe? It is actually very simple. In fact, we will dive deep into some complex trading concepts, like
the first derivative of the option price. But do not panic, we will come back up for air and dive more slowly into all
the partial derivatives that matter.
First note that to work perfectly you would have to have perfect knowledge of the instantaneous volatility of the
exchange rate at every instant of the life of the option. Assume we do. Suppose the notional on the option is €10
million and that the strike is 1.06. Now, recall in the Black–Scholes formula, N(d2). This number is called the delta
of the option. Also recall that it represents the probability that the option will expire in the money. Delta also has
another meaning. If you buy EURUSD via a forward contract (with a value date equal to the settlement date of the
option) and with a notional equal to €10 million × N(d1), then something interesting is true of a portfolio composed
of the option and the forward contract. In an infinitesimal time interval, the value of such a portfolio does not change
if the EURUSD exchange rate changes.11 You might think this is impossible, because both the option value and the
value of the forward contract will change. You’re right, they both change but in opposite directions and by equal
amounts.
Assume a bank trader sold the option to a client. The line ① in Figure 8.19 depicts the changing value of the delta
as a percentage of the notional today. As EURUSD rises immediately, the option will rise in value as EURUSD will
get closer to 1.06, and this will increase the probability that the option will be exercised. This will have a negative
impact on the trader’s portfolio because they are short (sold) the option to the client. However, as EURUSD rises
our long position in the forward contract (having a forward of long euros/short US dollars) will become an asset for
them, and therefore its impact on the portfolio value will be positive. Miraculously, those two changes in the value
of the option and the forward will exactly offset each other, and the portfolio value will not change. The same
reasoning will apply in case EURUSD falls.
Figure 8.20 shows an option trade, leg 1, where the delta can be read on the third-last row called Delta. Leg 2
represents the hedging forward trade, where the trader buys euros in a notional equal to the delta of leg 1. The
resulting aggregate delta can be read in the first column from the left, equal to zero.
Here is the dramatic turn of events. As soon as EURUSD has climbed by an infinitesimal amount, the calculation
of N(d1) will return a different number, so what was true then will not be true now, and the portfolio will very soon
start drifting in value, meaning that the two instruments will no longer move in precisely opposite directions. The
only way to keep the portfolio value unchanged is to continuously compute the delta of the option and to make sure,
by selling or buying EURUSD forwards, that the aggregate notional position in EURUSD forwards in the portfolio
always equals €10 million × N(d1). This process, called “delta hedging”, is central to proving that the Black–Scholes
formula is correct. In addition, the argument is at the core of the job of any market-making trader. Keep in mind that
when people say “Delta”, they may mean the probability that the option ends up in the money, which is N(d2), or the
hedging ratio, which is N(d1). The N(d2) of an ATMF call option is not 50%. We knew that from the exercise above.
The N(d2) is smaller than 50%, because the probability to exercise an ATMF call option is less than 50%. This is
because the mean of the lognormal distribution is the forward rate, but the median (the important number to count
the probability of exercising) is below the forward rate. Likewise, the N(d1) for an ATMF option is not equal to
50%.

Realise theoretical profit, the job of a market-maker

FX options market-makers typically work for large financial institutions, with access to lots of credit, client reach
and capacity to take risk. These market-makers are increasingly working for large hedge funds and are being
assisted by computers much more than in the past.
The market-maker is not necessarily paid to have a hunch on where the market will be heading. They are not paid
to know if the euro will rise or fall. They will have an opinion, and usually some residual position that will be
exposed to spot euro movements as well as forward point movements. However, their main job is to separate all the
other elements (spot and forward points) and focus on where volatility is headed. As risk-taking and capital
abundance has been restricted for all the reasons mentioned in previous chapters, they are primarily paid to buy and
sell options, and always be there to make a price for the institution’s clients. We will explore how one can
speculatively trade volatility to make money in later chapters. For now, how do they make money out of this client
activity? Suppose that the price of an option is US$40,000. If a client calls asking for a price to buy an option, the
trader will show the ask (offer) price, which will be higher than US$40,000. If a client calls and ask to sell an option,
the trader will show the bid price, which will be lower than US$40,000. If the client does not want to say what their
intention is, the trader will show two prices, the bid, lower than US$40,000, and the ask, higher than US$40,000.
Suppose the trader has just sold the option for US$41,000. This is good, because they sold something for more
than it is worth, but where is the money? Well, that is an entirely different story. If the trader can buy back the
option for anything less than US$41,000, then they have made money. However, the money is not there until the
trader buys back the exact option. This is where the complications start. The first challenge is to find somebody who
wants to sell the exact option at a reasonable price. The trader could call another bank in the interbank market. In
principle this is easy, but in the interbank market nobody does you any favours. Another bank will likely know that
the value of the option is US$40,000 and that the ask price is US$41,000, and therefore show an ask price of
US$42,000. In fact, it’s not uncommon that the client called both banks and asked for a price. When the client dealt
with our bank trader, they probably told the other bank that they no longer had any interest because they dealt with
someone else. Therefore, the second bank will most likely look to take advantage of the fact that our trader has to
buy back the option, and by the way they’re not all too happy they lost the deal to them. So, the first bank may very
well choose to just manage this risk itself, and not try to unwind the transaction right away.
What does it mean to manage the risk? One thing it could mean is to just leave it on the books and hope for the
best. In that case, the trader will have a short option position for one year and at the end the option will expire, and
this could potentially be good or bad news. The option could be OTM, and therefore not exercised, so the trader
would have a P&L of US$41,000 on that trade. However, the option could be ITM and exercised and, depending on
the value of EURUSD at the time of expiry, the trader might have to pay out, say, US$100,000 to the client. This
would then be a loss of US$59,000. You may have suspected already that banks don’t hire market-making traders to
take the other end of huge EURUSD bets created by their clients. This is a very volatile business. Recall the chart of
short option characteristics. Being short an option means the most you can make is the premium, and you can lose
an infinite amount.
In the “good old days”, some traders at banks did take bets of this nature. However, they typically were not
market-making traders. They were proprietary traders and constituted what amounted to a hedge fund inside the
bank. In other words, they may have been short options but it was options they chose to sell, not some random risk
somebody else did not want. Often, it was the same person who made markets and also maintained a risk portfolio.
If they saw something coming from the clients, they just kept the position for their portfolio. Removing this conflict
of interest, of being a market-maker and a proprietary trader at the same time, was why the Volcker Rule was
introduced. If all your clients are buying, then it is an easy decision to go ahead and buy as well. As a market-maker,
there is nobody in a better position to spot when the tide turns.
We digress. Bank option portfolios are not filled with short option positions that can have unlimited loss if the FX
rate moves dramatically. Managing the risk requires a constant adjustment of the delta so that the portfolio remains
indifferent to spot movements but may have a point of view on volatility. As you may surmise, this means that to
earn that US$1,000 the trader could rely on delta hedging, buying and selling euros for an entire year, theoretically
every second. That is why the only perfect hedge to an option is to buy back the exact same option so that you do
not need to keep the option on the books and you are not bothered by the need to delta hedge. Again, we are moving
too far ahead. In Chapter 9, we will acquire all the tools to start using the delta hedge to manage the risk of the
portfolio effectively so that we do not have to run out and buy the exact same option.

IT’S ALL ABOUT THE PREMIUM

Let’s now dive back into the premium and take a more practical and less theoretical approach. Following is a
prelude to the next chapter where we look at the Greeks and describe the topography of where mathematical theory
does combat with sentiment. In the previous section, we told you that the premium is important and that we have
some intuitive and theoretical ways to calculate it. In this section, we will construct and deconstruct the premium to
understand how we can get some practical guideposts to what happens when the various market parameters move.
The premium can be broken down into two components (see Figure 8.21) – a part that is easily calculated from
market observations, and a part that depends on the distribution, and is not so easy to figure out. The first is called
intrinsic value and the second is called time or extrinsic value.
Intrinsic value is the difference between strike price and forward, if positive, and zero otherwise; this value can
never be less than zero. Time value (optionality), on the other hand, depends on the time to maturity and the
volatility level; this value can also never be less than zero.
Recall the previous brief discussion about ITM, ATM and OTM. If, when subtracting the strike from the forward,
the result is zero, then this is an ATM option. When the result is negative, ignore it and conclude that the option is
OTM. If the result is a positive number, the option is ITM.

The intrinsic value is, frankly, not that exciting. It’s easy to calculate, moves in a linear manner and the reason it
exists is because we can lock that value in at any time during the life of the option. Let’s do a quick example to show
you how that works.
Figure 8.22 is the price of a six-month forward (leg 1), and the price of a six-month euro call/US dollar put struck
at 1.04. The market conditions are as follows. Spot is 1.06, and interest rates in both Europe and the US are zero, so
the forward rate is same as spot, 1.06. One can buy the option and pay US$0.0398/€ (398.624 on the price line in
Figure 8.22), then immediately turn and sell euros forward six months at 1.0600. What will happen in six months
(see Table 8.1)?
 When one can lock in profit from an existing option exposure by doing a forward, that locked-in profit is the
intrinsic value of the option.
Now to the most interesting part of the premium: time, or as it is sometimes called, extrinsic value. Table 8.2
shows the prices for a US$1.06/€ euro call/US dollar put, expiring in one month. The premium is broken down into
intrinsic and time value. For example, if the one-month forward rate (the rate where today one can buy and sell
euros to be delivered in one month) is 1.04, then the price of the option is 68 pips/€. Since the option allows the
owner to buy euros at 1.06, the option is OTM and has zero intrinsic value. Therefore, the entire premium is time
value. Now, assume the euro rises in value to 1.05. Note how the price increases, which reflects the increased
probability that the option struck at 1.06 will end up being valuable. However, the forward is still not above 1.06,
and therefore there is no intrinsic value. Not until the forward rises above 1.06 does the option start to have intrinsic
value.
Compare these prices with an option that gives you the right to buy euros at 1.06 in one year instead of one month
(see Table 8.3). How should the new one-year prices compare with the one-month prices?
What stands out immediately is that at every forward rate the premium is higher for the longer-term option.
Second, note there is no straight-line relationship between the price of the one-month and the one-year option at the
same forward rate. The intrinsic value is exactly the same for both options, but the time value has a more
complicated relationship.
 Let’s plot the time value and get a visual (see Figure 8.23). The important thing to realise is that the area under the
curve is the time value still left for the option. This is what we will explore next in detail.
It is obvious by now that the longer the tenor and the more lifetime an option has, the greater the price, all else
being equal. Time value is also greatly impacted by the volatility, or uncertainty, in the market. This is where the
terminology gets a little confusing. Time value includes the effect of the time to expiration (tenor) plus the effect of
volatility. That’s why it would have been better if the market called this part extrinsic and not time value. We will
explore the time and volatility impact in greater detail in the next chapter. For now, here is a more complete view of
the option price as the forward changes and how this view compares to the option price at expiration.

Figure 8.24 is the scenario tab from the Bloomberg option pricing page OVML. The x-axis describes where spot
is, the y-axis is profitability. Let’s take a moment to clarify how we will use spot and forward terminology
interchangeably from here on. Option prices are based on the forward rate! However, more commonly we relate
option prices, delta hedging and the conversation around spot. The reasons are very practical. Spot prices are easily
observable, but forward prices are not as commonly accessible. Given that the relationship between spot and
forward is well-understood and relatively stable over short periods of time, it is just easier to substitute spot when
discussing changes in option price. For traders and portfolio managers, the reason to use spot has to do with
eliminating some of the transactional friction. Spot is far more liquid across most situations. Therefore, it makes
sense for traders to transact spot with lower bid/offer spreads and then, at the end of the day or another appropriate
period, to manage the forward book. So from here on we will use spot, although we should be using the forward in a
perfect world.
 Figure 8.24 shows the price of a 1.06 euro call when spot is 1.06, interest rates are zero and volatility is 9.472%.
The line ① is the price of the option at expiration. You are now an expert at calculating any point on that line. The
line ② is the option price right now as a function of the spot level. For example, should the euro collapse to
US$0.95/€, the option will be worth close to zero. The intuitive thinking is: there’s a month left but it’s very unlikely
that the euro will rise beyond 1.06 between now and then. Similarly, if the euro jumps to 1.16 it is very unlikely this
option will expire ITM. The prices in between are not that easy to calculate, and are calculated by our now very
familiar Black–Scholes formula.
 Take a look at the one-year 1.06 call using the same axis (Figure 8.25). Even at the extremes there is still some
chance that, in a year, the euro can rise or fall to make the option precious or worthless.
Suppose a trader sells an option instead of buying one. What does that look like? It is the mirror image of the
above figures. Take a look at the two sold option profiles in the following figures. Figures 8.26 and 8.27 display the
pricing page and chart of a sold three-month, yen call/US dollar put. The line ① is at expiration and the line ②
represents the price right now. Note the maximum profit this trader will ever have is the premium of
US$201,972.75. The maximum loss increases, in theory, until the US dollar has no value against the Japanese yen.
Similarly with the yen put/US dollar call, the losses continue at expiration infinitely (see Figures 8.28 and 8.29).
Although the extremes provide some intuitive understanding, the prices in between are not that easy to calculate. As
mentioned, we need to go to our hallowed Black–Scholes formula for that. Many sharp traders have a sixth sense of
what a price should be. Actually, there are also simple shortcut formulas that get you some of the way there.
In the mid-1990s, there was a high-net-worth individual who traded huge amounts in overnight options. He would
call up the bank and typically sell euro puts expiring at 10am the next day. As we will see later, he could collect the
premium if the euro does not fall, and if the euro did fall he was happy to be given (put) the euros because he was a
long-term euro bull and content to own them. Now, calculating overnight options is very tricky as the volatility is so
volatile (pun intended). Figuring out where the option premium is, especially in large amounts, was a very stressful
job. He was a very polite and gracious man with a heavy Swiss accent, but also incredibly sharp. So we would take
great pleasure in putting on the telephone salespeople that may not know him. Once they got over the stress (fear),
greed kicked in and the salesperson would typically look at this opportunity as a way to make their day’s P&L.
Everyone knew he was a seller, so the salesperson would skew the price after punching the numbers into the pricer.
Then our Swiss friend would say: “Oh, that’s a lovely price, but can you please check again?” The salesperson
would posture and apply wonderfully practised intonations such as: “Of course it’s right. How dare you question
me? Do you not realise I have a computer here with tremendous computing capability, and I work for Mega Bank?”
Our Swiss friend would ask for the volatility and in an instant tell the salesperson the right price. He could calculate
overnight option prices in his head instantly.
 There are many traders and hedge funds that sell short-term options, hoping that in the long run they will collect
all these small premiums and be smart enough to avoid any massive move that will wipe them out. Although there is
evidence that selling options can be a smart thing to do, this behaviour is definitely akin to picking up pennies on the
railroad tracks and hoping to jump out of the way when the train comes down the track at 100mph.

CONCLUSION
This chapter introduced numerous concepts. The underlying and unifying force to all of them is probability.
Specifically, what is the probability that an option will be exercised? In that quest, we moved from the simple dice
experiment of the previous chapter into a more complex methodology employing the central limit theorem.
However, we discovered that we no longer had easy answers. How the distribution looked depended on subjective
assumptions. History provided some comfort, but in the end the distribution shape had lots of art as well as science.
At the heart of probability was the standard deviation, or what is known in the market as volatility. This chapter
took a perfunctory look at volatility, so that we have the language to talk about it in the following chapters. Using
our newly discovered maths, we asked whether Black–Scholes is actually correct. In the process of discovering the
answer, we hit upon an activity called delta hedging. Delta hedging is an integral part of what option participants do.
In fact, it is a methodology that can replicate the price of an option.
Finally, in this chapter we decomposed the price of an option into intrinsic and time value, only to discover that
the most interesting part of the option price is time value, which has a direct connection to volatility and time to
expiration. The next section will be dedicated to understanding that area underneath the curves of the figures we
looked at. Getting that right is the day job of tens of thousands of people across financial markets, across the world,
every single day.

1 Of course, different people might have different opinions about what this future distribution should be. Here we will only be concerned with what is
called “risk-neutral” distribution, which corresponds to the case where all asset prices equal their expected value under such distribution.
2 σ is the symbol most commonly used to represent standard deviation.
3 Standard deviation is a way for us to ultimately describe mathematically the narrowness or width of the curve referred to in the previous chapter. Note
that the “shape” of the curve is uniquely determined by just one parameter, the standard deviation. If you shift the centre of the curve, the curve shifts
but the shape is the same. However, if you change the standard deviation the shape changes. Note that it is false that if you double the standard
deviation then the curve just swells to double the width in a stretchy motion. In other words, the new shape of the curve will be genuinely different,
not just the stretched image of the original curve.
4 Keep in mind Figure 8.7 shows daily returns, but we will have to scale them up to yearly returns – just a technical point …
5 The settlement date is the date at which the currencies we agree to exchange at expiration must be delivered to their respective checking accounts.
6 To rephrase: the average, or mean, P&L of a forward trade is zero. This is not the same as saying that half the time you make money and half the time
you lose money, which is instead the definition of the median.
 7 And of course the probability of being below 1.06 will be denoted by N(1.06).
8 See Lars Tyge Nielsen, Understanding d1 and d2 : Risk-Adjusted Probabilities in the Black-Scholes Model, INSEAD, October 1992 (at
http://www.ltnielsen.com/wp-content/uploads/Understanding.pdf).
9 It would be rigorous but it makes the assumption that we know the probability distribution of the exchange rate is the lognormal curve.
10 FX options are not like equities. With equities you expect a return higher than the risk free rate. With FX options we shall see that you could make sure
via your hedging activity that you break even each time. Therefore no expectation of higher returns.
11 In other words, the derivative of the portfolio value with respect to the exchange rate is zero.
 9
The Greeks – A Practical Approach

Time to dive deep again, and put a more definitive structure around the way in which the premium changes as
different market parameters move. We briefly discussed the first derivative of the option price with respect to
movements in the forward.

The change in the option premium divided by the corresponding change in the forward rate is called the delta (δ,
lowercase, or Δ for capital) and indicated by N(d1). It is also the slope of the option price when plotted as a function
of the forward rate. The second derivative is called gamma (γ, Γ), and it is also the change in the slope of the option
price when plotted against the forward rate, then the change in option price with respect to one interest rate is rho (ρ,
P) and with respect to the other rate is phi (ϕ, Θ). Got that?
Don’t worry, we will review what is important gradually, starting with why these derivatives are called “the
greeks”. It is just that Greek letters are used to represent these derivatives. Therefore, when speaking about the
derivatives we will use lowercase, “g” for greeks. Having said that, one of the most important derivatives is the
change in option premium with respect to volatility. That is called vega or sometimes kappa. This is very upsetting
because vega is not a Greek letter, or even a Latin letter, it is just a pseudo-ancient-sounding word. It also happens to
be a star, a lunar crater, a programme of space missions and a place in Canada, Norway and California.
I remember sitting next to someone a few decades back who insisted that he and his colleagues at Midland Bank
London had come up and popularised vega as the way to describe sensitivity to volatility. He had worked for
Midland Bank in the late 1980s at a time when the market was grappling with ways to understand volatility, and
Midland was a powerhouse in the option market after Swiss Bank Corporation (SBC), Bankers Trust and others.
One quiet afternoon, having nothing better to do, they started throwing names around for this derivative and as a
joke started calling it vega. Before they knew it the whole market was calling it vega. I cannot vouch for the
authenticity of the story.
In this chapter, we will review how the greeks decompose the risks contained in an option price into their various
constituent parts. This in turn allows the trader to decide which risks to retain and which to hedge. Stated differently,
when a trader buys or sells an option, they take on a lot of moving pieces. The price is affected by the passage of
time, volatility, interest rate and forward movements. That’s a lot of disparate moving parts. What a trader wants to
do is separate, identify and unwind the risks they do not want and keep the risks they do want. Typically, an option
trader keeps the exposure to volatility and aggressively manages the exposure to forward movements, which
includes spot and interest rate movements implied by forward moves.
The other reason why you would look at this alphabet soup of Greek letters is the decomposition of the P&L of
any option portfolio. What follows is true of both an option and an option portfolio, just the same.
Suppose that you are trading a portfolio of options (let us call its value Π) and overnight you dropped 20 million
dollars. Surely you would want to run a full analysis of where the loss came? As all market values moved overnight,
you will have to look into each single input to the prices in your portfolio. Suppose that your portfolio value depends
on the level of spot, S, on the implied volatility, σ, on the interest rate, r (there will be more factors, but for the sake
of our example these will suffice).
If you were paying attention in class then you may remember Taylor’s formula from your maths lessons. If the
value of your portfolio is written with the dependency on the above market levels, it looks like this:

The meaning of this is: the value of the portfolio today equals the value of the portfolio yesterday plus delta
multiplied by the change in spot, plus vega multiplied by the change in volatility, plus rho multiplied by the change
in the interest rate, plus gamma multiplied by the square of the change in spot, plus d-vega d-vol (or vol convexity)
multiplied by the square of the change in volatility plus an error.
First of all, you now see the tremendous importance of delta, gamma and all the greeks. Secondly, you can see
that there is a method to the madness: you first differentiate once with respect to each variable, then you move on to
the second order. You now differentiate twice but then multiply by the square of the market change (to be honest at
the second order you should also include the second derivative with respect to the interest rate, and then the
combinations of the derivative with respect to spot of the derivative with respect to the volatility (called d-vega d-
spot) and other combinations of two inputs) then you should move on to the third order.
To conclude this rudimentary tutorial: two comments. Firstly, only the first two orders are ever calculated by
traders, and sometimes not even all of those. Secondly, as you compute more and more terms, the ε, the error term,
will tend to zero (this is true only if the portfolio value is expressed by an analytical function, but that is just a
detail). Getting an approximation as good as you wish, will allow you to explain to your boss why you lost all that
money overnight. For a reasonable approximation, you can just compute the first two terms (delta and vega) and
those should explain most of the P&L change.
Hopefully we have given you an idea of why the greeks are so important.

DELTA: δ

Below are listed the delta (δ) properties.

Delta is expressed as a percentage and has the following properties:

ATM options have a delta of approximately 50%, which indicates that (statistically) there is:
• a 50% chance that the option will mature in the money and be exercised; and
• a 50% chance that the option will mature OTM and expire worthless.

During the life of the option delta moves:

from 50% to 100% as the option goes ITM:

• for call options this requires the spot rate to rise; and
• for put options this requires the spot rate to fall.1

from 50% to 0% as the option goes OTM:

• for call options this requires the spot rate to fall; and
• for put options this requires the spot rate to rise.

At maturity the option delta must be either:

100% because the option has expired ITM and will be exercised; and
0% because the option has expired OTM and is worthless.
Delta describes the relationship between the spot rate and the option price, and is usually defined in one of three
ways:

change in the option value for a change in the spot rate;

percentage probability that the option will expire ITM and be exercised as a broad approximation, and not
strictly true but helpful; and
the percentage amount of hedge required in the underlying instrument to neutralise the impact of small spot
rate movements.

Starting with the first property, it is important to note that delta is expressed as a percentage of the notional. There is
nothing particularly sacred about that convention, but having agreed to it the market uses those percentages to
communicate many ideas. For example, when we say an option is 50 delta we mean it’s an ATM option.
Furthermore, the market has settled on making 25 delta options special as a benchmark indicator of an OTM option.
If you are interested in the actual formulas (here P(S) represents the price of the option when the spot rate equals
S):

The second and third properties will become self-evident as we dig deeper. However, consider that the option’s delta
is mirroring the probability that the option will be in or out of the money. As spot move to extremes, the probability
of the option being in or out the money by expiration becomes more evident. Finally, this concept will manifest
itself at expiration when the option will definitively be in or out of the money, and will have either a zero chance or
a 100% chance of being ITM. Given the strike, for put options the delta is obviously not the same as the delta of the
call option. In the case of the put option, delta equals N(d1) – 1, which is always a negative number.

Delta hedging
The most important delta characteristic is the last point in the fourth property (ie, delta describes the percentage
amount of hedge required in the underlying instrument to neutralise the impact of small spot rate movements). We
touched on this briefly in the previous chapter, but now we will revisit in a more deliberate way. You’ll have a third
chance to see it in action when we talk about how bank traders or hedge funds trading volatility actually manage
their books.
 Suppose a trader sells a US$ call/¥ put expiring in three months. What does their position look like the instant
after they sell it? In Figure 9.1, just like those in Chapter 8, the line ① describes what happens to the option price if
spot moves instantaneously to that level. The “at expiration” line we had previously is removed from this chart.
An options trader will realise that this exposure can be impacted on by a lot of things – eg, volatility, interest
rates, time and spot (forward) movements. The most dramatic and immediate impact will arise from forward
movements. As mentioned, forwards are just not as liquid as spot. Since spot is directly linked to the forward by
interest rates, which do not change as quickly, the real and imminent danger is spot. Therefore, from here on we will
talk about changes to spot but will ultimately be concerned with the forward. Okay, what can the trader do
immediately to reduce the spot effect? The option loses money as the US dollar increases in value. So, to protect
against that, they should buy some US dollars versus yen. But how much? You guessed it! Buy dollars in the
amount of the delta!
Figure 9.2, a pricing page from the Bloomberg OVML function, shows all the inputs and the important outputs.
The delta can be found in the results section. It is 48.883% of the notional. So, US$10 million × 48.8834% =
US$4,888,342. Market participants would make it easier and change the strike to 107.60 and make the delta 49%.
For convenience of calculation, however, to prevent confusion we will tend to keep the extra decimals and the exact
strikes. As soon as a corporate hedger, say, tells their bank salesperson they would like to execute this deal, the
salesperson will immediately turn to their spot desk and purchase the ~US$4.9 million against yen for the benefit of
his option trader’s book.
 There’s an awful lot of games and risk associated with the above scenario. The salesperson that priced this option
assumed that they would be able to buy the delta hedge at the spot rate they entered into their pricing formula,
108.00. If in that instant the rate moved higher, to say 108.15, they will lose money or at least not be selling it at the
price they thought. The risk is mitigated when the salesperson adds a little cushion before showing the price. The
client, of course, will want to get the best price possible. Leaving those elements aside for now, what happens next?
Both transactions go into the trader’s book. Therefore, they now have:

trade 1 – short US$ call/¥ put, expires three months, strike 107.58 on US$10 million, premium ~US$203,000; and
trade 2 – long US$ versus ¥, US$4,888,342 bought at 108 value in two days.

If we look at the delta hedge by itself, its P&L profile looks like that in Figure 9.3.
Since it is a purchase of US dollars, as the US dollar rises it is more profitable. You can easily calculate the profit
at any point on this graph. Let’s take 112.00:

Profit = US$4,888,342 × (112–108)/112 = US$174,583.64.

Now, we can simply add the P&L of those two trades together at every point on the x-axis. Figure 9.4 is what the
combined graph looks like.
By selling an option on US$10 million and buying the delta amount against it, the trader has an unlimited loss in
both directions – ie, if the US dollar rises and if the US dollar falls. Great job! Give it up, and see if you can become
a forest ranger.
 Having traded and structured options for major financial institutions my entire adult life, I have many times been
on the not so pleasant end of the turmoil that can happen in banking. In one of those transitions, the bank offered
career counselling services, so I talked to a consultant about just leaving it all behind and doing something
completely different. After talking to me for a while, he leaned over his desk and said: “After decades in this
business you are not going to be a forest ranger; why don’t you take some time off and then get yourself back in the
game?” I’m not so sure that’s the best advice I ever received. Who knows? If I hadn’t listened, this book could be
about the interesting ways to create art from mushrooms found in our national forests!
Seriously, all is not lost. Note two impacts of combining the option with a spot transaction.

Although the combined position loses an unlimited amount in both directions, the speed with which they lose is
greatly slowed. This does not show on the chart, but you can check it on the vertical axis: the losses are now
much smaller. Note also that the upside of this situation is that the trader received a premium by selling the
option in the first place!
The spot transaction combined with the option gave a temporary reprieve from any P&L impact for tiny spot
moves. Look at the area around 108.00 – very little money is lost as spot moves.
On second thoughts, well done! One factor neutralised, for now. By the way, the hedge was executed using a spot
transaction. As mentioned earlier, it should be hedged using a forward transaction. There is no room for spot in our
magnificent Black–Scholes formula. If we use a forward transaction the amount is slightly different. How different
the forward hedge is from a spot hedge depends on absolute interest rates, the differentials and the time to maturity.
You can use the chain rule for differentiation you learned in high school. As the forward is just the spot multiplied
by the ratio of the discount factors in the two currencies, the two deltas only differ by a constant equal to the ratio of
the two discount factors. Although the difference between the spot and forward hedge can end up being much more
significant for, say, long-tenor emerging market options, for now suffice it to say that if hedged with a forward, the
hedge would be to buy US$4,898,781 against yen.
There are at least two real-world considerations with hedging. If spot is used as an indication to get an option
price, or if there was an exchange of the delta hedge with the counterpart,2 there is an implicit understanding of the
current market quote for the swap. The swap would be to get out of the spot position and enter into a forward
position. The problem arises when there is a disagreement of where the forward market can be transacted
effectively. This is typically not an issue with, say, EURUSD, but may be an issue with USDBRL where the
bid/offer spread is wider or a market may not be available. In this case, the transacting counterparts will agree to the
option deal with the understanding that they will also exchange a forward delta at an agreed-upon rate.
 The second practical consideration is that traders want to deal as few times as possible in the forward market so as
not to incur extra bid/offer spread. What option dealers do is hedge all day with spot, and then at the end of the day
look at their net exposures in the forward market and execute a few deals that hedge the forward risk. This reduces
transactions that incur a higher bid/offer spread.
Let’s continue with the above example. Assume the trader sells the option and does a forward hedge out to match
the option expiration. The position looks like Figure 9.5. The distance between the two lines shows the path the
option will follow over time (more on that when we talk about theta).
For now let’s consider what happens as spot continues to move. The delta will change again. The trader then
needs to adjust their hedged amount outstanding. If spot moves to 109.00, they now need to have on hand
55.3018%. If spot keeps moving higher, then they will need to buy more US dollars to meet their delta hedge. Let’s
do one more at 112.00. Now they need to have 71.8708% of the US dollars on hand. You see where this is going?
As spot moves higher, the person selling the option will have bought all the US dollars they will need to deliver at
expiration.
Suppose spot moves lower again. Then they need fewer US dollars on hand and the delta is lower. In that case,
the trader will sell some US dollars and buy yen. Assume that this trader delta-hedged again at the same levels as
spot went lower, and let’s add one more level at 107.00. Figure 9.6 and Table 9.1 show how the delta hedge amount
will change as spot moves, with 91 days left to expiration for this option.
Therefore, the trader executed the deals shown in Table 9.2 to hedge.
All this trading has caused the trader to be long US dollars and short yen at an average rate of 109.38. That’s
terrible. Remember that if this option is exercised, the trader will need to deliver US dollars for yen at 107.58. When
the deal was priced the assumption was that you can hedge at 108.00. Can you guess what will ease the trader’s
pain? If you said the premium paid then you are right. So, in addition to all the delta trading, we need to account for
the US$203,182.55, received up front, which is ¥2.19/US$ ((US$203,182.55 × 108)/US$10 million). Therefore,
109.38 – 2.19 = 107.19 is the effective rate at which the trader bought US dollars. So far so good, the trader is
actually doing better than the original 108.00.
 The levels at which one executes a delta hedge, and how often, is the art in this business. Some shops are very
quantitatively driven – they will delta-hedge at specific levels and in specific quantities with military discipline.
Others will use that old-fashioned instinct to make market decisions. Think of a situation where something happens
in the political world and there is widespread panic for an instant. The dollar is higher and the delta skyrockets, and
the machine is screaming: “BUY DOLLARS IMMEDIATELY”. A seasoned trader may think: “Not so fast, the
market will calm down again and I will just be whiplashed.”
I used to work at a major bank structuring FX option hedges for clients. The bank had a deal with a famous
trading shop that did all the trading and risk management. So we would structure and sell the deal but get a price
from this other shop, while the risk was handled by these trading shop traders. The traders were under strict orders to
delta hedge under very rigid rules no matter what. They were the forerunners to today’s algorithmic trading shops. I
have tremendous respect for their risk taking and their discipline. The fact they are still around and doing well is
testament to that. However, I will never forget how they hedged by selling US dollars at just below ¥80/US$ in
April 1995, having priced a massive deal for a large hedge fund. These were the days when inside market colour on
currency news and potential moves made you a hero, not a villain. This level was not seen again for another six
years. They were so focused and geekie that we used to joke that you needed a social lobotomy to work at this shop.
Now, back to our example from above. The more ITM the option becomes, the closer the delta gets to 100%.
Another way to say it is that, if the option goes well ITM, the trader will have accumulated 100% of all the currency
they need to deliver at expiration. Therefore, the seller of the option will be able to deliver the US dollars promised,
or zero if the option is OTM and expires worthless. However, here is the punch line. If you sold the option and got
an upfront premium, every delta hedging deal you do is a losing deal! You are buying high and selling low.
If you bought an option, then every delta-hedging deal is a winning trade. Here is the most counterintuitive fact
about all this – the option market-makers do not care if the option they sold or bought is ITM or OTM. What they
care about is that the rate at which they bought or sold the FX hedge cost them less than the premium they received
or paid. If they bought the option, it is important that the money they made delta-hedging is more than the premium
they paid. If they sold the option, then they care that all the losses by delta hedging are less than the premium. Think
of the extremes – from a speculator’s perspective, one sells an option and there is no more FX trading. The market
stays exactly where it is for the entire time. The seller keeps the premium, and has a nice day. What about the person
that buys the option? They hate that. They paid all this money expecting movement (why else would you buy it) and
nothing happens, you just lost your premium and have nothing to show for it.

GAMMA: Γ
Option traders are geeks, or wannabe geeks, at heart. Consequently, they need to have a geekie name for everything.
One of the behaviours that are very important to monitor and control is how quickly the delta changes. That rate of
delta change is named after the third letter in the Greek alphabet: gamma. In other words, if the exact option we’ve
been playing with in the previous section had a different expiration date, or maybe a different volatility level, the
speed with which the delta changes will be different.
More formally, gamma is:

the rate of change of an option’s delta with respect to the price of the underlying instrument; and
the second derivative of the option price with respect to spot (forward).
Some gamma characteristics are as follows:

An option, or a portfolio for that matter, is said to have high gamma if the delta changes quickly as spot moves. It
has low gamma if the portfolio’s delta is relatively insensitive to spot moves.
For short-term options, where the strike is close to spot then the delta changes very quickly as spot moves, gamma
is high. Think of an option struck at EURUSD 1.0500. At 9:59am,3 with one minute to go, spot is hovering right
around 1.0500. One second it is 1.0501 and the next it is 1.0499. At 10am exactly, the option will either be
OTM and have a zero delta or ITM and have 100% delta. As the seconds tick away, the delta will change
quickly from close to zero to close to 100%. In market jargon, to be long gamma means you own this euro
option and benefit as spot moves; you can trade your delta around this and have a winning trade every time. To
be short gamma means you are short the euro option and are hurt by market moves.
Mathematically, delta is the slope of the option price as spot moves, and gamma is the slope of the delta curve we
showed in the previous section. This also means that gamma is the convexity of the price curve.

Figure 9.7 is a depiction of the gamma. Stated more plainly, it is the difference of the delta from one point to the
next. Unlike the delta, which goes from 0 to 100%, gamma goes from zero to zero and looks like a normal
distribution centred around the forward.
The reference to the expiration time reminds me of the first few months I was on the options trading desk. These
were the ancient days of the mid-1980s when the telephone and Reuters were the only forms of communication. I
was the junior person on the desk (as my boss was six months older and had been on the desk six weeks longer than
I had). It was my job at 10am to call and exercise all the options we were long and ITM, and to accept calls from all
the dealers with options we were short and ITM. The head trader of a smallish bank called to exercise a deep ITM
option. By the way, there were many players in this game before the big banks dominated. This head trader was an
arrogant man with a very heavy French accent. When I picked up the phone, he said: “I am so and so and I exercise
this option.” I do not know what possessed me, but I thought it was a good time to have fun. I said in my best
Queens accent: “Ray’s Pizza, what options you want on your pizza?” He hung up and called back repeatedly and
frantically demanding to speak to my manager. Every time I would pretend it was a pizzeria. He kept calling the
Reuters code and I would respond that the option guys were not in today. Around 12:30, the desk gets a call from
the chairman’s office of my bank. They wanted to understand why the chairman from the other bank was furious
and demanding an apology. Needless to say, from that day forth I took 10am NY time as the most sacred time of the
day, never to be defiled again with humour.

Here’s a summary of the thinking on the premium so far. The option’s value at any time can be split into two.
First is a component that is easily calculated and measures the part of the value that can be locked in: the intrinsic
value. The second is a complicated and less-predictable change to the premium: the time value. The most urgent
factor to consider that affects the premium is spot. The changes in the value of the option as spot moves are
measured by the delta. It is important to understand how quickly delta changes, and for that we use gamma.

THETA: Θ

What other conditions affect the value of an option, and specifically the time-value component? The passage of
time. We discussed previously the idea that the longer the tenor of an option, the more valuable it is and therefore
the higher its cost. However, now we want to understand more fully how the passage of time makes the premium go
up or down. Actually, as time passes and all else is constant there is only one way to go – alas, one of life’s
unalterable rules. As time passes, all else being equal, time value goes down. To be precise, there are some cases,
with deep ITM options in currencies with high interest rates where as time passes the value of the option rises, but
those do not occur often.
Look at the option we’ve been working with, the USDJPY dollar call/yen put struck at our odd strike of 107.58,
expiring in 91 days, leg 1 (see Figure 9.8). We’ve put in a flat volatility (②) for this example just to keep things
constant, so we can focus on time. For leg 2 to leg 4, note that the main change is that these legs have less time until
expiration (①), from 91 days to 60 days to 30 days to one day. Note how the delta (⑤) is affected by time. When
there is almost no time left and the option is OTM, then it is close to zero. The more time the option has, the more it
looks like the graph we drew in the previous section.
 Spot and the forward are set to 107.00 or less, so that at no point does this option have any intrinsic value. This
allows us to focus on the time value, which is the entire premium (④). As you can see, the premium (the value of
the option) decreases as time passes. The decrease of the value for one day is called theta (③). Look at leg 1: the
theta is US$1,083.57. Move across the theta values, and the theta or time decay increases as the option has less and
less time.
Finally, with one day left this option will be ITM and have intrinsic value, or will be OTM and worth zero. Since
107 spot and forward in fact says there is little chance of its being in the money (delta is 15%), then the entire value
of the option must be the theta on the last day: US$4,041.03.
Note how the theta increases abruptly near the end. Stated differently, the price of the option drops abruptly.
When we examine a real portfolio later, this will have important implications for the trader’s decisions as to what
options to hold until expiration and when to get out of given positions.
In Figure 9.9, the solid line is the value of an ATM option, and the dotted lines are the values of OTM and ITM
options. Note how the time value of the option drops off abruptly near the end of the option for ATM options. OTM
and ITM options lose their time value in a (quasi) straight-line basis.
 Let’s combine theta and gamma in our thinking:

paying theta is the cost of being long gamma;

receiving theta is the benefit of being short gamma;
every day the value of the option decreases; and
if the P&L from gamma trading is less than the time depreciation of the option (time decay), this position loses
money. Assuming the implied volatility stays the same.

Now combine delta, gamma and theta. Figure 9.10 makes a first attempt to move three levers at the same time. The
figure examines how delta, in-the-moneyness and time affect each other. This is our ultimate destination, and is
what a portfolio manager of options has to do – work out how all these components moving at once affect the value
of the portfolio. Think in multiple dimensions and distil the impact down to one number, the value of your portfolio.
We will look at this effect in a more realistic scenario in Chapter 11, but for now let’s return to the theta alone and
another perspective. Recall what our option looked like today and at expiration, and let’s fill in the dates in between
(see Figure 9.11).
Line ① traces the P&L of a delta-hedged option at expiration. We also know that line ② is the delta-hedged
position today, allowing for some stability around the current spot market. Line ③ is the future price of the option if
time passes. In this case, we simulate the passage of around seven weeks. Note how the value at the extremes is
more or less the same for all the curves, but if we remain at the same spot level the option continues to lose value at
the theta rate we described above. The seller of the option is very happy to keep earning money as time passes. Of
course, the opposite is true for the buyer, as Figure 9.12 depicts.
 To summarise, as time passes:

the distribution of possible outcomes shrinks;

the value of all options declines;
OTM options become less likely to be exercised (delta declines);
ITM options become more likely to be exercised (delta increases);
at expiration, an option is worth only its intrinsic value;
time decay occurs slowly at first, then more rapidly as the option approaches expiration;
theta is the change in premium for a unit change in time to maturity (usually one day);
ATM options have the most time value and therefore the highest theta – this theta increases as the option
approaches maturity;
OTM and ITM options have less time value and therefore a smaller theta – this theta is almost constant over the
lifetime of the option; and
when volatility increases, theta increases! (Why?).

VEGA OR KAPPA: VOLATILITY

We’ve already discussed volatility in previous chapters. In this section, we want to focus on itemising the most
important characteristics and continue to tie together the other greeks and highlight their interdependency.

Volatility measures the jumpiness of the market:

volatility is not concerned with whether a currency pair goes from A to Z, but more how it gets there.

A measure of the fluctuation in the market price of a given currency pair. Mathematically, volatility is the
annualised standard deviation of the lognormal returns:
Return: Rt = log (Pt/Pt-1)
 Number of days = N. Days in a year: 255
Standard deviation:

Two types of volatility:

historical (realised); and
implied.

How do changes in volatility affect the premium of our option? The volatility we will focus on is implied volatility.
As you know by now, this is a key component that anyone must put into the model if they want to see an accurate
option price.
Option traders are volatility traders for a couple of reasons.

The parameters, other than volatility, that go into the pricing model are either common and easy to agree on, or are
set at inception – such as strike, term and put/call. Traders, however, focus on volatility when they communicate
with other traders, so a typical conversation may be: “Where are you on a 25 delta, euro call on 50?” This
means: “What is your price for a one-year euro call/US dollar put, struck at a level where the delta of the option
will be 25, and for a notional equal to 50 million euros?” The response may be “10.25 at 10.5”, which is a
volatility quote. If they agree to buy that option at 10.5%, for example, then they pass the trades on to the junior
traders to calculate the exact details.
Option traders trade volatility by design. They are not spot dealers, trying to forecast spot moves, nor are they
forward traders trying to understand the course of interest rate markets, although they do both by default. Their
main focus is to intermediate all of the spot and forward risk into the market, and be left with the volatility
alone. We got a sense of how they do that with the delta hedging. Similarly, they hedge out the forward risk, but
that is beyond the scope of this book.

Vega is usually quoted as a number that relates to that option at that point in time. Therefore, we may say that the
vega of this option is US$100,000. Meaning that if volatility increases (decreases) by 1% instantaneously, the price
of the option will rise (fall) by US$100,000.
Given the importance of volatility, how does it affect the premium at inception and as the other parameters
change? All options are not equally impacted on by changes in volatility. The longer the term of the option, the more
sensitive it is to volatility. Since the measure of volatility we are using is an annualised measure across all options, a
change of 1% in volatility will impact on a one-year option much more than a one-day option. Look at Figure 9.13.
Vega is highest in longer-dated ATM options and decreases as expiry approaches. The sensitivity to volatility is
greatest for ATM options. As the strike moves away from the forward, for OTM and ITM options the sensitivity
persists but the difference between short-term and long-term options is less.
Look at our tortured option example again. Just above the theta (Figure 9.8), the vega is a US dollar measure of
how much the option price will change if the volatility changes by 1%. Nice and simple. Note that leg 1, which is
due to expire in 91 days, has a vega of US$19,617.20. Leg 2, which is a 60-day option, has a vega of US$15,903.30;
leg 3, US$11,166.37, and finally the one-day option has a vega of only US$1,217.78! Think of it this way – for the
price of the one-day option to increase by US$11,000 due to volatility changes, volatility has to increase by 9%!
That is massive. For the 30-day option, leg 3, to increase by US$11,000, volatility has to only increase by 1%.
Let’s summarise vega.

Historic volatility:
an actual measure of past fluctuations in the market price of a given currency pair.
Implied volatility (the focus of this section):
a trader’s prediction of future fluctuations in the market price of a given currency pair;
the traded price of a currency option determined by supply and demand in the market; and
option traders are volatility traders.
Vega is the change in premium for a unit change in volatility (usually 1%):
long an option (call or put): positive number
• almost zero for short-dated OTM or ITM options.
vega is higher when:
• the option strike is near the forward; and
• the maturity is long.
for ATM options, vega is linear: same price increase for each increase in volatility; and
vega decreases with passage of time.

As expectations of future volatility rise:

the distribution of possible outcomes broadens;

the values of all options rise;
all options become less certain of exercise/expiry (delta tends towards 50%);
increased volatility can retard or even reverse the effect of time decay, but only temporarily; and
a decline in expected volatility accelerates the effect of time decay.

INTEREST RATE SENSITIVITY: PHI Φ, RHO Γ AND MORE ADVANCED GREEKS

EXERCISE 9.1: aDDING UP ThE GREEKS4

Here’s a really hard exercise given the material we have covered so far. If the answer comes easily, you will
only want to read this book for the anecdotes, the option stuff will be too easy for you …

Suppose for a US$ call ¥ put, strike 108.00 in US$100 million, you observe the following greeks for the
different expiries shown:

Using the information above, how would you go about creating a long gamma/short vega position?
Does the structure net pay or receive time value?

Solution to Exercise 9.1

This exercise expects you to jump to the next level of sophistication, where you’re buying one option and
selling the other.
If you buy the one-month call and sell the three-month call, you will create a long gamma/short vega
position.
The net position costs theta, or loses time value every day.

Remember that gamma increases and vega decreases with the passage of time; and that an option is a
decaying asset.
We will not spend any time analysing the effect of interest rate changes to option prices, other than to let you know
that this is an interesting derivative that impacts option prices, particularly longer-term options. Since the focus of
this book is on the short to medium term, interest rate changes are just not that impactful.
There are many more greeks used to really understand how option values are impacted by changes in volatility,
such as the volatility of volatility, how the delta changes are affected by volatility and how the risk reversals (to be
discussed shortly, get excited!) are affected by volatility changes. Some of those are beyond the scope of this book,
but we will touch upon others in a very practical way in the following chapters.

CONCLUSION
This chapter took a pulse of the most important levers affecting option premiums, and tried to give you a sense of
how the premium changes as you pull at these levers. As we added levers to our repertoire, we stressed the
interdependency between these switches. It is the ability to move these levers across a portfolio of thousands of
options that makes or loses money for a trader, and allows a speculator or a hedger to enter into the best position to
accomplish their goals.
The following is a list of what we discussed and a summary table of how a purchase or sale of an option impacts
on the greeks.
Currency options are multidimensional instruments, and their price in the secondary market responds to:

changes in the spot rate;

the passage of time;
changes in the level of implied volatility;
changes in the base currency interest rates; and
changes in the foreign currency interest rate.

Which greeks does the market use?

Delta:
change in the option value for a change in the spot rate;
percentage probability that the option will expire ITM and be exercised (approximately); and
the percentage amount of hedge required in the underlying instrument to neutralise the impact of small spot
rate movements.
Gamma:
change in the delta for a change in the spot rate; and
may be defined by either a 1% change in the spot rate or a basis point change in the spot rate.
Theta:
change in the option value for the passage of one calendar day.
Vega/kappa:
change in the option value for a 1% change in ATM implied volatility.
Phi:
change in the option value for a 1% increase in the base currency interest rate.
Rho:
change in the option value for a 1% increase in the pricing currency interest rate.

Not discussed here, but will be explored in Chapter 10:

dVega/dVol – change in vega for a change in volatility; and

dVega/dSpot – change in vega for a change in spot.

If the trader’s position is:

delta positive – they want the underlying spot to rise;5

gamma positive – they want the underlying spot to move swiftly, regardless of the direction (they want actual
volatility);
theta positive – the passage of time will increase the value of the trader’s position;
vega positive – they want implied volatility to rise; and
rho/phi positive – they want domestic/foreign interest rates to rise.
1 For put options the delta will technically be negative. This is because, as spot rises, the option value decreases. In what follows it will be clear from the
context whether we are talking of the absolute magnitude of the delta for the put option or just of the delta of the put option. In the present case, we
are talking of the absolute magnitude (the absolute value of the negative number).
2 Between professional traders, and often with clients, both the option and the delta-hedged are agreed to simultaneously. Therefore, if the market taker
buys an option, they agree to two trades. More on that when we discuss hedge funds.
3 Recall that options expire at 10am New York time, 3pm London time or 3pm Tokyo time, and some other less popular variations for specific emerging
markets currencies. This is agreed between the parties at inception of the deal.
4 Over the years, and hundreds (if not thousands) of presentations of this nature, we have borrowed examples and phraseology from many colleagues to
the point where much of this is really in the public domain. I think this exercise came from collaboration with Jerome C., who is one of the smartest
option people I’ve had the pleasure to work with.
5 In the interest of bookkeeping, earlier we referred to the delta of a put as to a positive number, an abuse of language. Here we are being more exact, and
mean that if you are long a call you hope spot will rise, and if you are long a put you want spot to fall.
 10
Portfolio Management and Second-order Greeks

The previous chapter armed us with powerful tools that enable us to peer into the heart of option valuation and
option risk management. These tools can be complex in their own right, but they get very challenging with the
awareness that to get anywhere you need to consider all of them at once. However, there is one key characteristic
that allows us to breathe easier – the results of each analytical tool can simply be added across many options. As we
saw in Exercise 9.1, this allowed us to engineer portfolios that react a certain way to market conditions. That is
powerful! In this chapter, we will take the closing realisations from the previous chapter and build on them. We will
begin by examining the greeks and issues of a particular combination of options, called the “risk reversal”. As a
consequence of that analysis, we will examine the market conventions around the risk reversal, and its impact in a
portfolio setting. As we’ve done before, as the chapter progresses we will ratchet up the complexity to include other
tools that are helpful in portfolio management.

OPTION PORTFOLIO MANAGEMENT: THE THEORY AND THE REALITY

Recall the situation we introduced back in Chapter 8 concerning the market-maker’s job. The issue dealt with was
about how a market-maker can make money from the option they sold. More broadly, how can a market-maker
manage risk effectively?
The solution to this puzzle is portfolio management. At any point in time, the trader will have a number of options
in their trading book – some options they bought from clients, some they sold to clients. Many options will have
between one-day and three-month maturities, but others will be six months, one year or in some cases will be a long-
term option going out a few years.1 The perfect way to manage the risk is to match off against every option. They
would have to buy back the exact option they sold and sell off the exact options they bought. That just doesn’t
happen in a real-world trading situation. It is not efficient, because at any point in time there might not be any
interest in selling, say, a 2.5-month, 1.06-euro call on US$35 million. Furthermore, even if there were interest in the
market for the opposite risk, they would have to pay the spread, hit the bid and take the offer, and would most likely
lose money on most trades.
Spoiler alert: the way to manage the risk is to use all that we learned in the previous chapter to mitigate and
control specific risks. The trader will calculate the exposures to each of the parameters we discussed in the previous
chapter for every option, then add these risks up. They will try to reduce their exposure to unwanted risk to zero.
They choose to keep certain risk because they think it gives them an advantage over the prevailing market sentiment.
If all the risks (all the greeks) in the portfolio are zero, then the portfolio value will not change with small changes in
time, spot FX or interest rates. As these market conditions change, the risk is no longer zero and they must start the
process again. Now that we’ve given you the solution up front, let’s go back and reconstruct the process so we can
understand it.
What is the portfolio value of all their options? Well, with just one option the portfolio value is the cash received
as premium together with the short option position – in other words, the option is an asset to the client but a liability
to the market-maker trader who sold it. Before the trader unwinds this risk, the value of the option is calculated at
mid-value, and is subtracted from the premium received. So, what is the value of the option, the mid-market value?
In the case where they sold the option and received US$41,000 for a premium, the portfolio value would be
US$41,000 minus US$40,000 (mid-market valuation) = US$1,000. In other words, the trader made a profit that
equals the mid-to-ask spread on the sold option. This is the mandate of the true market-maker trader. It is not to call
the market direction, or to make money by taking the opposite side of all the trades that the bank’s clients enter into.
The mandate is to make the mid-to-ask spread. The value of the portfolio equals that spread. So, is the trader done?
Can they now surf the Internet for the rest of the time until expiration? No, because the US$1,000 is still just a
mathematical fiction until the liability (ie, the sold option) disappears. In fact, if EURUSD suddenly goes to 1.50,
then the value of the option will skyrocket and the US$1,000 in profit would turn into a huge loss. The true job of
the trader is to protect the money that has been hypothetically made until the uncertainty linked to the liability is
gone. This only happens when the option actually expires.
What is the easiest way to make sure the value of the portfolio stays at US$1,000? To have the value of the
portfolio never move. But this is impossible! No, not at all just think of our greeks discussion. We saw that if the
trader buys € Notional × N(d1) versus US$ via a forward contract, then the portfolio value will not change for an
infinitesimal change in the value of the EURUSD exchange rate. This is pure magic.
How come? As we discussed in the previous section, if EURUSD rises then the value of the option rises (which is
bad for the trader), but at the same time the long EURUSD position rises as well (which is good for the trader). The
two changes in value offset each other. If EURUSD falls, the value of the option falls – but, at the same time, the
long EURUSD position also falls. The two changes in value offset each other again. If the trader can keep the
portfolio in this state of immunisation from EURUSD movements, which is “a hedged portfolio”, then the value of
the portfolio does not change and stays at US$1,000. At the expiration of the option, the liability is gone and we are
left with the money in cash. That’s a lot of work for US$1,000. In the previous chapter, we showed how by delta
hedging we will accumulate or get rid of the hedge to end up with either 100% or zero of the underlying at
expiration. That is a consequence of our focus on keeping the portfolio in balance as the option and the proper
hedges dance around each other.
The problem with this theory is that it does not work in the long run. After some amount of time, the portfolio will
not be hedged. Why is that? Recall the effect of time on the delta, or theta. The delta of the option, the N(d1), will
have changed with time. As time ticks, the EURUSD exchange rate, the time to the expiry of the option, the interest
rate and the implied volatility all change. All the inputs in the Black–Scholes formula change. What can the trader
do? Very simple – as we have already seen, keep calculating the delta and when it is found to differ from the
forward amount on hand, execute another forward trade to adjust the aggregate forward position so that it is always
matching the total delta of the option.
This is great and relatively easy for one option, but what if they have a portfolio of options? Surely you cannot
reasonably assume anyone can do this instantaneously for 100 or 1,000 options? Here is where things become easy
for the portfolio. Although they may have 1,000 options in their book – some long (bought), some short (sold), some
short term and some long term – the greeks can be simply summed up! This is huge. If for a 1,000 options the delta
is to be long US$20 million versus the euro, if the euro moves simply add up all the deltas again and transact the
difference just like we did for one option. It is the same for theta. If I am long 500 options and the total time decay
for one day is –US$100,000, but I notice that the total time decay for all my short options is +US$250,000, I know
that the portfolio will gain US$150,000 tomorrow if nothing else changes.
How realistic is it to assume that one can keep all this in balance to earn the initial spread? Specifically, can the
trader:

execute an infinite number of rebalancing trades at every split second?

execute all such trades at the mid-market rate?
have a God-given knowledge of the instantaneous volatility of the EURUSD exchange rate and such a number
being constant over time? and
have the realised volatility number as exactly the same number that was used as implied volatility in the Black–
Scholes formula?

Very unlikely – it would be a miracle! Once again we find ourselves out over our skis. Let’s backtrack and get to the
answer slowly.

DELTA HEDGING AS A MECHANISM TO LOSE MONEY

Of the US$41,000 that the trader originally received for selling the option, at expiry only US$1,000 will be left,
which exactly equals the mid-to-ask spread on the transaction. Stated differently, the transactions to delta hedge
during the life of the option will cost US$40,000. Therefore, under the above conditions the mark-to-market value of
the portfolio during the entire life of the option would have been constantly equal to US$1,000. You might be
puzzled by the statement that delta hedging would have generated an aggregate loss of US$40,000. After all, it is not
at all obvious why there should be a loss. EURUSD can gyrate along a multitude of different paths that we cannot
predict. Some of those paths must lead to a gain.
You most likely suspected that pocketing US$41,000 was not going to be so easy. In some final scenarios, they
would pay nothing, in other scenarios a million dollars. It’s only the sum of the weighted probabilities of all the
scenarios that results in the expectation of paying US$40,000. The trader can choose instead to eliminate the risk of
this mad passive gambling and carry out dynamic delta hedging, which we’ve briefly described in previous chapters.
That activity will change the risk profile of the trading book, but will not change the expected P&L, which stays at a
loss of US$40,000 at the expiry of the option. Recall that each of the hedging forward trades are performed at
market levels, and therefore each of them has a P&L expectation of zero, so they don’t muddle the P&L picture at
the time of their execution.
How is it that the hedging activity turns out to generate sure losses amounting to US$40,000? As we’ve
demonstrated in a few cases already, this is due to the way delta behaves. As mentioned above, when EURUSD rises
the option becomes more valuable, so its probability of exercise rises, which means its delta rises and the trader will
have to adjust their hedges by buying a few more euros. When EURUSD falls, the same but opposite reasoning will
tell you that the trader will have to sell a little of the euros that were bought previously. This does not sound like a
great investing strategy, buy high and sell low. Recall our discussion from the previous chapter – that if they are
short the option, losses are generated by the hedging activity.2 If they are long the option, gains are generated with
every trade.
By now you will have understood the main and most fundamental difference between cash market and option
markets from a risk manager’s perspective. In a cash market, a trader is making a price and after the client trades the
trader will have a position, short or long, which gives the trader risk. The trader can turn around and execute what is
known as a “back-to-back” trade, which consists in unwinding the same position with another bank. In so doing, the
trader is lucky to break even, which means that they will sell at the ask and will buy back from another bank at the
ask as well. Net net – no money is lost or made. What the trader is really trying to do is to sell when they think the
rate will go down in the next few seconds, or minutes or hours, and to buy when they think the rate will go up in the
next few seconds, or minutes or hours. Therefore, when a spot trader gets a risky position from a client, they have to
somehow eliminate the risk by purchasing back from the market what they have sold to the client. We know that
there is only one guaranteed way to eliminate the risk completely: buy back the exact position you sold. How, when
and from whom is the reason why the cash trader makes a buck or not.
This is tough to impossible in the option market. Options are less liquid, and so if the trader just tries to buy back
the option that they have just sold to the client, the other banks will show very wide spreads and the trader will end
up losing money. The trader can also wait until another client of the bank comes around who wants to sell exactly
the same option that the first client has bought. That does not happen very often because cash is just cash, but
options have strike, notional, expiry date, and it is therefore very unlikely that such a saving grace will materialise.
Therefore, what is the option trader to do?
We have gone to great lengths to explain the concept of delta in the above pages, and here is where this concept
pays out. Suppose that a trader has sold an option and then has hedged its delta. Now, the portfolio (the trading
book) has the same value even if EURUSD moves slightly. The trader has to continue recalculating the delta to
make sure the forward position matches the delta on the sold option, but it’s doable. Suppose, now, that another
client comes along and wants to buy a different option. The client sells the second option and will now continue the
calculations but aggregate (literally, sum together) the delta of the first and second option, and will trade the
forwards according to the aggregate. This “portfolio approach” frees the trader from the challenge of buying back
from the market exactly the same instrument the client bought.
Here is the next problem the trader will face. At times the value of the portfolio will still move, possibly against
the trader. After a thorough investigation, the trader will discover that they are short two options and assume implied
volatility increased. This caused the value of both options (liabilities for the trader) to increase. On the other hand,
the value of the forward contracts they are holding in the trading book as delta hedges will not have been affected
because forward contracts are not options, and their value does not change at all when implied volatility changes. So
the trader needs a new solution.
The trader will calculate how much they lose every time implied volatility rises by 1%. As you now know, this is
called the vega of the trading book. Suppose vega is a negative US$15,000, meaning that if implied volatility rises
1% the portfolio will lose US$15,000. To eliminate this newly discovered risk, the trader will purchase a straddle. A
straddle is an option strategy composed of a call option and a put option with the same strike. The strike is usually
such that the aggregate delta of the straddle combination is zero. The purchased call option with a strike close to the
forward rate will have a positive delta of about 50%, and the purchased put option with the same strike will have a
negative delta of about 50%,3 equal in magnitude to the delta of the call option. The trader will buy options with
notionals so that the vega of the straddle equals US$15,000. Adding the straddle to the portfolio does not change the
delta, but the aggregate vega will now be zero.
Therefore, if implied volatility rises or falls then the value of the portfolio is unchanged. Of course, just as before,
if spot rises or falls, for small increments, the value of the portfolio does not change either. So, for a brief moment in
time and spot movements, we’ve neutralised the portfolio to movements of spot and volatility. However, just as we
need to trade to account for delta changes, as time goes by, as spot moves, as interest rate differentials change, the
vega balance of the aggregate portfolio will also change. Therefore, in theory the trader will have to continuously
sell or buy straddles to ensure that the vega stays at zero. It is getting pretty intense at this point to monitor all these
moving parts. Let’s add some specifics with an exercise.

EXERCISE 10.1

A hedge fund trader is bullish EURUSD (thinks EURUSD will rise), but they are bearish volatility (thinks
the EURUSD-implied volatility will fall). If this sounds odd to you, that a volatility in a currency pair will
actually cause expectations of movements to go down, think of the following situation. The EURUSD move
higher brings the EURUSD rate out of a technical extremity where the market and policy makers expressed
concern, back into a range that most strategists and traders believe is a sustainable rate for a long time.
Which of the four trades below should the client enter into?

buy a EUR call/USD put;

buy a EUR put/USD call;
sell a EUR call/USD put; or
sell a EUR put/USD call

Solution to Exercise 10.1:

The answer is to sell a EUR put/USD call. In fact, as spot rises the put option will go OTM (the probability
that it will be exercised will decrease), thus making its value lower. The option is a liability for the trader
who sold it, so the position increases in value. If implied volatility decreases, the value of any vanilla option
decreases as well. Think of it this way. The trader sold the option and received US$X. As spot moved higher,
they could buy it back for something less than US$X. As volatility moves lower, there is an additional
reduction in price.

EVERY OPTION HAS ITS OWN VOLATILITY: THE VOL SMILE

On October 19, 1987, the US stock market crashed and fell 22% in one day (see Figure 10.1). Of course,
overvaluation and mob panic were part of the causes of the crash, but programme trading was also blamed. Legend
has it that the crash was the first time traders noticed that OTM put options (the strike was below the current market
value) were much more in demand than the corresponding call options. The reason is driven by supply and demand,
but it is also eminently psychological and very subtle. This contributed to a fundamental change in how the Black–
Scholes formula is applied.
With the advent of behavioural economics, much emphasis has been placed on the well-known fact that humans
suffer much more intensely from financial losses than they enjoy financial gains. In other words, the pain felt for the
loss of one dollar is much more intense than the joy felt for the gain of one dollar. In 1987, this caused market
operators to be much more concerned about the possibility that the equity market would plunge further, and
therefore to be willing to pay very dearly for put options. Looking at the option market, you would calculate the
implied volatility that caused the Black–Scholes formula to match the price of ATM options (those whose strike is
equal to current market level), then you would use the same implied volatility to price an OTM put option and the
price would not match the market price for that option. It would invariably be too low, because the puts were so
much in demand!
 The efforts of Bruno Dupire, Emanuel Derman, Iraj Kani and other legendary quants allowed the development of
a new model to price options where the implied volatility used in the Black–Scholes formula is different for each
strike. The “local volatility” model was born. Of course, whereas before you just needed to know that implied
volatility equals 10% to price all options, now you need to look at the strike first and then remember when was the
last time that anybody traded that strike and what volatility was used. This is not easy.
The way market convention developed in FX was to actually make the volatility depend on the delta of the option
instead of on the strike of the option. We will skip some of the more technical details on this last point, but suffice to
say that currently the interbank market regularly quotes 50 delta volatility, ATM, and two OTM strikes, 25 delta
puts, 25 delta calls, and then 10 delta puts and 10 delta calls. In total, if you call an interbank options broker they
should be able to quote you those five volatilities for each currency pair and each standard tenor. What happens if
the option you are pricing is 37.5 delta? Then you will take for implied volatility the average of the ATM and the 25
delta. In other words, for any delta level that is not one of those five pillars that are regularly quoted, you will use an
interpolation (it is a matter of hot dispute which interpolation technique is the best one, but that is another story). For
deltas smaller than 10, you will use some sophisticated extrapolation methodology not discussed in this book.
It took a while for this revelation to sink into the entire FX market mindset. Although the traders felt in their
bones that you did not use the same volatility across all strikes, risk officers, for example, kept using one vol across
all options. The concept that you need accurate volatility for every strike (volatility surfaces) to properly mark-to-
market portfolios took years to be implemented. Many times there was frantic activity in OTM options at month-end
to create artificial P&L, and very often at great risk to future earnings. For example, suppose month-end was
approaching and you had a terrible month. You needed a silver bullet to boost earnings. Well, if your USDJPY book
was being marked at 10% but the 25 delta options were 12% and the 10 delta options at 14%, you sold a ton of both
of those and “realised” the difference as profit this month. This happened because the back office would value all
options at 10%, even the ones you had sold at 12% or 14%. However, you just ended up with a large number of
short options and potentially lots of negative gamma that will kill you if spot went crazy. The opposite activity
would ensue if you had a great month and wanted to put some money aside for the following month.
A chart showing the implied volatility for each delta of options for a particular maturity, going from very small
delta puts to very small delta calls and showing the ATM strike in the middle is commonly called “the volatility
smile”. If you wondered how implied volatility depends on the delta of options, a good look at such a chart should
answer that question. This volatility smile in Figure 10.2 shows the implied volatility used to price options with their
corresponding delta shown on the x-axis.
All these adjustments to the Black–Scholes model show how the model is flawed, or at least incomplete. It does
not make things any easier in pricing options, but it does practically account for the fundamental human intolerance
to pain. Come to think of it, why is it that the pain happens when the equity market goes down and not up? After all,
for the people who shorted the market the pain is on the way up, so for them the protection is to buy calls. In the
equity markets specifically, puts are more expensive! The main reason is that most cases revolve around supply and
demand. Investors in a stock can earn “excess” profit by selling (writing) options for OTM calls. This is called
“covered call writing”, and we will explore something similar in the FX market in a later chapter. So, the extra
supply of people selling calls, and not selling but buying puts, creates this volatility smile in equities. For now,
however, let’s explore something more nuanced.

Another way to look at the same issue is quite subtle and has to do with history. The economy can be seen as the
ability of people in aggregate to forecast what will happen when you launch a business. Everybody is able to write a
business plan and get a loan from a bank to launch the business, but then the execution is the difficult part. The
business plan will lay out how things are supposed to work so that the business can make money and the loan can be
repaid. If everything goes without surprises, then the business will be successful, the bank will make a small profit
on the loan and the business owners will make a living, or become wealthier. However, if things go wrong they
could go very wrong, the business will flounder, the loan will not be repaid, and the bank will lose the whole
notional on the loan and the business owners will lose their shirt.
In other words, things can turn out well, we can have small positive surprises, or we can have huge negative
surprises. It is very rare that a business plan makes double the expected money, but very frequently a business plan
makes no money at all. Good news comes in small increments but bad news comes in truckloads at a time. It is life.
This is wholly reflected in the stock market, and in any market for that matter. We might find it difficult to model it
mathematically, but people have it in their blood. This is why stock prices can crumble, tank, plummet, crater,
slump, collapse, but they can only inch up, climb a little every day, grind higher. Think of the business news media’s
love of making statements such as: “This drop in the stock market erased US$X trillion of market capitalisation”. I
do not recall any reporter saying: “today’s up moves added so many trillions to the stock market.” Consequently, put
options on equity markets are more expensive than call options.
Follow this line of thinking into the emerging markets FX world. In any market you have to ask yourself, where is
the risk side? Look at USDBRL, the exchange rate expressing the number of reals that buys one US dollar. If the
Brazilian real strengthens, USDBRL goes down. This is a market where investors buy the Brazilian real to put
money in a developing country with potentially higher GDP growth. This may turn out to be the case, but it will not
happen overnight. It’s unlikely you will open up the paper tomorrow morning and read that, against all expectations,
something massively positive has happened in Brazil and therefore the GDP has increased by 10% from
expectations and the FX rate has reacted accordingly. On the other hand, something can – and often does – go
wrong, and the real plummets. History has demonstrated this repeatedly in dramatic fashion. The stable, more
predictable, side of the trade is the US side. Most of the investment goes from the US to Brazil, not from Brazil to
the US, and any bad news about the US dollar is likely to only generate a small fall in USDBRL. In addition to the
in-country investment of US dollar-, euro- and yen-based investors, many Brazilian entities look to borrow at lower
dollar rates, so many of their loans are dollar-denominated. At the first sign of trouble in the currency, these entities
must rush in to buy dollars and sell reals to meet their interest and principal payments, exacerbating the crisis.
Therefore, it’s no surprise that USDBRL calls are more expensive than USDBRL puts – or, more precisely, US$
calls R$ puts are more expensive than US$ puts R$ calls of similar deltas.

Figure 10.3 shows the three-month USDBRL risk reversal. This is the difference between the implied volatility
quoted for 25 delta US$ calls R$ puts and the implied volatility quoted for 25 delta US$ puts R$ calls. This is also a
proxy for the risk perceived by the market. As you can see in the figure, risk peaked during the great financial crisis
of 2008, during the sovereign credit crisis of 2011 and the spike in 2017 that corresponds to legal troubles for the
Brazilian government. Note that it has never gone even close to zero, let alone negative. When we compare US$
puts with US$ calls, the calls’ implied volatility is always higher.
On the other hand, let’s look at USDJPY. Interest rates in Japan are typically much lower than in the US, which is
a reason for many Japanese investors to pour money in US dollar-denominated assets. In addition, the savings rate in
Japan is much higher historically than in the US, so institutional investors (pensions and mutual funds) in Japan have
a lot of money to invest. Finally, to compound the effect, the Japanese economy is much more export-oriented than
the US economy. Exporters who sell their wares in the US collect dollars that will eventually be sold for yen to
repatriate revenues. The preferred hedging strategy for these exporters is to purchase OTM, US$ puts ¥ calls, giving
them the right to buy yen at a minimum, predetermined rate. As they are corporations and have a hard time
explaining to management why they paid any money for a premium, they sell OTM US$ calls ¥ puts to take in some
premium, and usually put on a hedge for no upfront premium. This combination of buying yen calls and selling yen
puts is also called, by an abuse of language, a risk reversal. We will talk extensively about this hedging strategy in
Part 3, but for now suffice to say that there is a huge demand for US dollar puts yen calls and an abundance of US
dollar calls yen puts.
Figure 10.4 shows the three-month USDJPY risk reversal, which is almost always negative. The US$ puts are
higher than US$ calls. This is the difference between the implied volatility quoted for 25 delta US$ calls ¥ puts and
the implied volatility quoted for 25 delta US$ puts ¥ calls. This is also a proxy for the risk perceived by the market,
but this time the lower the risk reversal quote the higher the market risk. Note the spike related to the Japanese
earthquake of March 2011.
 In summary, for a variety of often unrelated factors there are very important differences between the USDJPY and
the USDBRL markets. In USDBRL, US$ calls cost more, but, in USDJPY, US$ puts cost more. This imbalance,
which gave rise to the “local volatility” option pricing model, is captured by the difference between the implied
volatility used to price 25 delta calls and the implied volatility used to price 25 delta puts. This spread is called the
risk reversal, and will give you an idea of the risk underlying any specific option market.

Risk reversals can jump dramatically in response to geopolitical events. Countries where the currency is free
floating, where there is a lot of foreign investment, where interest rates are high and FX reserves are low, are
candidates for panicked waves of investors that are more likely to pull their money out at the first whiff of bad news.
Those countries will have very pronounced risk reversals. Countries similar to the US, with stable and open
economies and a similar level of interest rates, will have a risk reversal close to zero.

RISK REVERSALS AND THEIR TRUE IMPLICATION

It is correct to think that the risk reversal is in some sense a proxy for the risk in a market, just as is the absolute
level of ATM-implied volatility. Plot the USDBRL spot, one-year ATM-implied volatility and one-year 25 delta risk
reversal. All express the level of fear in the market, and will move with a high degree of correlation. When bad news
about the Brazilian economy spreads, investors pull money out, they sell reals and buy US dollars, so USDBRL
rises. If they want to stay but do not want unlimited risk, investors buy ATM US$ calls R$ puts to protect those
investments. As demand for options rises, traders will respond by raising their prices. If the price of the same option
is higher but nothing has changed in spot or interest rates or time, the only thing left to do is say implied volatility
has risen. This is a chicken-and-egg thing. The traders may not believe in their heart that things will get dicier and
therefore that future expectations of volatility will go higher. They may simply look at all the demand and say, “I
just don’t want all this risk on my books”, or “If I will have all this risk I want more money for it.” It really sounds
so much nicer to say the price is higher because implied volatility is higher. This reminds me of airlines when they
say: “We are experiencing an overbooked condition.” As if the condition arose by deus ex machina. No, you have an
overbooked condition because you overbooked the plane.
Along with purchases of ATM options, many investors may think this is not a really bad situation – I do not want
to spend all that money for an ATM option, I just want some protection in case things go really, really badly. They
will choose not to buy ATM because they are expensive, and may be able to stomach USDBRL a little higher before
they start to panic. Those will buy 25 delta US$ call R$ puts instead, and in so doing will cause the risk reversal to
rise as well – meaning the greater demand for OTM US$ calls will cause market makers to charge more.
Figure 10.6 shows the USDBRL spot exchange rate, the USDBRL one-year implied volatility and the USDBRL
one-year 25 delta risk reversal. All three are behaving similarly, reflecting a changing perception of risk at different
times.
There is a paradox here. Many commentators like to say that a high risk reversal indicates that the market expects
the exchange rate to rise above the forward rate. Many business news outlets like to also say that if the risk reversal
is biased this way or that, it reflects which way the market is expected to go. Both of these are wrong. In the above
simple situation, one can see intuitively that investors may be buying US$ calls not because they think it will go
higher, but just in case. In fact, those buying OTM US$ calls think the opposite, they just can’t stomach unlimited
risk if they are wrong.
 Let’s make a more rigorous defence of our above statements. The mathematical model for the FX market has the
forward rate as the unbiased market-implied expectation of the exchange rate in the future. This means that the
expected P&L of a forward trade executed today is zero. Hopefully, 10 chapters on, you are convinced that logically
the market cannot imply any expectation different from the forward rate.
Nevertheless, many are not convinced and, although they are wrong, we will try to explain their argument. They
say that higher prices for puts imply more people buy puts than buy calls, and certainly this must mean that more
people think the rate will go down than think the rate will go up. They are wrong for a simple reason. If one is
bearish, one may buy a put: that is true. However, suppose that you are bullish and you went long by purchasing a
forward contract. Of course you have a risk limit, and you realise that if the rate falls 3% then you will hit your risk
limit. So, what you do is buy a put struck 3% OTM. Does the fact that you bought a put mean necessarily that you
are bearish? No, it does not. Therefore, the level of the risk reversal does not give you any information on whether
market participants are bullish or bearish.
Another way people create an erroneous mathematical proof is the following. If the risk reversal is negative, it
means that the purchase of a 25 delta put and sell of a 25 delta call will net cost money, as the put volatility is higher
(in this example it is crucial that deltas be exactly 25%). The reason the trade costs money, they argue, is that this
trade has a positive expectation in terms of P&L. This must mean that future spot is more likely to be below the
current forward level than it is to be above it. This argument is also wrong. It is true that the trade has a positive
P&L expectation. If spot rises, the trade loses US$X and the sold call option will potentially be exercised. If spot
falls, the trade makes US$Y and the bought put option will potentially be exercised. It may be that US$X is smaller
than US$Y, but it is not true that US$X is less likely than US$Y.
To see the fault in the argument in yet another way, imagine that you bought EURUSD via a forward contract,
sold a 25 delta call and bought a 25 delta put.4 Then, on an aggregate basis, you would have to pay to enter into this
three-postions position, but this time you are bullish, not bearish. In other words, the fact that you have to pay to
enter into a bullish position does not mean that the market is bullish.
What is it then the risk reversal tells us? How can we interpret the risk reversal in common language? From a
mathematical point of view, the risk reversal tells us that when spot moves a lot (a lot defined as an amount that
causes those OTM puts to become ITM, so that the current spot is now below the strike of these puts), then the
mathematical model requires that volatility will increase. The risk reversal tells you by how much. The implied
volatility for 25 delta puts represents approximately how the volatility level will rise if spot moves down to the strike
on that option. That is a big “if”. The risk reversal does not tell you that spot will move down – it tells you what
volatility will do “if” spot moves down.

RISK REVERSALS, BUTTERFLIES AND PORTFOLIO MANAGEMENT

Now that we have explained the definition and implications of the risk reversal, we will return to the main reason
why we have decided to talk about it: portfolio management. Recall our trader, who is busy calculating delta and
vega in the portfolio and adjusting the forward and straddle hedges so that the portfolio value is not changing when
spot changes or when ATM implied volatility changes. Assume that the trader is short an ATM call option, plus long
a 25 delta call and short a 25 delta put. This is a portfolio with three options. Also assume that delta and vega are
perfectly hedged. Then something unexpected happens, causing the risk reversal to fall by 1%. This means that the
spread between the implied volatility of a 25 delta call and a 25 delta put decreases by 1%. What happens to the
portfolio in this case? Nothing else moved, so the value of the straddle and the forward are unchanged, and the value
of the ATM call option is unchanged. Only the value of the two 25 delta options has changed.
We do not know if the volatility of 25 delta calls falls or if the volatility of 25 delta puts rises, or if it is a
combination of the two, we only know that the spread fell by 1%. This has real-world consequences for our trader. It
may be that net value of the combination of the two 25 delta options falls and the trader experiences a loss of P&L.
They also realise that their hedging activity, the act of executing trades for the sole purpose of immunising the P&L
of the trading book, needs to include the risk reversal as one of the hedged variables, together with delta and vega.
Just when you thought you were getting a handle on all the moving parts.
In the above example, if the trader wanted to not have this risk, they should have sold the 25 delta call option and
bought back the 25 delta put. Doing this would have no effect (or very marginal effect) on their vega position
because the sensitivity to the absolute level of implied volatility is roughly the same for those two options. The delta
of the trading book would become negative, which would require an additional forward trade to adjust the delta
hedge and make sure it reverts to zero. After this, the trading book would be hedged for delta, vega and the risk
reversal.
The above is a very simple example, where it is obvious that if you get rid of the two OTM options the problem
goes away. In more realistic situations, the trader has dozens and dozens of options in the trading book. After
hedging delta and vega, the trader runs some simulations. They hypothetically change the risk reversal and revalue
the trading book, calculating how much money they will lose or make if the risk reversal falls by 1%. This amount is
sometimes called “d-vega d-spot” (this is one of the second-order components in the Taylor formula that we showed
in Chapter 9), and represents the change in portfolio value due to two infinitesimal changes, one in spot level and
one in volatility level, but do not worry about that just now.
After having calculated d-vega d-spot, the trader knows that they are facing danger if the risk reversal moves. The
trader then proceeds to trade in the interbank market, selling the 25 delta call option and buying the 25 delta put
option. As mentioned earlier in this chapter, this combination of option trades is also called a “risk reversal”, or a
“collar”. The notional on the two options is calculated in a way such that if the risk reversal falls by 1%, this
combination will have a positive P&L equal to the d-vega d-vol just calculated for the existing trading book. The
addition of the collar to the trading book, by design, will result in the overall portfolio having zero P&L
consequences if the risk reversal falls by 1%.
However, as with any other carefully balanced situation, after adding the risk reversal you’ve now unbalanced the
delta hedge and the vega, and they are now different from zero. Don’t fret, it turns out that the purchased collar has
roughly no vega. The vega on the sold call is equal and opposite to the vega of the bought put. Although the delta
will be impacted by the collar, that has a simple solution: execute a forward trade to make sure the delta is now zero
again. The forward trade has no impact on the d-vega d-spot or the vega of the trading book. In the above example
the trader had a perfect 25 delta collar position, but the above is also valid if the trader only has a 35 delta put option
position, or any other combination of options.
Suppose now that the trader is long a 10 delta call option and short a 10 delta put option. Nothing in the market
moves, but the implied volatility for the 10 delta call option falls by 1%. The P&L of the trading book falls, but the
trader would see minimal to no change in delta, in vega and no change in the 25 delta d-vega d-spot calculation. This
could be potentially problematic in a trading book with hundreds of options. Traders also look at the 10 delta risk
reversal, defined as the difference between the implied volatility of a 10 delta call minus the implied volatility of a
10 delta put. This is yet another set of dials on the trader’s dashboard. So far we have delta, vega, d-vega d-spot for
25 delta and d-vega d-spot for 10 delta, theta, rho and psi.
Let’s keep climbing. Suppose that suddenly the 25 delta put-implied volatility drops by 1% at the same time as
the 25 delta call-implied volatility drops by 1% (the ATM volatility is unchanged). In such a scenario, the 25 and 10
delta risk reversal would not change but the trader will likely experience a change in the value of their book, even if
they are delta-, vega- and d-vega-dspot-hedged. The thing to do is to create yet another measure, another dashboard
dial, called the butterfly. The butterfly will detect these tricky changes in the volatility smile. The definition of the
25 delta butterfly is the average of the implied volatility of a 25 delta put and a 25 delta call, analogously to the 10
delta butterfly. The butterfly is also called the d-vega d-vol. It is the second-order derivative of the book value with
respect to changes in implied volatility (another second-order component in the Taylor formula).
 Figure 10.7 shows the 25 delta one-year butterfly in EURUSD. Compare this chart to the risk reversal chart and
you’ll note that the butterfly does not vary as much as the risk reversal over time.

EXERCISE 10.2

How can you compute the expected absolute value of the change in the EURUSD exchange rate over the
next month? Specifically, what is the expectation of |EURUSDf – EURUSD0|?

Solution to Exercise 10.2:

The answer is the price of a one-month EURUSD straddle (buy a EUR call/USD put with strike equal to spot
and also buy a EUR put/USD call with strike equal to spot) expressed in pips, or units of exchange rate. In
fact, if EURUSD rises then the EUR call you bought will pay you the absolute move of EURUSD and the
EUR put will pay you zero; if EURUSD falls, then the EUR put you bought will pay you the absolute move
of EURUSD and the EUR call will pay you zero. Either way, in aggregate you will be paid the absolute
move in EURUSD. Just keep in mind that if the cost of the straddle is expressed in percentage terms, then
you need to translate that into units of exchange rate. So, if EURUSD is trading at 1.10 and the cost is 10%
(that would be a very expensive one-month straddle), then the cost expressed in units of exchange rate is 1.10
× 10% = 0.11.

All the measures we reviewed help traders manage the risk in their trading portfolio. It is hopefully clear by now
that there are many tools that are collectively called the greeks. However, the importance of the newer ones is less
that those discussed in Chapter 9. Delta and vega, for example, correspond to the first-order coefficients in the
Taylor expansion of the value of the trading book, for those who remember high school maths. The risk reversal and
the butterfly would be second-order coefficients, and therefore their change would have a lesser impact than a
change in delta or vega. The real point in managing all those greeks is that if your trading book has hundreds of
options, you can manage the greeks and the value of the trading book will not fluctuate uncontrollably. They are an
incredibly powerful tool, as long as the trader keeps the greeks close to zero.
As a final comment, we will circle back to the beginning of this chapter, and recall the comparison between
trading spot and trading options. Recall that spot traders can easily balance their books. They can decide how and
when to do it, and for how long to keep open positions. Now you can appreciate that option traders do not have that
luxury. There are so many parameters in any option portfolio that it is very unlikely that the trader will find the other
side to unwind all their positions. Even if the trader decides that for each option sold they will try to buy it back
from the interbank market, the other banks will soon enough know that the trader is desperate to offload the risk and
their spreads will widen. This is a recipe for losses.
Option traders usually wait much longer, and only think of their greeks: they will never say that they are short
such and such option. They will say that they are short delta, short vega and short gamma. Typically, in the
interbank market they will trade ATM straddles, 25 delta and 10 delta collars, and 25 delta and 10 delta butterflies
so as not to show their hand. Here again the language fails us. A butterfly in this context is a combination of options
consisting of selling a straddle, buying a 25 delta call and buying a 25 delta put, not related to the relationship of
equidistant puts and calls. Look, if there was no confusing terminology there would be no job security.

EXERCISE 10.3

A trader is given quotes from the interbank broker for the ATM volatility, 25 delta RR and 25 delta butterfly.
How can the trader calculate the 25 delta call implied volatility?

Solution to Exercise 10.3

25 delta call-implied volatility = ATM + 0.5 × RR + butterfly.

Trading in options is intrinsically different from trading cash (or spot). Option market-maker traders are obliged
to manage a portfolio, and are also obliged to manage their greeks collectively and not every single position
independently. They also will tell you time and time again that they cannot be completely flat, meaning that it is a
losing battle to try to have a trading book where all the greeks are zero.
I was once chief option dealer for a major non-US bank that wanted, in theory, to be a big player in these markets.
However, they could not stomach the spectrum of uncertainty that encompasses any complex option book. The head
of all trading in New York was a wonderful man who just wanted consistent daily profit and really hated down days.
It used to drive him nuts when I would come in some mornings and the portfolio had lost money overnight. He
would say to me: “Why did we lose money? You told me that you were long Canada and USDCAD went up.” I
would explain that, yes, my delta was long USDCAD but USDCAD moved up a lot, and I was actually short gamma
at those levels, so the delta would switch to short USDCAD. Not only that, but since USDCAD moved so much, vol
went up and I was short vega, so I lost on many fronts. He would just shake his head and tell me to be flat overnight
and take positions during the day. Something an option trader simply cannot do!

EXERCISE 10.4: SIEGEL’S PARADOX

Siegel’s Paradox is a true FX classic. A European trader, let’s call him Hans, trades with a US trader, let’s
call him Hank, a very special type of financial contract that Hans just made up. The EURUSD spot rate is
1.00 and the one-year forward rate is also 1.00, to simplify the exercise. The financial bet is structured in the
following way. If in one-year’s time EURUSD is below 1.00, then Hans receives US$1 million from Hank;
above 1.00, Hans pays Hank €1 million.
In this exercise you are Hans. You reason that your expected payout is positive. In fact you are right – a
50% chance for you to pay €1 million, and a 50% chance to receive US$1 million. However, hold on, by
definition in every scenario where you receive US$1 million that amount is greater than €1 million, because
you receive it only if EURUSD is below 1.00. This means that your expected payout is greater than zero.
Therefore, you conclude that this financial bet must be unfavourable for Hank. Not so, claims Hank.
Flipping the argument upside down and reasoning that, for Hank, the expected payout is also greater than
zero – a 50% chance of having to pay US$1 million and a 50% chance of receiving €1 million. But each time
Hank has to pay, the value of the dollars he has to pay is lower than €1 million as the payout only happens if
EURUSD is lower than 1.00.
 Solution to Exercise 10.4:
So, who is right? Or maybe we have just discovered a little machine to make money? After the discovery that
crystal balls do exist, this would not be such a great surprise. This is the greatest and the most famous of FX
paradoxes, and it has the good taste of not having an equity version. Table 10.1 shows all the possible
outcomes.

As long as Hans expects to make money in euros and Hank expects to make money in US dollars, or in other
words as long as Hans measures his wealth in euros and Hank measures his wealth in US dollars, then they
are both happy. As soon as you want to compare apples to apples, the free money disappears into thin air.5
Similar cognitive biases are used in creating some interesting investment vehicles, which we will not
cover in this book. They offer what appears to be low- or no-risk investment with high returns. These are
particularly popular in private wealth management where individuals are more focused on the currency they
tend to consume. A typical condition will have something like: “You invest US$1 million and your principle
is guaranteed, but, if some condition materialises, the bank has the right to return to you either US dollars or
euros as they choose.”

CONCLUSION
Chapter 10 expanded on the characteristics of the first-order derivatives reviewed in Chapter 9. This was done in the
context of a portfolio, so that we could drive home the interdependency of the greeks. Then we examined a couple
more second-order greeks, such as the risk reversal and the butterfly. These terms are used to refer to two different
situations in the option market, and can be very confusing.
A risk reversal is a number that defines the relationship between equal delta put and calls, and is the way we
define risk reversal when talking about risk. It is also used to describe a combination of actual trades, specifically
when one buys a call and sells a put at the same time. That is also called a risk reversal. In hedging circles, this
combination of options is also called a collar. We will see why in Part III of this book.
Similarly, when describing the risk of the portfolio, another second-order derivative is called the butterfly. This is
the average between certain delta (usually 25 or 10) puts and calls. Butterfly is also used to describe the purchase of
two ATM options, a put and a call, and the sale of an OTM put and an OTM call. Actually, there are other ways to
construct butterflies but this will suffice for now. I suspect option traders drew a graph of what this combination of
trades looks like and decided it looked like a butterfly. To economise on words, they kept the name.
The discussion in this chapter was more theoretical than practical. The next chapter will take these concepts and
apply them to the specifics, and show the kind of systems real-world portfolio managers use to control their
portfolio.

1 FX options that go out to five years and beyond are significantly impacted upon by interest rates, and their liquidity is constrained so that it becomes
more of an interest rate play than an FX volatility play.
2 A separate issue is that delta-hedging trades are not transacted at mid-market, so the bid/ask spread also contributes to the losses for the delta-hedging
trader.
3 Just a friendly reminder that put options have negative deltas, even if in the previous pages we have sometimes ignored the sign.
4 More on this combination in Part III.
5 A very similar paradox consists in noting that if Fwd is the forward rate in EURUSD then 1/Fwd is the forward rate in USDEUR. This is all very
logical and good. Nonetheless: the median of the lognormal probability distribution corresponds to a return of μ = rd – rf – \ σ2/2. This unluckily does
not flip as easily, in the sense that the median of USDEUR is not the reciprocal of the median of USDEUR … but how is it possible? The rate that
splits 50%/50% all possible future scenarios for EURUSD must also split 50%/50% all the possible scenarios for USDEUR, right? The solution to this
version of the paradox is that if you look at USDEUR you are actually looking at a world where we measure wealth in euros, and in this world the
probabilities are different from those in the world where we measure wealth in dollars.
 11
FX Options Trading Book & Risk Measurement

Now that we have talked about the theoretical underpinnings of a portfolio approach to managing an FX options
trading book, let’s put it all together and, as you may be used to by now, kick it up a notch to the next level of
complexity. The first sections of this chapter will repeat elements of Chapter 9 but from a more practical
perspective. We look at a theoretical reaction of a trader who sells an option to a client the day before the French
elections, and their subsequent adventure in trying not to lose money. Then we combine those elements with those
of Chapter 10, and look at some new concepts by the end of the chapter. Specifically, we begin by reviewing the
basics of delta hedging, which you’ve seen in a few different contexts, and then use real-world examples to discuss
gamma, vega bucketing, value-at-risk and portfolio stress testing.

DELTA, GAMMA AND VEGA CONSIDERATIONS IN HEDGING THE FRENCH

ELECTIONS
Let’s start with an example of historical significance. Suppose that a bank trader just sold an option to a client on the
eve of the first round of the 2017 French presidential election. The following figures are actual screens and
information a trader would be looking at.
Figure 11.1, Figure 11.2 and Figure 11.3 show an overview of option details. The tables below are samples of the
type of screens that include all the relevant details like: expiry, strike, P&L on the trade, which should be zero when
the trade is initiated, the value of the option in US dollars, the premium of the option (when the option was traded),
the fact that the option is a put on the euro, and therefore a call on the US dollars, the notional in euros, the delta, the
gamma, the theta and the vega of the option, together with the P&L of the trade in euros.
 Notice that Figure 11.3 also includes the delta hedge which we will show below.
Figure 11.4 shows how the option looked at the time of trading, with all the details and the price date and time.
For added drama, we assume that this option was traded OTC on the Saturday preceding the first round of the
French presidential election in 2017. Note the spot reference on top is 1.0729. Now relive the moment from the
trader’s perspective.

The bank client bought a six-month €1.0550 put US$ call with a notional of €10 million to protect against a
sudden decline in the euro in case of an election result that could spook the markets. Immediately after the sale, the
trader has a number of decisions to make very fast. The markets are nervous and moving. Assume that this is the
only position the trader has – which is, of course, a big simplification. The first problem for the trader is to quantify
the risks, since we know an option position is a composite instrument. In Figures 11.1–11.4, we can see the option
parameters and the different greeks that we described in the previous chapters. At the very moment the trade is
consummated, the trader has taken on the opposite risk profile to the client’s. This means that if the markets don’t
like the results of the election and the euro tanks, the trader will have a loss. Note the sign of the delta, which is
negative. We can also run a quick calculation called a “spot ladder”. A spot ladder is like a stethoscope to a GP. It is
the most fundamental tool for any option trader. You would never go home at night until you had a correct spot
ladder to understand how your delta changes as spot moves. Spot ladders are usually a table with the following
columns: spot percentage change; spot level; delta at that level; and P&L from when you closed the books.

In Figure 11.5, details of the option are on the top part of the screen, and the lower half chart shows the value of
the portfolio (composed of just one option) as the spot rate moves. The scenarios considered, start from a negative
move of 5%, and increase with a step of 0.5% until the highest scenario corresponds to a positive 5% move. The
percentages are shown in the horizontal axis. A fall of 5% in EURUSD will result in a loss of more than
US$200,000. A rise of 5% in spot, on the other hand, will result in a gain of just shy of US$100,000. The loss and
the gain are not symmetric as the position is not a spot or forward position but an option position.

Delta
Suppose the trader does not like this risk/reward profile. They may actually agree with the client, that there is
downside risk to the EURUSD. In general, good traders do not just take on the other side of whatever risk they are
randomly given and pray for positive results. They take the risk apart and bend it to reflect their risk/reward profile.
This trader decides to eliminate the spot/forward component of the risk by delta hedging the option. They enter into
a forward trade with the same notional as the delta of the option and in the opposite direction, which means the
trader sells euros.

Figure 11.6 shows the hedging trade. Note that the notional on the trade (in euros) is equal to the “hedge” amount
shown on the bottom line of one of the previous figures, and also note the premium is zero.
Now assume at a latter part of the day spot euro has moved higher, making our sold option less valuable: it is now
only a 25 delta. Keeping up with the changes the trader has reduced the delta to approximately 25%. The two tables
in Figure 11.7 show the portfolio after the delta hedge has been added. Note that the market value (MktVal) is not
zero for the forward trade because spot has moved since trading. The two delta figures are almost exactly equal and
opposite. Also, the gamma, theta and vega of the hedging trade are all zeros since the hedging trade is not an option
trade and has no convexity.
The first row in Figure 11.8 shows the aggregate portfolio with delta close to zero. At this point the trader can
take a breath because the delta is close to zero. Let us re-run the spot ladder just as before with the hedge included.
Figure 11.9 shows a spot ladder of the portfolio after delta hedging. Appreciate the beauty of this image. Now, for
small changes in the spot rate, the portfolio value is roughly unchanged. For a more extensive discussion on delta
hedging, go back to Chapter 9 and follow that example. In mathematical parlance, the derivative of the portfolio
value with respect to the spot rate is zero. The first and most immediate risk has successfully been temporarily
neutralised. They exchanged fast, unlimited loss in one direction for slower, unlimited risk in both directions. Still
not a situation they can ignore for long.
 This seems unfair to the trader. More to the point, why did they put themself in this position? Keep in mind that
the trader received a premium upfront as an advance payment for taking on the job of hedging. The terms of the
understanding are that, by hedging this trade during its life of six months, the trader will be compensated by the
client with the premium. Mathematical models tell us that if the trader is diligent during those six months and the
volatility turns out to be 9.75% (which is the implied volatility used to price this option, see above figures), then the
trader will just make the ask-mid spread at the time of the sale. The hedging activity will cost the trader an amount
equal to the premium received minus the ask-mid spread. It is the job of the trader to deviate from this mathematical
fiction and manage to lose less than the premium.
For example, a 3% move in EURUSD, either way, will generate a loss of about US$20,000. This is where the dry
mathematics goes out of the window and human nature takes the helm. This will be a game of chicken: the trader
versus the market. Maybe US$20,000 is too much of a risk – suppose their trading year has not been good and they
need to stop all losses, they may be desperate (by the way, the amounts we are dealing with here are really very
small for most of the big bank traders involved in this business). Alternatively, maybe they have had a great year
and they are now trying to protect the P&L and play it safe. Given that our scenario happens in April 2017, it’s
unlikely that they would be in a defensive mode this early in the year. The point is that the trader’s disposition
towards risk will impact on how their position is hedged.

Gamma
Another way to describe this particular situation, where any spot move will generate a loss, is that the trader is
short gamma. The gamma column is negative a half a million US dollars. This means that, if the spot rate moves by
1%, then the delta of the option changes by half a million dollars. Stated differently, for the portfolio to be delta-
neutral, again the trader needs to transact another forward for US$500,000. Figure 11.10 shows the relationship
between delta and spot.
For another more delta-focused example for how delta hedging works, look at the appropriately named section in
Chapter 9. We will review briefly here the dynamics and then make it more complicated.

Is it a good or a bad thing that delta changes? Why should we care? Well, as we showed during our theoretical
explanation in the preceding pages, it can be good or bad, but in this case it is bad for the trader. Let us see why. If
spot rises, then delta decreases; to remain delta-hedged, the trader will have to trim their delta hedge, which was
originally to sell euros. Trimming means buy a little bit of euros. If spot decreases, the situation is in reverse – the
delta increases, which means that the delta hedge amount on hand is not enough and needs to be increased a bit with
a further sale of euros. In essence, if spot rises the trader buys; if spot drops, the trader sells. Does this seem like a
good strategy to you? “Buy high, sell low” is a sure way to the poor house, a sure recipe for losses. However, there
is nothing the trader can do as long as the gamma is negative. So negative gamma by definition means every trade is
a losing trade.
The sure way out would be for the trader to buy a corresponding six-month option with equal and opposite
gamma, so the gamma of the portfolio sums to zero. The problem with that is that the premium of another six-month
option will likely be similar to the US$174,000 the trader received from the client. The trader would basically spend
it all to buy back the same, or a similar, option from the market – hardly a viable recipe for making money. But there
is hope. The trader can buy in the market an option with similar gamma but that costs much less. The trader knows
that options with a short tenor have very high gamma as long as the strike is close to the current market. Why is
that? Let’s try to figure that out.
In a six-month option, when spot moves the delta moves, but the moves are slow and smooth. How do things
change at the expiration of the option, at exactly 10am EST on the expiry day? Remember, at that point the value of
the option looks like that in Figure 11.11.
The value of the option is zero as long as spot is higher than 1.0550. For any spot reference higher than 1.0550,
the delta is zero, given that the option is reduced to nil. If spot is lower than 1.0550, then the option will be
exercised, which means that being short the option is like having a forward position to buy the euro at 1.0550, not a
great position when spot is, say, at parity. Therefore, just as for a forward trade, the delta of the option is equal to
100%, give or take the interest rate differential.

What about when spot is at 1.0550? There, the mathematical definition of the delta breaks down. To visualise this,
note that at a kink point a function does not have a defined “slope” and therefore there is no derivative. In essence, at
1.0549 the delta is 100%, at 1.0551 the delta is zero, so 1.0550 is the level where the change in delta is the most
extreme; as the change in delta is gamma, then gamma is the highest if the strike is close to the spot level and if the
expiry is very close. In addition, we know that the option price is an increasing function of time – when time is very
small, the price of the option must be very low. What a great piece of thinking. We just discovered that short-tenor
ATM options have stratospherically high gamma and are cheap.
The trader will then decide to splurge some money by buying a short-term option to reduce their gamma position.
The only problem if they buy another option is the delta will now be off by the amount of the delta of the newly
bought option. So, when the trader buys the option, they need to do a delta hedge as well so that the total delta will
continue to be close to zero. Although individual vanilla options have positive or negative delta, straddles do not.
Straddles are the simultaneous purchase of a put and a call with the same strike (usually ATM) and the same details
for everything else. In Part III of this book, we will examine some of the most basic combinations in the market and
talk about some of the characteristics of the greeks that make them special.
Figure 11.12 shows a three-day ATM EURUSD straddle with a notional of €577,132, or at the strike of 1.0871 is
US$627,400 per leg (meaning that each one of the € puts and the € calls will have a notional of US$627,400). The
total gamma of these options of €476,650.87 is very close to and offsetting the euro put’s gamma of €518,099 and
the price is a paltry US$4,836.26. The trader buys these options in the market (in our fiction we continue to assume
all prices are mid-market). The introduction of bid and ask prices would not change anything in our analysis. Figure
11.13 presents the portfolio after we add the straddles.
Figures 11.12–11.15 show the portfolio value after the portfolio has been delta and gamma hedged. The three
screens show different columns to give a full portrait of the characteristics of the portfolio.
A few comments are in order. First, note from the scenario chart in Figure 11.14, the monochrome line (①) – so,
for a rise in spot of 5% the loss is now close to US$30,000; for a fall of 5%, the loss is now US$60,000. A great deal
has been done to immunise the portfolio against spot moves related to the electoral results in France.
Congratulations!
Look at the gamma columns Figures 11.13 and 11.15. Notice spot constantly moves so they are slightly different.
The net gamma in the portfolio is now very close to zero. In addition, of the original US$174,000 received as
premium from the sale of the 1.0550 € put, US$4,836 is gone because it has been spent to buy the short maturity
straddle. It was sacrificed for the purchase of the short-tenor gamma-boosting straddle. The other thing to note is
that in Figure 11.15 the monochrome line (①) is not symmetric with respect to the zero spot change line. Why? you
ask. It’s because our original option had a strike of 1.0550 and a delta of about 35%. Our straddle has a strike of
1.0871.
 Consider that every time the trader makes a move, they have to consider that they are partially destroying their
precious previous trades. They bought the straddle to make sure gamma is close to zero, but before they did that,
they had to think of how it would impact all the other greeks. They had to consider that the delta would potentially
move away from zero and that they would have to add an adjustment trade to the delta hedge after the purchase of
the gamma-boosting option structure. It turns out that this was not necessary because there was a way to purchase an
option combination with a lot of gamma and no delta, which is the straddle.

Vega
OK, the delta and the gamma are close to zero – no real fear if spot moves in small increments. What about if
volatility moves? The trader is short vega by virtue of the sale of the 1.0550 option. What was their thinking? Was
this a conscious decision not to care about volatility moves? More likely, they thought volatility would fall and
wanted to be short. If vega exposure was the primary concern, then the second trade would have been different. In
that case, the trader could have purchased a straddle with almost no gamma but a lot of vega, like a long-tenor
straddle. In our analysis, we will assume they were concerned about both and wanted no part in these exposures.
They simply wanted to facilitate the client’s hedging business. If in fact they did not want to have any vega or
gamma position, the simplest solution would have been to sell the 1.0550 to the client and immediately purchase it
back from another market-making bank. As we mentioned, this would have addressed delta, gamma, vega and all
the other greeks. The problem with this approach, however, is that the interbank market for options is not very
friendly. At best, the bid/ask spread, on a direct request for a 1.0550 € put, would have been too wide and most
likely they would be paying away the spread. There are two solutions to this problem. First, charge the client more
to begin with. If they know that they will not be able to buy back the option for US$174,000, but for, say,
US$190,000, then charge the client US$200,000 to begin with, and then buy back the option from another bank for
US$190,000 and go home happy. The problem with this scenario is that, if the client puts a couple of banks in
competition, US$200,000 will likely not be the best offer, and therefore there won’t be any trade at all. This
scenario, though, works very well with local banks, which offer white labelling of option prices. This is because no
other banks service or extend credit to their customers. Therefore, local banks have the luxury of being able to back-
to-back their trades and still turn a profit.
The second way to deal with the problem is to still sell the option at US$174,000 but then keep a low profile. The
trader will keep their eyes open for any option offer in the market where the details are somewhat similar to the
option they just sold. It is always better to wait for price requests so that you can be a price maker instead of a price
taker. The key concept here is that you will have to wait, and that is not a luxury they have, given the timing of the
elections.
 As mentioned, in January 2015 one of the large FX megabanks was doing exactly that. They were waiting for the
best time to replace their market hedge on EURCHF, as the previous one matured. At that precise time, the Swiss
National Bank decided to stop its three-year intervention activity to keep the Swiss franc artificially weak versus the
euro, and EURCHF plummeted by 40% in the space of seconds. According to the Wall Street Journal, the bank,
which at that point was without a hedge, lost US$150 million.1
What type of options is the most likely to be offered on the interbank market? The tenor, which is six months, is
very liquid, so options with that expiry will be the subject of interbank trading. The strike is a 35 delta and is not
very liquid: 25 delta, 50 delta and 10 delta are more liquid. It is therefore likely that the trader will end up buying
back from the market a 25 delta option. In this case, the net exposure to all the greeks we mentioned above will be
something other than zero but not as much as the situation described thus far. The trader can run another portfolio
analysis that will tell them what greeks are still outstanding and, in stages, hedging trade after hedging trade, they
will get closer and closer to having zero greeks. Realistically, the portfolio will never have all the greeks at zero. To
achieve that, a high number of trades would be needed, and a lot of the bid/ask spread will have to be paid, and that
is not a great way to make money. Instead, they will buy and sell options that are most liquid and/or consistent with
their risk appetite and market view.
Let’s return to our base case. The trader is comfortable being short volatility but was worried about being long
EURUSD and being short gamma. Now the portfolio is just short vega. But for how much, and which periods? Yep,
we are introducing another dimension: vega not as one number, but vega over different buckets of time. The vega
column in Figure 11.13 reads –€24,368. This means that, if implied volatility rises by 1%, the trader will have a loss
of that amount. Remember that the trader also bought the straddle, so that piece will make them long the three-day
vega. In addition to the spot ladder, the vega bucket report is the next most valued report for a trader. They would
check this by distributing the vega into tenor buckets. In reality, they will have hundreds or thousands of options in
this portfolio, so one can easily see the benefit of looking at vega exposure in buckets rather than one option at a
time.
Figure 11.16 shows that at a spot level of 1.0868 the three-day straddle contributes US$418 of vega. In other
words, if the one-week volatility rises by 1%, then the portfolio value rises by US$418. The 1.0550 contributes a
negative vega of –US$22,839 in the six-month bucket.
This is roughly the thought process that the trader will go through every time an option is bought or sold. It
actually is more complicated, as the risk reversal and the butterfly come into play. Even this toy portfolio has a bit of
risk-reversal risk because, as we mentioned above, the 1.0550 is OTM but the straddle is ATM. Therefore, the next
step may be to do a small risk reversal to balance out the dvega/dspot risk. We will not go through the exercise in its
entirety, but just sketch the concepts. At this point, we hope the reader understands how gingerly these portfolios
should be approached and have a sense of the amount of work necessary just to maintain balance.
RISK ACROSS THE INSTITUTION: VAR
Above, we explained how a trader would look at the risk in the trading book. However, the institution as a whole has
other constraints on the amount and type of risk it takes. These other ways of accounting for risk are mostly used by
risk managers and asset managers. Traders concentrate on the ability to hedge the risk they don’t want and keep the
risk that they feel will earn them a profit. Asset managers and risk managers tend to look at what risk is left in the
portfolio and how that risk would be impacted upon by broader shifts in the markets.
Let’s expand our portfolio. In the example below, the portfolio has:

sold a £ put US$ call ATM forward, with a tenor of one month;
bought a one-year ATM US$ call MXN (Mexican peso) put;
sold a two-week straddle in EURCHF (euro Swiss); and
sold a six-month risk reversal in USDCNY (US dollar China), which means we have bought a US$ put CNY call
with a delta of 25% and sold a US$ call CNY put with a delta of 25%.
For all the trades together, see Figure 11.17. In the figure, you can see various screens showing all the details of our
sample option portfolio. The next step is to rethink risk away from specific greeks with a single-minded focus on the
instruments at hand. It is an attempt to look at all the instruments in our portfolio and ask a much broader question –
suppose the world changes substantially from its current trajectory, what happens to the portfolio’s value? The
question to answer then is, what changes are you expecting?
A first step may be similar to what we have done above. Suppose this or that currency moves by so much, what
happens to our portfolio? A more dramatic question may be: look at history and assume that the worst-case calamity
we have witnessed actually happens. Then what happens to your portfolio? This second question involves moving a
ton of potentially indirectly related items and seeing the effect on your portfolio. This type of analysis is called
value-at-risk. A more rigorous definition would be: VaR estimates how much a portfolio of financial instruments
would lose, given normal market conditions, in a given timeframe. It is a loss number that tells you that, in most
cases – usually 95% or 99% of the time – your portfolio is likely to not lose more than a certain amount. A related
number that is always of interest is: what is the percentage of the time that your loss will exceed that number? VaR
was popularised by the JPMorgan incubator RiskMetrics in 1994, and has been growing in popularity ever since.
Institutions such as banks, insurance companies, asset managers and especially regulators love VaR for a few
reasons.

Most of all, they love it because it’s so easy to explain. It is easy to go in front of any board, CEO or congress and
say: “All the hundreds of thousands of incredibly complicated positions we have will lose only US$X should we
have another Lehman-type crisis” (pick your period of terrific turmoil). Everybody gets a warm and fuzzy
feeling because with that one statement you are lulled into believing you can predict the future from the past or
from a simulation.
The second reason is that you can throw everything you have into the computer bucket, and it will crunch it all
together. If you have currencies, bonds, equities, equity derivatives and credit instruments, they can all be
valued the same way. Not only that, VaR calculations can take correlations into consideration so that assets that
move in opposite directions offset some of the risk.
There are many software packages that can run VaR calculations inexpensively. If you choose to calculate VaR by
computer simulations versus statistical methods, the sheer massive increase in computer processing power
makes it easy to calculate VaR even on massive portfolios.
Every cool kid in town is using it – you want to be the only one not using it?

However, there is also a massive body of academic and empirical work that argues VaR is not the elixir many
believe it to be. There are many reasons for their admonitions.

The results are presented as if we can predict such things with great certainty. When someone says: “I guarantee
with 99% confidence you will not lose more than US$X”, it is human nature to assume that it is a mathematical
truth that your losses are limited. Not true – your losses are not limited.
By definition, 1% of the time your losses will be worse than your worst nightmare.
Typically, many of these models assume distributions for asset classes that are not justified. For example, it is
assumed that returns on FX are normally distributed. Mathematically, then, we can say that FX spot must be
lognormally distributed. If that was the case, you would be incredulous on hearing news analysts saying things
like: “This currency pair is six standard deviations away from the mean.” That just should never happen once,
let alone often in periods of crisis.
The sophistication of how VaR is calculated and applied leads to different ways of calculating it, and every method
comes up with different results.

An in-depth discussion of VaR is beyond the scope of this book. Suffice it to say, whether you like it or not, as a
portfolio manager or a trader, or a corporate treasurer for that matter, one needs to be aware of it and know how to
interpret the results.
Now, back to our portfolio. Suppose that we are calculating the historical VaR on the portfolio with a confidence
interval of 99%, a tenor of one day, and over a historical interval of five years (the same calculation can be done
using simulations instead of historical returns). This is how the computation is done. For each day in the past five
years, we look at all the market parameters that changed that day. For example, if GBPUSD moved by 1% that day,
then we calculate the value of all the trades in our portfolio by assuming that GBPUSD now moved by 1%. Of
course, in the past the spot reference would have been very different, but we apply the 1% change from the current
market level. That day, GBPUSD volatility also changed. Suppose it fell by 1% – then we apply that change to the
current level of GBPUSD volatility. The change in spot and in volatility will have an impact on our GBPUSD
option, but not on any of our other options. Then we examine the changes in EURCHF, USDCNY and USDMXN
for spot and volatility, and incorporate the changes in our calculation.
Other asset classes will have changed that day, such as the US stock market. Since we do not have equities in our
portfolio, those changes will not impact on our valuation. Therefore, for every day in the last five years our portfolio
will show a change in value consistent with the market changes that have happened on that day. The final list of all
the P&Ls for those portfolios is depicted in Figure 11.18.
Figure 11.18 shows, for each day in the last five years, the P&L that is generated if we subject our current
portfolio to the same market changes that happened that day in history. On the left is a chart showing the worst loss
happening in January 2015 (deep monochrome line (①)) and the greatest gain happening in November 2016. Can
you tell what happened on those two fateful days? On the right is a table of the same P&Ls, one per day.

The history of the world is written in these numbers, if you know how to read it. The large drop in the middle of
2016 occurred as the £ put US$ call went suddenly ITM, as a consequence of the Brexit vote in the UK. The largest
drop, the January 2015 loss of US$1,250,000, is when the Swiss National Bank suddenly ended its three-year effort
to keep the national currency from appreciating. The drop in August 2015 was due to the People’s Bank of China
devaluing the yuan by about 2%, increasing the value of the US$ call Yn put we sold and decreasing the value of the
US$ put Yn call we had bought. Finally, you can see our largest gain, the US$0.4 million in November 2016 as the
market reacted to the election of Donald Trump to the presidency of the US and the subsequent collapse of the
Mexican peso, which pushed our US$ call Ps put ITM.
Let’s isolate a specific historical period – for example, April 2016 to April 2017 (see Figure 11.19). The next step
is to take all the data points calculated above, and mush them together into a neat histogram (see Figure 11.20).
Figure 11.20 shows a histogram for the results from the study carried out in the previous simulations. The chart to
the left tells us that there was one day when the P&L had a loss of about US$590,000. We can see the precise
number on the right, under “worst days”. It also tells us that there was one day, all the way to the right in the chart,
with a P&L gain of about US$418,000, listed under “best days”. The histogram also tells us that there were about 51
days with a P&L in the region of –US$18,000.
 That’s a nice, understandable and usable histogram. It contains one year of data, or about 255 days. There is a
confidence interval of 99%, which means that we can exclude 1% of the days from our VaR, or 1% of the days is
255 × 1% = 2.55 days. Start from all the way to the left of the histogram and count two or three days. In the table, on
the bottom right start with –US$589,466, then the second day at –US$299,512, the third is –US$204,130. That is
about our VaR with 99% confidence interval and a one-day time horizon. This means that, if we base our analysis
on historical data, every 100 days our portfolio should have a loss of US$204,130 or greater.
Figure 11.21 shows that the VaR with a one-day time horizon, a one-year historical dataset and confidence
interval of 99% is US$204,130. Other combinations of historical basis interval and confidence interval are shown on
the top table. The bottom table shows the VaR of each individual currency position. For example our USDCNY, two
vanilla option position, has a VaR of US$18,844.
There are different historical time periods one can consider. A shorter period may reflect more of the state of the
world as it is today. However, a longer period will include more historically significant market episodes. Also,
consider different confidence intervals. Corporations usually consider a 95% confidence interval, while financial
institutions usually go for 99%, for the good reason that financial markets’ gyrations are supposed to impact on
financial institutions more than ordinary real-economy corporations. Alarger confidence interval means that the
calculation is more conservative and has a greater concern for larger losses that only happen once every 100 days.
Banks usually need to be concerned about a one-day time horizon, as they can literally go bankrupt in a day.
Corporations usually look at a one-year time horizon, as sometimes their product cycle might take a year and they
need to carry out the calculations for budgeting purposes.

Note the matrix at the bottom of the table in Figure 11.21. The VaR column shows the VaR of the entire portfolio
on top, US$87,820. Below that number are the individual VaR of the four components, corresponding to the four
currencies to which our portfolio has exposure. Those are calculated as if each currency position constituted a
portfolio by itself. Note that the sum of the individual VaR is larger than the VaR of the portfolio. This happens due
to “portfolio diversification” – the risk on a portfolio is usually less than the sum of the risks of each component of
the portfolio.
To give you some intuition for this concept, imagine a portfolio that is long euros and short Danish krone in equal
amounts. The correlation between those two currencies is very high, in the order of 98%. Every day the euro surges,
the Danish krone surges as well. As a consequence, the euro component of the portfolio has a positive P&L, while
the Danish component, because it’s a short position, has a large negative component. Those two components tend to
offset. Therefore, the total risk on the portfolio is very small, and close to zero. Nonetheless, individually each
component, the euro or the Danish krone, has a normal-sized risk but the sum of the two individual VaRs (VaR is
always a positive number) will vastly exceed the VaR of the portfolio.
This example can also serve to introduce the concept that sometimes hedging can increase risk. Imagine that, in a
misguided effort to reduce the risk on the portfolio, only the euro component is hedged, reasoning that reducing
some exposure is better than doing nothing. In so doing, the portfolio is now left with only the Danish component
and the risk on the portfolio is equal to the relatively large risk on the Danish component alone. Hedging one
component in this scenario actually significantly increased the risk involved.
This brings us to the column labelled “MVaR” in the table in Figure 11.21. This means “Marginal VaR”, and
shows the decrease in VaR when a particular line item is eliminated from the whole portfolio. In the above example,
the MVaRs of both the euro and the Danish component are negative, meaning that both components act as
diversifiers. In other common situations, each component may add to the risk. In such cases, all the MVaRs will be
positive. The MVaR is a very useful notion for corporates that are considering hedging a particular position. They
should make sure MVaR is positive before they hedge.
 The last column to the right is IVaR (or “Incremental VaR”). This equals the derivative of the portfolio VaR with
respect to the notional of each specific component. Another way to look at it is to say that IVaR is the change in
VaR for an infinitesimal change in the size of a specific component, and then re-scaled (otherwise it would be a
minuscule number) by multiplying by the aggregate size of the portfolio and dividing by such an infinitesimal
change. A third way to look at IVaR is as the covariance of a specific component versus the whole portfolio, divided
by the volatility of the whole portfolio. This last is the easiest – the more the component and the portfolio are
correlated, the higher the IVaR. In addition, the notion of IVaR has, by virtue of its own definition, the good taste of
summing up to the aggregate VaR. In other words, the sum of all the individual IVaRs equals the VaR of the
portfolio, which is great when people want to see a breakdown of VaR.
For better or worse, VaR is a very popular tool to quantify risk. However, financial risk is a very difficult notion
to understand, maybe on a par with entropy. It would be good to here go back to where the discussion started, and
talk more about the risks of using VaR. Many people are in positions that require they act on risk data but they do
not actually understand risk. VaR is the bait that lures them into using one number to estimate and manage. In
particular, during the 2008 financial crisis – and even before that, at the time of the demise of Long-Term Capital
Management’s legendary hedge fund – catastrophic mishandling and misunderstanding of financial risk has been
attributed to the shortcomings of VaR as an all-encompassing proxy for “risk”. One of the most obvious issues with
the notion is that it reduces risk to just one number. Human nature being what it is, people start managing VaR
instead of managing risk. Those are the pitfalls hiding behind sentences that “feel right”, such as: “You cannot
manage it if you cannot measure it.”
Another problem with VaR is that, in general, it is not true that the sum of the individual VaR components is
always larger than the VaR of the portfolio. Table 11.1 provides a brief example.2

Then we have the following VaR calculations, as shown in Table 11.2.

As we can see from Table 11.2, the 27.8 VaR on the portfolio exceeds the sum of 8.9 and 8.9 individual VaRs.
How do we come to this conclusion? If we look at Table 11.1, we can calculate the current price of A as 3% × 70 +
2% × 90 + 3% × 100 + 2% × 100 + 90% × 100 = 98.9, and similarly for B. The 5% VaR of A is calculated as the
minimal loss in scenario 1 and 2, which together sum up to 3% + 2% = 5%. The loss is the price paid minus the
highest value in these two scenarios, so 98.9 – max(70, 90) = 8.9. Same for B. Now for the nice part. In those two
scenarios, we calculate the expected loss as 3% × (98.9 – 70) + 2% × (98.9 – 90) = 20.9.
Finally, there is another problem with VaR. Suppose that I ended up in the 1% of unlucky scenarios. In other
words, I am now in one of the worst 12 days over a five-year time period. VaR tells me how bad the 12th day is.
However, how bad does it get on the 10th or fifth worst day? The notion of VaR says nothing about that. If I am a
risk manager, I want to know. There is a notion that can help me with that: Conditional VaR (CVaR), or expected
shortfall. This corresponds to the expectation when you are within those 12 worst days. In other words, assuming
that I am not within the 99% confidence interval, what is my average loss? You can see that number in the CVaR
column of Figure 11.21. That number is always larger than the VaR number as the CVaR is more conservative. In
addition, for extra beauty the CVaR has the good taste to always be smaller than the sum of the component CVaRs.

RISK ACROSS THE INSTITUTION: SCENARIO ANALYSIS AND STRESS TESTING

In addition to using VaR to understand risk, there is another equally popular and similarly intuitive way to explore
worst-case scenarios when things go badly: scenario analysis. When looking at a portfolio, many would ask: what
happens to my portfolio if the euro drops by 3%, or if the Treasury yield curve steepens by 25 basis points, or if the
S&P 500 index rises by 5%? All those questions can easily be answered by changing the inputs in our pricing
models and re-running the valuations. We already explored such questions when we shocked the EURUSD spot rate
and looked at the value of the portfolio. We can create all sorts of scenarios to answer all these questions.
The fundamental question is, do all these calculations really help us to better understand the risk in the portfolio?
Yes and no. On the one hand, the euro moves by 3% quite frequently over periods varying from a week to a few
months, so knowing what to expect is certainly helpful. On the other hand, running such reports can be a waste of
electrons. For example, on one occasion the CEO of a large US bank asked me to calculate the change in portfolio
value should the US dollar lose half its value overnight. Admittedly, it was our fault for having buried him under a
mountain of mathematical mumbo jumbo and simulations. Nonetheless, knowing that answer is probably completely
useless. The chances of the US dollar losing half of its value against all other currencies overnight can safely be
ignored. The bottom line is that one can have a lot of fun running scenarios such as in Figure 11.22.

Figure 11.22 shows what happens to the portfolio if we shock the spot rate. We shift all four currency pairs that
compose our exposure: GBPUSD, USDMXN, EURCHF, USDCNY and the volatility for each. The shocks applied
are –10%, then –5%, then no shock, then +5% and finally +10%. The columns represent volatility scenarios and the
rows spot scenarios. For a zero shock in both spot and volatility, the current value of the portfolio is US$379,465.
There are also valuations for changes in each scenario combination, the delta and the vega of the portfolio. Vega is
positive, so logically if volatility increases you can see that the portfolio value increases. The delta is also positive,
but keep in mind that a positive delta for USDMXN, due to the quotation convention, means that if USDMXN rates
rise, then the Mexican peso weakens, and if your delta is positive then your portfolio will lose money. Note the
column with the heading “0”, which shows losses for positive shocks to the underlying prices.
Tables like that in Figure 11.22 are very helpful to get an idea of the type of risk faced. For example, a move of
5% in the spot rate is not impossible, and could cause (if it is an up move) a loss of about US$411,538, from
US$379,465 to a negative –US$32,073. If this is too much, then we need to do some delta hedging, as described
above. The table is also useful if a portfolio manager at a hedge fund believes that spot rates will increase and
volatility will decline. They can set up their portfolio to take advantage of this incredible gift of being able to
forecast financial markets gyrations. In that case, the table shows the world that they might be great at divination but
not at setting up a portfolio to make money out of it.
There is another flavour of scenario analysis that is very interesting and incredibly appealing to our instincts.
Instead of having individuals guess what shocks the market may experience, consult history. Here the benefit is
clear. Individuals may be very bad at estimating how likely it is that the US dollar will lose 50% of its value
overnight, especially if you’re the CEO of a very large US bank, but individuals are very good at remembering past
crises that have shaped our collective subconscious. Some of our readers will have memories of intense experiences
around the Lehman Brothers default, the collapse of Long-Term Capital Management, the Tequila Crisis, the Greek
crisis (the country, not negative gamma) and so on. Everybody remembers the pain and the drama that such
calamities brought. The concept is to create in current markets the same price shocks that happened at the time of
such calamities, and to see how our portfolio would fare in the face of such events.
Figure 11.23 shows a list of dramatic events or crises of the past and the impact that such events would have on
our current portfolio, in the column headed “P&L Port Ccy”. The fact that we are long a US$ call Ps put would have
netted us about US$8 million in a repeat of the Tequila Crisis. A repeat of the Lehman Brothers default would have
netted us about US$1,500,000. In all the above scenarios, the shocks are applied to all market variables, including
equity prices, equity volatility, rates, rates volatility, currencies and relative volatility, and so on. Therefore, each of
the scenarios is composed of hundreds of market shocks (see Figure 11.24)
CONCLUSION
There are many ways to think about and assess risk across the financial markets. In this chapter, we focused on
where we can use all the “greek” we learned in previous chapters. There was a further demonstration of how delicate
the balance is for any portfolio manager to control the risks they want and mitigate or the risks they do not like. We
hope you got a palpable sense of how a trader needs to think to get this right.
However, “no man is an island”. These risks exist in a bigger universe of investments and institutional interests
that must be considered and monitored. To do that, institutions have concluded that the two most effective tools are
VaR and scenario analysis. VaR calculates the worst-case loss for a predetermined period within a predefined
confidence interval, usually 95% or 99%. As mentioned at both the beginning and end of our VaR discussion, VaR
has important issues and limitations! It must be used with care, and there needs to be careful analysis beyond the
single number it offers.
Scenario analysis takes a very different tact. It is not a statistical test that relies on tons of data and correlations.
Scenario analysis shocks existing positions across all asset classes based on realistic situations. We know they are
realistic because they happened. On the other hand, just because they happened in the past does not mean these
situations will happen again, and certainly may not happen again in the same way. As with VaR, use with great
caution.

1 https://www.wsj.com/articles/swiss-franc-move-cripples-currency-brokers-1421371654
2 For further details, see Carlo Acerbi, Claudio Nordio and Carlo Sirtori, 2001, “Expected Shortfall as a Tool for Financial Risk Management” (available
at https://arxiv.org/abs/cond-mat/0102304).
 12
Hedging FX Risk at Corporations

This chapter will examine the foundations of how to form an FX hedging strategy by exploring broad approaches to
policy. It will then focus on the world of corporate treasury, and the practical (and sometimes personal) business
drivers for those hedging FX. Finally, we look at some specific tools that can help create parameters around the best
practices of hedging, with the aid of specific examples that clarify how the tools are used, and sometimes misused.

THE CORPORATE TREASURER AND THE FX CONUNDRUM

It is not always clear whether FX risk should be hedged, and how. In this section, we will explore those perspectives
and discuss some unconventional ways that corporations approach hedging.

Hedging policy perspectives

Does hedging matter in the long run?
Before getting into the nitty gritty, it is worth remembering that one of the most legendary investors, Warren Buffett,
would argue that this part of the book is mostly useless. He does not believe in hedging. Here is an excerpt from an
interview: “When you are investing or buying a business with the intent of owning that business for years or
decades, your focus rests solely on its fundamental attributes.”1 In the late 1980s, Buffett invested in Coke. Yet, “he
couldn’t have cared less about the effect of the British pound, the Russian ruble or the Iraqi dinar”. He also said,
“Coke’s incredible performance was not related to currency effects: but rather volume growth. Sure, there may have
been certain years where currency translations were harmful to results. But over time, those issues matter less and
less.” He added:
A long-term perspective allows you to eliminate or rely less on factors that are minor over time. Investing becomes a lot easier. Think about
my last statement carefully – investing becomes a lot easier.

Warren Buffett is not the only one. I recall a friend, Charles H., who was the CFO of a large multinational
corporation. He was happy for me to make a good living selling sophisticated derivatives to hedge FX risk, but he
would have none of it for his companies. He believed it was a sham created by banks to make money.
There are a few reasons to not rely on this theory.

Corporations and investment shops may last for hundreds of years, but your expected career span with that
company or your job as a portfolio manager will most likely be a lot shorter. Metrics and evaluation of
performance will most likely be considerably shorter than even 20 years (or even 20 months).
Currencies are in fact not reverting, at least in our sample of the last 40 years. Take a longer sample if you wish. At
the beginning of the 20th century the value of the pound was around five dollars. Where is the mean reversion
now? Take a look at Figure 12.1. We took the top six currency pairs against the US dollar in terms of flow and
economic activity, and normalised their movements back to 1975. Only the euro (from 1975–99, we combined
its constituent currencies) appears to be close to 100 again, and that after spending decades well below 100.
With emerging-market currencies, the effect is significantly more skewed and non-reverting.
This theory assumes the company will be able to hold steadfast and not hedge during good times and bad. There are
two problems with that. As mentioned in Part II of this book, it is a fact of human nature that we are just not as
good at handling pain as we are at dealing with pleasure. On a practical level, it is also true that the market can
always stay “mispriced” longer than you can stay solvent. For example, what does a company do if it appears
that a severe change in a given currency can put it out of business? Hedge of course!
The final assumption is that the company will be hedging similar amounts of exposure during both good and bad
 times. This is not true on many levels, particularly if the company is involved in a cyclical business. If, for no
other reason, we can logically expect a company to grow over 10 or 20 years, the relevant exposure will
definitely be different over that period.

Do shareholders want their currency exposure?

When discussing hedging at this level, someone will inevitably raise the issue of investor intent. When one buys a
company’s stock, in theory they are aware of what percentage of earnings comes from overseas, from which
countries and what potential impact it will have on earnings. Therefore, if the treasurer actively eliminates that
exposure they are tampering with what the shareholders understand to be the risks and rewards as advertised.
So much for theory. We question whether investors will consider the FX risk of, say, a manufacturing company,
assess which way the FX market will move and then make the investment decision. We are even more suspicious of
the corollary argument that investors take FX risk by buying stocks in companies with operations outside their base
currency. There are a myriad more efficient ways to take FX risk.2 For a sophisticated investor, the investment
decision is based on the company’s ability to grow in its chosen field of endeavour across all the countries it is
operating in. There are some that argue that investors actually expect the treasurer to hedge away FX risk, and in fact
companies that do hedge are valued more than companies that do not.
We will not address this topic in a serious way here, but simply want to acknowledge it, particularly since CEOs
and CFOs love to ignore this argument when it suits them. When they don’t hedge, they are all happy to blame poor
results on adverse currency moves. In fact, CEOs and CFOs of exporting companies have made a ritual out of
complaining that it is not their management that has caused the negative results but the strong dollar. Of course,
when the dollar is falling it is their brilliant strategy that added to the bottom line. Importers make the opposite
arguments. Paradoxically, when hedging away the FX exposure, they are scared to death that losses on the hedges
might end up on the front page of a newspaper with the word “speculative” in the same sentence. You will not be
surprised to hear that there is a dearth of stories where CEOs credit the FX move and not their brilliance for
favourable performance.

Organic hedging
Wouldn’t it be great if a global business, across many continents and many different currencies, did not have to
worry about FX at all? That’s easy. Structure the business so that it is immunised against FX risk – in essence, create
cashflows so that cashflow shortcomings are met by cashflow surpluses in the same currency. There a few ways to
do that without using the traditional hedging instruments. These techniques are organic, in that they can evolve from
the company’s overall activity across FX and interest rates. They are sustainable in the sense that as long as you
continue to have ongoing operations, it is likely that these can be paired off and regenerated.
In terms of matching cashflows, the most intuitive technique is to match the currency of some expenses and
revenues. Therefore, if a company has revenues in Switzerland, it would want to consider incurring expenses in
Switzerland. This can involve moving operations or other physical activities there, but we will only focus on the
financial activities that can be shifted. The most obvious is to borrow in Switzerland. As revenue comes in, it will
offset the interest and principal expense, all in Swiss francs. This is easier said than done for many companies.
Deciding where to borrow involves a much more complex set of factors than just the currency, the most obvious
being the interest rate environment in Switzerland and also this specific company’s access to that market. Where an
IBM may be a name that is equally well received in New York, London, Zurich and Tokyo, a smaller, local, US-
based company will end up paying a premium to investors, which consequently may make Swiss borrowing too
expensive.

One way to clarify if issuing debt to offset FX risk makes sense is to understand the market level of the firm’s
credit spread. A company can examine where its debt is trading and creates its own specific yield curve. In Figure
12.2, bonds for IBM are used to generate a curve that reflects where IBM can issue for different maturities.
Once a company can see where it can issue debt in the US, it can see the basis for a comparably rated entity in the
country of interest. Then it would still need to work with its investment bankers to get an idea of how well its name
will be received in that country.
One thing is to calculate the mathematics of the respective yields, but supply/demand is another game altogether.
Despite what the calculations say, appetite for a specific US issuer might be high in Europe, possibly because there
are not a lot of European issuers in that specific sector, or at that specific credit level. In many cases, large
institutional asset managers can make or break an issuance. This is where the ability of a good syndicated sales team
at the debt underwriter can shine, as they will be in charge of marketing and distributing the issuance.
In terms of revenues and currency risk sharing, procurement contracts can be structured in such a way that the
transaction price is contingent upon the FX rate being lower (higher) than a specified level(s). For example, many
US companies have such agreements with Chinese suppliers providing that, if yuan appreciates beyond a certain
point, they will pay a fixed predetermined amount. It may sound like this: “Over the next year, we will buy 10
million widgets at CNY62 per widget when the Shanghai close of the BFIX rate is above CNY6.2/USD. When the
closing rate is less than CNY6.2/USD, we will pay US$10 per widget.”
How can one evaluate such an agreement? You may recognise, in this language, something familiar from our
previous conversations. This company has the right to buy widgets at CNY6.2/US$ or better. Beyond CNY6.2/USD,
it is guaranteed a fixed rate. This of course is nothing less than a currency option – which, in this case, the Chinese
vendor has sold probably for free.3
For other internally sourced risk-management techniques, there are many different schools of thought. Evaluating
complex commercial risks with FX components systematically is outside the scope of this book. Suffice to say that
they typically involve some form of self-insurance. It can be very broad, such as a huge multinational combining
their FX, interest rate, commodity, fire, terrorism, wage, employee life insurance, etc, into one pool and managing
these risks as one. Self-insurance can also be very targeted, where there is financial risk across many asset classes.
This targeted, packaged risk can then be offset to the market using correlation contracts.
All the tools mentioned may have their limited application depending on the company, and are being used by
many corporations in a limited way. As a general rule, however, they require corporate decision-making structures,
resources and a risk profile that is not consistent with the current corporate reality.

FX and the world of a corporate treasurer

The treasurer’s fundamental business is very often impacted on by factors outside their control: their job is to have a
response and to guide the company through them. In the end, they will be held accountable. These factors range
from the strength of the base currency, to the level of technological change, to conflicting internal objectives and key
vendor relationships. In this section, we will lightly touch on these topics.

Dollar levels affect trade

In the spring of 1254, two Venetian merchants, Niccolò and Maffeo Polo, found themselves stranded in Sudak, on
the Crimean peninsula, because westward travel back home was simply too unsafe to attempt. They were so
desperate that they instead travelled east, on what became the Silk Road, and reached the court of the great Khan of
the Mongols. Niccolò’s son, Marco, many years later narrated their travels and contributed to open communications
between Europe and Asia, a dialogue that changed the course of history. The Silk Road passed through
municipalities, fiefs and kingdoms that all required merchants to pay bribes, taxes and protection; all these
middlemen made commerce very difficult. Many millions of beheadings later, the Mongol conquest eliminated them
and made travel so safe that people used to joke that a maiden carrying a basket of gold on her head could walk from
Xanadu, the capital of Kublai Khan’s Yuan dynasty, to Constantinople without being disturbed.
Eight hundred years later, we consider another type of middlemen, also dealing with some unpredictable new
trends that, just like the Mongol conquest, find their livelihood threatened. Let’s set the stage for our story. For most
of this century, globalisation, emerging-market performance and investment abroad have caused the US dollar to
weaken (see Figure 12.3). This helped US exports rise from 9.13% of US GDP in 2002 to 13.58% in 2013,4 and to
33% of large listed company revenues.
 Unexpectedly, in the summer of 2014 the trend came to an end, and the dollar strengthened by 21% by 2015, and
more than 30% by December 2016 (see Figure 12.5). This wiped out about US$160 billion in export revenues and
ushered in a new era where FX markets favourable to US corporations were a thing of the past.
As an aside, the most commonly used dollar index, DXY, has been around since 1973. It is outdated as a measure
of the US dollar’s value relative to global trade. It has only been updated once, at the advent of the euro. There are
better indicators for the dollar’s strength relative to trade, such as the Bloomberg dollar index, BDXY, which is
linked to trade flows and updated annually (see Figure 12.6).
Conflicting perspectives at the C-suite and technological change create stress
There are a few stress points that make the job of the treasury FX staff difficult, including the fact that a modern
corporation looks at FX risk from two differing perspectives. On the one hand, the C-suite is mostly concerned with
the share price, mergers and analyst reactions to earnings calls, none of which are very sensitive to the vagaries of
exchange rates. CFO and CEO relations with banks develop mostly through investment bankers, who know little
about currency markets. On the other hand, treasurers, assistant treasurers and FX managers mostly interact with
salespeople who sit on trading floors. Bankers and salespeople are the middlemen, the same type of people who
were put at risk by the emergence of a safe Silk Road; more on this theme later.
 In the next section, we will address one way the divergence between C-suite and FX managers manifests itself:
which is more important, earnings-at-risk (EaR) or cashflow-at-risk (CFaR)? We will offer a way to conceptually
think about that choice, and help create an efficient frontier that balances the different objectives.
However, returning to the differing hedging perspectives, treasurers worry about cash availability and economic
losses due to FX moves. They live inside a paradox – if they do not hedge, revenues might suffer; if they do hedge,
they might be facing headline risk in case the hedge turns sour. Historically, corporate FX managers are jacks-of-all-
trades: very busy with accounting, cash management, payments, risk quantification, trade execution and bearing an
oversized responsibility. They rely on the banks to get the FX puzzle sorted out. It is true that banks act as
middlemen, but the complex web of functional relationships is closer to a symbiosis.
The second stress point is the massive amount of information that is now instantaneously available. Even in the
early 2000s, you could still hear corporate FX managers asking sales for “a good level in the yen”. The market was
so opaque and mysterious that corporate hedgers needed to ask roughly where the exchange rate was. In today’s
electronic real-time market, such a question is equivalent to sending letters via camels travelling the Silk Road.
Today, you can summon a good level on yen to your screen with five keystrokes. Corporate FX staff not only can
get a good level, but just as quickly they can check the prices from all their banks. Figure 12.7, for example,
provides instantaneously competitive price indications from all potential counterparts.
 By the late 2000s, you could hear corporate FX managers asking bankers for “a chart of six-month EURUSD-
implied volatility for the last three years”. Today, they can put that chart together in seconds (see Figure 12.8).
The stress comes from having to digest and react to all this information that is now accessible and increasingly
sophisticated. The assistant treasurer, who would typically call their bankers for any information they needed, now
has to learn how to use effectively information on macroeconomics, spot and forward rates, volatility expectations
and history, and draw a conclusion on what they should do.
There is value in smart relationships
Banks used to have armies of people to help arrive at the appropriate recommendation for each of their clients.
We’ve spoken about this phenomenon earlier in the book, but suffice to say that banks have greatly reduced the
depth of sales expertise, economic analysis and structuring on demand that was a hallmark service offered to
corporations. Nevertheless, bank/client relationships can be invaluable. In fact, the value-added of a banking
relationship can be very critical to buy-side clients, and especially corporations. Despite so much information being
commoditised, the relationship cannot be reduced to the transfer of data and pricing information. The market-savvy
people on the trading floor can offer invaluable insight.
We will now examine some previous situations to give a palpable sense of the value of such relationships. For
instance, in November 2011, June 2012 and again in June 2015, many companies worried about revenues from
Greece. On advice from the banks, they ended up sweeping their euro revenues from their Greek accounts into their
German entities accounts each night. Data and analytics providers jumped in to streamline the communication of
information. Specific screens focused on monitoring the Greek crisis (see Figure 12.9). This Bloomberg screen was
created to allow quick display of economic indicators and market data related to the financial and political crises in
Europe.
Another more tragic example happened on Saturday, September 13, 2008, when the corporate sales team at one
large bank, concerned about the fate of Lehman Brothers, called a large US corporate to alert them. Unfortunately,
the client still proceeded to send Lehman Brothers a cheque for about US$100 million first thing Monday morning,
September 15, as the outbound leg of an FX settlement. The inbound leg was never received, as Lehman was busy
filing for Chapter 11 and going out of business. These are the cases where the banking relationship could be
priceless.

The type of risk referred to above is called settlement risk. Let’s take a small detour to address this FX
phenomenon. In any FX transaction, the two parties agree to exchange one currency for another. For example, if the
exchange is between US dollars and Japanese yen, the yen will be delivered into an account in Tokyo and the dollars
into an account in, say, New York, when New York opens. There is a timing difference, and consequently the risk
that one leg will get deposited while the other counterpart does not make good on their leg of the contract. That risk
has a famous history. A privately owned German bank called Herstatt Bank was forced into liquidation by the
German regulators in June 1974. During the European morning of that day, banks that had FX deals with Herstatt
sent payments of Deutschemarks to Herstatt to its Frankfurt account, expecting US dollars into their US dollar
accounts when New York opened. The authorities closed the bank at 4:30pm Frankfurt time, which was 10:30am
New York time. Herstatt ceased operations before delivering the dollars, and of course the counterparty banks did
not receive their US dollar payments. Risking the entire notional amount in an FX deal is a very rare circumstance.
FX risk due to market movements will never be this extreme as exchange rates rarely go to zero. Furthermore,
settlement risk has been dealt with very effectively since 1974, and is no longer seen as an issue. Many transactions
are now cleared through the CLS Group, which systematically settles transactions and mitigates the associated
settlement risk. Additionally, the new regulatory environment, especially with MiFID II, will force more
transactions into a cleared, exchange-type platform, reducing settlement risk further.
Another publicly acknowledged example of good advice from bankers is Southwest Airlines, who enjoyed a long-
term competitive advantage by being the only major airline to hedge jet fuel in the early 2000s. These are situations
where bankers’ sound judgement and advice are priceless. Sometimes it cuts the other way. A legendary example is
the unwinding of a hedge for an aborted cross-border merger with UK rival BOC by Air Products in 2000. This
unwind cost the firm US$300 million5 on a deal size that amounted to about US$11 billion. The deal seemed to Air
Products to be on track. It decided to hedge its anticipated need for sterling by buying pounds forward. However, a
few months after the announcement, the US Federal Trade Commission did not approve the merger, and this
decision effectively forced the deal to be abandoned. Air Products had a huge GBPUSD forward hedge, long
sterling. The use of a forward hedge is appropriate since it felt the merger was a done deal, and especially as there
was potential for the pound to appreciate, making the cost rise. The situation did not unfold as expected. Not only
was the merger abandoned, but the pound collapsed. To unwind the hedge, Air Products had to pay the hedge
counterparty nearly US$300 million. The hedge was actually a combination of options and forwards, but in later
interviews former treasury management stated that a strategy of purchased options would have rendered the
mechanics of the merger deal too expensive. However, it would have rendered the unwind of the hedge costless! In
fact, the company would have received money for the unwind.
In all fairness to the FX salesperson advising Air Products, the key issues to consider in the type and timing of the
hedge are the certainty level and timing of the merger. There is no realistic way for the bank salesperson to have a
really good understanding of those probabilities. Since 2002, the sophistication of products offered to help
corporations hedge mergers has evolved dramatically. At the major banks, the entire bank team, including the
investment banker, will typically get together and decide what the deal completion probability is. They will create a
timeline, and offer some very creative products to the client.6
For example, banks may offer a conventional option hedge to the acquiring company that has zero upfront
premium and no obligations if the deal is rejected (these are called “deal-contingent” hedges). However, should the
deal be consummated, the company will pay a massively higher premium than if they had bought the option
normally and risked losing their premium. This takes us back to the discussion on how “each participant should stick
to their knitting”. We will discuss this further later when we talk about the source of excess profits. From the
corporate perspective, it makes a lot of sense, especially when they know that if the deal is consummated, they will
roll the excessive premium cost into the deal borrowings that typically end up on the target’s balance sheet. The
bank, on the other hand, takes the risk and runs the mathematical analysis to figure out if it can extract more value
by transacting FX at a more favourable rate.
In some cases, corporates understand the value of a good relationship and bankers can openly discuss with them
the amount they will charge above market prices, also known as “mark-up”. The mark-up is justified by advisory,
extension of credit and the sound financial condition of the bank. Today, more than ever before, these elements can
be deconstructed and estimated. For example, the extra cost due to liquidity considerations is available on any FX
trading platform. The amount a bank should charge a corporate due to the credit exposure on a forward contract
hedge is easily calculated using data provider tools. Consider how a treasurer can use the Bloomberg terminal for
that. To compute the debit valuation adjustment that a bank might charge a corporate client takes just a few
keystrokes (see Figure 12.10).
An example of credit valuation adjustment calculations on a portfolio with two counterparties is the adjustment to
the value of positions visible in the CVA column. Nonetheless, many factors conspire to undermine the
bank/corporate relationship, chiefly a misguided perception that banks are no more than middlemen and that the
natural evolution of things is that everything now needs to be disintermediated. Surely one of the reasons is the loss
of confidence in banks after the scandals following the global financial crisis. Bank reputations being dragged in the
mud, the Occupy Wall Street movement, the rigging of FX fixings, the mysteries of high-frequency trading all
contributed to such mistrust.
 One head of the corporate sales team at a major bank told me that their efforts to warn corporate clients of the
imminent demise of the euro in 2014/15 were widely ignored by distrustful clients (see Figure 12.11). The
frustration contributed to their decision to resign, and instead manage a chain of fitness centres. At the same bank, a
successful structurer, whose job was to advise clients on volatility relative value trades, resigned to pursue a career
as a chef and the highest revenue producer on the FX corporate sales desk left for a career in real estate.
From the treasurer’s side, the fear of scandals deprived some corporate treasurers of the ability to make decisions
based on judgement, and relegated their activity to a video game where a few banks are summoned electronically to
compete on a transaction, and the green square on the screen needs to be clicked so as to trade at the best price. At
the extreme, there’s little attention to sound risk management as long as the compliance department is happy, and
that the fiduciary duty has been rubber-stamped. This tends to fuel a vicious circle. With prices so easily compared
and a myopic attention to eliminating any mark-up over market rate, and with the function of the FX manager
morphing into a mechanistic executioner of corporate policy instead of being the traditional guardian of corporate
financial safety, the largest corporations are pushed to save pennies while refusing to reward banks for any help.
This makes it increasingly difficult for banks to fund advisory resources, and to earmark any of them for small
clients. It also generates a proliferation of FX bucket shops that cater to unbanked corporations. These smaller shops
are more likely to have predatory execution practices, particularly as regulatory institutions focus their attention on
large banks.
One head of a sales desk at a major bank lamented a lost generation of corporate treasury professionals who have
no idea how the market works. Famously, as early as 2009, during their earnings call a US airline declared: “We
decided to stop hedging because it is too risky.” This is to date our favourite, most absurd, FX utterance.
In bank/corporate relationships where the client is large enough and the bank feels sufficiently compensated for
its complex services, very often the relationship experiences one more mortal enemy, the infamous “analysis
paralysis”. This tends to tie the finance department in a never-ending cycle of requests for banks to run numerical
scenarios for the benefit of the board of directors or the hedging committee. As mentioned, such committees do not
have the technical expertise to understand complex market situations, while the finance department does not have
the authority to make decisions. Therefore, the market keeps running away from the budget rate, sometimes
threatening earnings. The organisation tries to work out decisions for which nobody will get a promotion, and the
bankers wait for a phone call to execute a transaction that never comes.
A few years ago, the FX committee of a top chemical US corporation had decided on a rule-based system to time
its sales of euros. When the time came, the treasurer called the FX manager to order them not to sell the euros as the
treasurer felt the euro would go up. It is not clear why the treasurer should have been privy to information or
analysis that escaped the smartest hedge funds, but for sure the treasurer had the authority to stop the hedging
programme. The euro proceeded to have a great time plummeting, and the FX manager proceeded to experience a
dramatic drop in the quality of their sleep, as they were blamed for the loss in revenues.
An expert in market behaviour working at a large bank has talked of the fact that “corporations are short
volatility”. In stable markets they do just fine, but as soon as trends develop and decisions are needed, they tend to
panic and drown. Where will the endgame of this grotesque situation take corporate treasuries? In our opening
analogy, Europe enjoyed immensely the new benefits of cheap and safe commerce, but in 1346 the Black Death,
travelling from China along the newly disintermediated Silk Road, spread to Europe and killed between 25% and
60% of its population. Venice, which owed its very existence and immense wealth to commerce and banking (yes, to
banking), almost perished entirely. Disintermediation sometimes has a way of ending unexpectedly.

Searching for outperformance, a misguided use of resources

We will now transition from the relation between banks and their corporate clients to some of the questions that
corporate clients are passionate about, but in reality should not be concerned with. In a typical US corporate
treasury, there is a lot of transition. Below the very top positions, many of the people executing and analysing the
risk and efficiency of the treasury are those that are circling through all of the company’s operational areas. This is
certainly the case for FX operations. Very often the function of the FX manager/trader is performed only for a few
years, after which the individual will continue on to other areas in the finance department. This means that the
corporate FX manager, during those few months, tries to add value to an existing process that is already in place,
and in most cases has been in place for a few years. There is a “reinventing the wheel” syndrome. Of course, the
periodic rethinking of processes is a great idea, and should be part of any well-run department. However, when it is
driven by the needs of the individual to show why they matter rather than the needs of the process to reform, it can
be counterproductive in the long term.
One of the simplest ways that new treasury staff can come up with to add value is to add money to the bottom line
– ie, turn treasury into a profit centre. This means that they try to find a better way to hedge, and of course an even
grander way to document that, thanks to their findings and implementation of hedging methodology enhancements,
the firm is better off financially. Your authors have lived for years in fear of the one meeting with a corporate client
where they will ask us to teach them how to outperform the market, how to make money while hedging.
There is an absurdity to the question itself. Asking “How can I make money off my hedging activity?” seems to
imply that anybody with experience should know the answer. The usual answer that any person with any experience
in financial markets would give is: “If I knew the answer, I would be so rich that I would not be here talking to you.”
Regretfully, corporate treasury professionals who ask this question are usually serious. They truly believe that sell-
side salespersons know how people make money, or have inside information about the FX market. The fact is,
nobody on earth knows how it is possible to consistently make money in the FX markets. That the question is posed
at all hints that the firm puts its faith in corporate treasury professionals with no idea of how the markets work. The
best answer that we have concocted over the years has been along the following lines.
If you are trying to beat the Street, to pick them off, to read the market, to punt and consistently be right, do yourself and your boss a favour
and forget about it. You are in corporate treasury – you are very busy with payments, accounting, execution, forecasts, gathering of
information of all sorts. Do you stand a chance of beating Wall Street hedge funds at their own game? You can devote to FX trading two
hours per day when you are lucky. What makes you think that you can do better than people who spend 20 hours a day watching the
EURUSD market? People whose entire life has been devoted not to accounting or supply chain or reporting, but to dissecting flows and
positioning and high-frequency correlations. For your own good, do not try to fight a battle that you cannot win.

The search for outperformance often entails additional risk. It is easy to boost performance by assuming more risk
and its twin, more volatility, and the potential for larger losses. Therefore, no changes to the current process should
be implemented just for the sake of financial performance without checking the impact on financial risk or market
risk. Change is usually pitched using numerical tools. Those are generally of the two types we discussed more fully
in the previous chapter: simulation and backtesting. Corporate treasuries consult with a few banks, and a few
different hedging strategies are compared. But how is the comparison carried out?

Unearthing the secrets of history to predict the future: The backtesting elixir
A corporate treasury professional may believe that one should compare the performance of different hedging
strategies over a historical time period, ideally the whole history of the FX market. If one particular strategy beats
the others, then they have discovered an intrinsic natural rule that nobody knew before. This pure lunacy flies in the
face of anything we have ever learnt about financial markets. Nonetheless, these zealous FX managers will ask
banks to run such huge backtesting exercises and report to them the findings, which will then be the object of a
hedging change proposal to the board.
History does not precisely repeat itself, and any mispricing that happened in the past is subject to disappearing in
the future. The same is valid for studies carried out that substitute simulations for backtesting, with the same
dramatic consequences. Banks will be very happy to indulge their clients and diligently work through massive
computational exercises as long as their clients’ gratitude has a reasonable chance of translating into increased
business volume. Where banks may not see the value of your potential business, technology has democratised these
processes that many think are valuable. Systems can now backtest historically very complex strategies in a matter of
seconds. They give corporate treasuries a sense of comfort but no real advantage.
Consider the following situation. A Japanese company that has US$1 million in receivables a quarter wants to
start using participating forwards to hedge. The treasurer suggests the idea to the CFO, and the CFO asks: “How
well would that strategy have performed in the past?” The company asks its banker for the answer. The banker runs
a backtest on the Bloomberg page OVML. They model a participating forward where the client buys a yen call/US
dollar put with a US$1 million notional, and sells a yen put/US dollar call with a US$500,000 notional, out three
months, with 50% participation and zero upfront premium (see Figure 12.12).
They then run a backtest to the first business day of 2015 out for 12 quarters, and ask the computer to determine
the P&L of a forward transaction where the client sold US$1 million every three months versus entering into a
participating forward with the above-mentioned criteria.
 Pause for a second to consider how far we have come in our capacity to access and process data! Think of the
magnitude of what one needs for this process. One needs to know exactly where spot prices were on each quarter,
the entire forward curve, the entire vol surface so that the right volatility can be chosen for the strike, and then have
the computer calculate a 50% participation, three-month, zero-cost strike. And, oh yes, this capability is available to
every salesperson on earth, with the results available in a few seconds! Of course, this power is also at the hands of
the corporate treasury person, but let’s return to our point about expertise and technologically induced stress. How
can one person that is rotating through treasury for a couple of years possibly master all this and the previously
mentioned tools of VaR and stress testing?
Back to our example. Once the backtesting is done, just as with the security blanket of VaR, the backtest gives the
user one number that says this strategy would have done better or worse than hedging with a forward for the period
requested. By now you know all the pitfalls of that statement. Obviously, changing the period, amounts and
frequency of hedging can give completely different results. Nevertheless, the elegance of one number that one can
tie back to is really powerful. Look at the output in Figure 12.13.
Every single occurrence is reconciled beautifully. In the first grouping you can see that the first trading day was
the January 5, 2015, the strike for zero-cost 50% participation is 117.74. At the end of the three months, on April 5,
2015, spot was 118.99. Therefore, if one sold US$1 million at the forward, the profit would be US$9,850.07. If,
however, one had entered into the participating forward, the yen call at 117.74 would have expired worthlessly and
the yen put would have been in-the-money on half the amount: the total loss would have been US$5,243.61.
Therefore, for that occurrence the forward would have outperformed the participating forward by US$4,606.46
(9,850.07 + 5,243.61). The computer does this for another 11 quarters, and tells the user that both the participating
forward and the forward would have positive results but the forward would have generated US$27,283 more. So, I
guess this means this treasurer should always do forwards?
 Here’s a little secret. If you tell me what the general path spot will follow, I will tell you which strategy you
should implement. Recall our discussion in Part II of this book – if spot rises with your underlying exposure, do not
hedge or, if you must hedge, buy some way OTM inexpensive options. If spot moves against the direction of the
underlying, then simply do forwards. If the spot movements is range-bound in nature, then some form of a forward-
like collar strategy works best.
The above backtest summary confirms this. As the US dollar falls between December 2015 and June 2016, the
forward outperforms. The reason we explore alternatives is because we do not know where spot will be and how it
will get there. It’s unlikely that it will repeat a previous path exactly and, even if it did, we certainly cannot predict
which one. Once again the company should hedge based on what suits its business best, and leave the business of
getting the optimal FX rate to those whose job it is to do that day in and day out.

The source of excess profits

Here’s the problem with the above reality. When corporations focus on their core business and FX hedging priorities
are subservient to the main business priorities, there is FX money left on the table. This is an odd situation in the FX
markets. As we saw in Chapter 2, corporations (non-financial participants, 7%) and asset managers (institutional
investors, 16%) constitute a small part of the total FX volume on any given day. Yet an argument can be made that
without that “real money”7 having an organic need for foreign exchange, there is no FX market. The odd thing is
that most of the real money is not interested in maximising their FX profits, as a rule, but to invest in international
stocks or bonds based on their investment criteria, or to sell, manufacture or offer services consistent with their
corporate charter. FX is a conduit to those ends, not an end in itself.
On the other hand, the market participants running this conduit – banks and hedge funds – are in the business of
extracting maximum profits from FX. Consequently, a US asset manager may simply buy a Brazilian corporate
bond and hedge using forwards or a cross-currency swap. This way they eliminate the FX risk and are exposed just
to the bond returns. A bank will get the FX (Brazilian reals) needed, possibly from a hedge fund that thinks the real
will devalue at some point in that timeframe. The bank will want to buy reals cheaply from the hedge fund and sell
them more expensively to the asset manager. The hedge fund may choose to keep the risk (short real) of that FX
forward on its books for a period of time in anticipation of the FX rate moving in its favour.
Here is some insight as to why, on a volume basis, real money is a small part of the total. Let’s assume the deal
was for R$30 million and then a series of transactions ensue. The total volume would be as follows.

Asset manager to bank: +R$30 million.

However, the bank would have +R$30 million with the asset manager.
And another +R$30 million with the hedge fund.
And the hedge fund would report another +R$30 million with the bank.

In reality, there may be a few more hands exchanging the risk back and forth causing total volume to skyrocket off
that one deal for R$30 million. The point is, if the real money player had no need to buy reals, none of the other
transactions would have happened.
In the following discussion, we will briefly explore the spectrum of thought and anecdotal implications of this
behaviour. We now know that real money players need to buy and sell foreign currencies to stay in business.
However, do they care where the FX rate is? On one level, the answer is no. For example, a Texas oil drilling
company will consider buying a German drill bit at the prevailing EURUSD rate. If the cost, in US dollars, means a
satisfactory return on investment, they will do it. Otherwise, they will look for a cheaper drill bit, maybe from
China.
What if this company is much more sophisticated and proactive, and wants to maximise its engagement with the
FX market? The company will try to enter into this FX transaction at the absolutely best rate. It will stretch, even
sacrifice, some of its business-driven criteria so that the very best rate in FX can be achieved. To do that, it has to
invest in resources to understand the FX market. For example, it may hire an FX trader or expert, or buy charting
and predictive analytic systems that will help with the timing. It may hire an economist to try to understand how the
German economy will develop versus the US economy. It will cultivate its banking relationships so that it gets as
much information as possible from the bank’s FX salespeople and traders. This company will compete with any FX
hedge fund or bank dealer. No excess profits here. However, if it decides that it will execute the FX transaction
around the need for drill bits, it will offer the opportunity to the executing dealer, who most likely has extensive FX
resources, to execute at a more favourable rate. The difference between the rate at which the company executes the
transaction and the rate at which the dealer was able to unwind their risk is what we call excess profits. Lest you
think: “How much money can there be in that business?”, think again. Every major bank and hundreds of hedge
funds are in the business of extracting many billions of dollars in profit per year.
Even in our simplistic example, it is obvious that an oil-drilling company with a few international transactions has
no business investing resources to become an FX expert. Is there a middle road? The same company may anticipate
needing those higher-quality drill bits a few years out, and may look for opportunities to buy euros cheaply and keep
them for the chance to buy the drill bits. This will give it a competitive advantage over its competitors who did not
plan ahead should the euro rise. Suppose, however, that in the interim the euro continues to collapse versus the US
dollar. Then the purchase of euros for a drilling company constitutes a foreign exchange loss, and the company is
now at a competitive disadvantage. Its competitors can buy the drill bits more cheaply since they do not have an FX
loss to contend with. The above example underscores the reason most companies engage in FX transactions only to
the extent they absolutely have to. Imagine the CFO of this company having to go before the board and explain why
they were speculating in FX. We explored this hedging conundrum in earlier chapters.
The more realistic situation is that FX risk for corporations arises from a variety of different categories and is not
typically or directly related to the company’s core business. Banks and hedge funds, and the majority of the volume
transacted, are relevant at the margin. Although this is what in fact happens, as we will see in the following sections
most corporations refuse to accept that they are leaving some money on the table and consequently keep trying to
get at it. Corporate FX risks are identified, separated and assigned to corporate treasury to manage. Overseeing
committees are set up to prevent speculation and adhere to core principles.
Most treasury areas are impacted by FX in three broad categories.

Transaction risk: The risk linked to variations in the expected functional currency cashflows.8 Typically, this would
involve payables, receivables, repatriation of profits, etc.
Translation risk: The risk that, when translated into the parent entity’s, the balance sheet of subsidiaries will deviate
from expectations due to the change in FX rates from one period to the next. Typically, this would only relate to
the assets and liabilities being translated. Most companies calculate the translation at an average rate over the
last quarter of the reporting year. Income statements are translated at the prevailing rate at the time of
translation. This risk is related to the impact of FX changes on earnings.
Economic risk: The risk that the value of the company will be affected by the movement of FX rates as the value of
future revenue will be impacted by FX rates.

CORPORATE HEDGING BEST PRACTICES

In this section, we get down to the specific questions that arise as a corporate treasurer tries to hedge. We explore
ways to think about the risk, while offering some tools to help get to real-world answers and implementation
solutions.
A little boy said to his mother: “I wish there were no market risk and no financial risk – just as in science, where
everything can be calculated with precision, I want the same thing for financial statements.” The mother kissed her
boy on the forehead and said: “Yes, you are right, how nice if there were no uncertainty.” I am told this story is
apocryphal because little kids don’t talk that way, but I have a different opinion. I think the kid can teach us a lot
about financial statements, but the story is apocryphal because the mother is wrong. Risk and uncertainty
intrinsically inhabit science and physics just as much as they inhabit the financial world. Randomness is as
unavoidable in the markets as in the determination of the position of electrons. The same equations governs the
vagaries of the EURUSD exchange rate and of subatomic particles. Once grown up, the kid will represent a new
breed of risk manager, one who embraces risk and tries to understand it instead of avoiding it. Here, we briefly
analyse the challenges facing a corporate treasury FX manager from a “best practice” point of view.
We begin by analysing the goals of the FX manager, before considering all the factors that complicate the job,
including market volatility and accounting considerations. We will end our survey by listing the different
approaches that are common in the treasury community, and identify a few “state-of-the-art” elements of the
workflow. In future chapters, we will establish how risk management can be made more robust and less fragile with
the aid of numerical tools. In our analysis, we will take the point of view of an FX manager at a US-based
multinational company with multiple foreign entities, not all of them US$-functional. We will also assume that the
company adopts generally accepted accounting principles (GAAP), but our concepts do not significantly depend on
any particular accounting standard.

The “big hedge theory”: What does it mean to “hedge”?

Here’s the partial description of the FX manager’s job and their challenges:

ensure accounting standards are applied correctly;

hedge cashflows due to revenues, expenses and net investment; and
report to the treasurer or the CFO, and align the FX management activity with their priorities (as briefly mentioned,
these two people can have conflicting priorities, as we will see in our discussion about CFaR versus EaR).

A few of the above tasks imply that hedge accounting treatment will be preferred when hedging. However, at the
CFO level the attention is usually on the earnings impact of the FX fluctuations. In the end, earnings predictability is
very important for a very practical reason. Every CFO wants to avoid situations such as the examples listed below,
which regularly happen every year the dollar move is significant.

Coca-Cola, 10-Q filing, June 27, 2014: During the three months ending June 27, 2014, foreign currency exchange
rates unfavourably impacted on consolidated operating income by 5% due to a stronger US dollar.
General Electric, 10-Q filing, June 30, 2014: The total pre-tax amount in accumulated other comprehensive income
(AOCI) related to cashflow hedges of forecast transactions was a US$121 million loss on June 30, 2014.
Apple, 10-Q filing, July 23, 2014: The company may choose not to hedge certain exposures for a variety of reasons
including, but not limited to, accounting considerations and the prohibitive economic cost of hedging particular
exposures.
Google, July 25, 2014: The notional principal of these foreign exchange cashflow hedges was approximately
US$10.0 billion and US$9.4 billion as of December 31, 2013, and June 30, 2014, respectively. Also, they had an
FX risk management programme designed to reduce their exposure to fluctuations in foreign currency exchange
rates. However, this programme would not fully offset the effect of fluctuations on their revenues and earnings.
Microsoft, 10-Q, July 31, 2014: Net losses on derivatives increased to US$328 million due to higher losses on FX
contracts, losses on equity derivatives as compared to gains in the prior period.
Berkshire Hathaway, 10-Q, August 1, 2014: BHRG’s underwriting results included pre-tax foreign currency
exchange rate losses of US$139 million in the first six months of 2014.
Ranbaxy Laboratories: They may post an FX loss of as much as US$200 million in the September 2013 quarter due
to wrong-way bets on currency hedges, analysts say. That loss may take their cumulative loss on account of
derivatives to over US$1 billion.

Hedge accounting: A very brief introduction

At the very top of the computational challenges faced by treasurers is the maddening complexity of claiming
favourable accounting treatment for hedges, known as “hedge accounting”. The original official rulebook, known as
FAS 133, is 873 pages long (2004 edition). A friend of ours, Matt Daniel, gave me a brand-new copy of this book
around 2005. He rightly insisted that, as a professional in this market, I needed to know it inside out. He kept asking
if I had read it, and always referred to such and such a page anytime I asked him complicated accounting questions.
The truth is I tried, but my brain is just not wired that way. I would much rather gnaw off my right arm than read this
book cover to cover. The other author of this book has read it all, so you can be comforted that this brief description
is accurate.
 As the celebrated physicist Richard Feynman purportedly and eloquently quipped about quantum mechanics,
“Nobody understands quantum mechanics,” we also could say that “Nobody understands hedge accounting.” Given
the complexity, sometimes multiple interpretations, and a constantly changing domestic and international accounting
landscape, companies have often reached out to their banks for help. And banks help the corporates lest the latter
give up the hedging business entirely. Large banks have hired the top accounting gurus to assist their clients in
running the rocket science calculations involved in hedge accounting. Although we will barely scratch the surface of
hedge accounting in this book, in what follows we want to provide a general understanding of what it purports to do
and how it affects corporate hedging policy. The truth is, many corporate treasurers have spent countless hours and
days on this activity.

Suppose a company needs to hedge forecast revenues from Europe that will be realised in a year’s time. One
potential hedge would be to sell euros forward. Until 2001, the accounting standards allowed you to not book the
hedge, which is a derivative (as in “it is derived from spot”). When the underlying revenue exposure was recognised,
the company would then book it together with the payout of the hedge in earnings and, ideally, the fluctuation of one
would counterbalance the fluctuation of the other (see Figure 12.15).
With the advent of the FAS 133 (now ASC 815) standard in 2001, the revenues are still not booked until
recognised, but the hedge needs to be booked and marked-to-market for its entire life, its value fluctuations flowing
into earnings all alone (see Figure 12.16). This had the effect of creating additional volatility in earnings, as the
hedge re-measurement was not counterbalanced by an opposite re-measurement of the underlying exposure, which
is still not booked and its market value fluctuation not reported. The paradox originates from the fact that the very
act of hedging creates the volatility that it was supposed to eliminate in the first place.
The solution was to formally link the hedge and the exposure in a hedging relationship. To qualify the whole
structure for hedge accounting treatment, there is a need to follow a set of complicated rules and calculations that
must be satisfied (see Figure 12.17). These rules are there to validate what the hedger purports to be the case. Once
this is achieved, the re-measurement impact of the hedge can be booked into “other comprehensive income”, a part
of equity, away from earnings, until the exposure is recognised. Then, and only then, are both the impact of the
hedge and the exposure recognised together and at the same time into earnings. The idea is that equity is typically
such a massively larger number that some relatively small changes in derivative valuation will be inconsequential to
the company’s operations, and equity analysts will not focus on it.

There you have it in a nutshell. Implementation is incredibly more complex and laborious. Today, all those tasks
are being commoditised by data and risk system providers, and there are many systems available. However, for
demonstration purposes, we will use the Bloomberg HEFF function. Corporate treasury staff can set up a hedge
accounting effectiveness testing designation (see Figure 12.18). Then the Bloomberg terminal can create the testing,
schedule monthly updates, and store all the reports ready for the auditor to review.
C-suite tug of war: EaR versus CFaR
We touched on this topic previously. Here we offer a framework to arrive at the optimal answers. Two very
important factors are common to the vast majority of US multinational corporations: the parent company has non-
US dollar functional subsidiaries; and the hedging activity is typically geared to be hedge accounting compliant.
Under the above assumptions, only the entity that bears the financial risk can hedge such risk and receive hedge
accounting treatment. Let’s rephrase this important fact. If a US-domiciled company has a euro-functional
subsidiary with an expense exposure of US$80 million, only the European subsidiary can receive hedge accounting
treatment when hedging this exposure. The hedge consists of selling euros and buying US dollars through a forward
contract or some other derivative. If the same European entity also has expenses of €100 million, those do not
represent an FX risk to the euro-functional entity, as the expense is denominated in the same currency as the
functional currency of the entity that assumes the exposure. Therefore, no hedge accounting treatment is possible for
that €100 million. The parent company can enter into a forward contract to buy euros/sell US dollars, but such a
contract would:

not receive hedge accounting;

not represent a cashflow hedge for the European entity;
be marked-to-market into earnings directly and immediately each quarter; and
indeed hedge the consolidated earnings of the parent company.

In other words, the parent company cannot hedge the translation exposure represented by the €100 million and
achieve hedge accounting treatment. The company as a whole can hedge the US$80 million cashflow exposure and
receive hedge accounting treatment. This paradoxical situation is the key to understanding a few important facts
about corporate risk management. First, that a corporation can receive favourable accounting treatment if hedging
cashflows but not if hedging earnings. This has been a topic of hot debate, as many corporations have tried to find
creative solutions to the problem of hedging earnings. A large Fortune 500 company used to apply hedge accounting
to hedge earnings, with the blessing of a partner at the local office of one of the major accounting firms. They
refused to hear anybody from a bank telling them they were in violation of FAS 133. Looking at it from another
point of view, the problem at its core amounts to saying that the FX risk is not the same for cashflows as it is for
earnings. Therefore, one cannot hedge them both at the same time. Second, in principle, priority should be given to
hedging earnings. However, there is no hedge accounting for hedging earnings, no easy solution and consequently
we have a problem.
 You may wonder: Why do we say that hedging earnings should be prioritised? The answer to that question is the
list of casualties described two sections back, where corporations have to report big FX losses. FX losses only make
headlines when they end up in the 10-Q and affect earnings. The research analysts only focus on earnings; the
investing public only focuses on earnings; in the vast majority of cases, the CEO and the CFO focus on earnings.
The FX manager is in the uneasy predicament of looking at cash risk, ready to hedge it with accounting-friendly
instruments, while at the same time knowing that the only way to capture the attention of senior management is to
hedge earnings.
Why not hedge the euro expenses, forego hedge accounting and make the analysts happy? Assume a company
hedges quarterly earnings out to two years. The aggregate notional of all those outstanding hedges would correspond
to two times yearly earnings. This is a huge amount. Every quarter, that huge two-year hedge would be marked-to-
market. Then, the quarterly re-measurement of this huge notional would flow into earnings at each quarter since
there is no hedge accounting. This will most likely create huge swings that may not be so easy to explain. In other
words, the cure would be worse than the illness. Is there a solution to the quandary? Yes and no. As in many cases,
the only solution is a compromise. Senior management and FX management should sit down and look at all their
options, and select to implement the solution that best addresses CFaR and EaR. The weapon they have to their
advantage is the hedging ratio – essentially, how much to hedge of the euro-functional entity’s US dollar exposure,
which of course receives hedge accounting treatment. Let us have a look at what their options are by using the chart
in Figure 12.19.
The best course of action the company can take is to look at Figure 12.19 and decide on what point of the efficient
frontier they want to be. Each point corresponds to a specific hedging ratio for the US dollar exposure, a specific
CFaR number and a specific EaR number. Those are calculated as the VaR of cashflows and the VaR of earnings.
Let’s go through a few examples. Our notation for these examples will be slightly different from the usual: €70 will
mean an exposure denominated in euros and with a magnitude equal to US$70 million. The number will always
denote the equivalent number of dollars, so that we can easiliy compare magnitudes, and the currency will indicate
in what currency the exposure is denominated. This will be extremely confusing at the beginning, but is by far the
best method to analyse the situation (for this approach we are indebted to Matthew N. Daniel).

Example 1: Simple case

US parent has euro-functional subsidiary;
sub has US$80 million in third-party expenses; and
sub has €100 million in third-party expenses.

Scenario 1 – unhedged:
CFaR is US$80 million; and
EaR is €100 million.9

Scenario 2 – Subsidiary hedges the US$80 million cashflow:

CFaR decreases to zero;

EaR increases by US$80 million because now the subsidiary has a forward contract where they sell euros to buy US
dollars, which will create P&L in translation; the original US dollar liability does not create any P&L in
translation.

Example 2: Intercompany exposures

euro-functional subsidiary has ¥100 million expenses to the Japanese yen-functional subsidiary; and
yen subsidiary has ¥100 million revenues from the euro subsidiary.

Scenario 1 – no hedge:

yen subsidiary CFaR is zero;

euros parent CFaR is ¥100 million;
total CFaR for entity is ¥100 million; and
EaR in total is zero, as the euros and yen subsidiary net out their translation risk.

Scenario 2 – euros subsidiary hedges the ¥100 million:

yen subsidiary CFaR is zero;

euros subsidiary CFaR is zero;
total CFaR is zero; and
EaR is €100 million plus ¥100 million.10

In Figure 12.20, all the numbers are expressed as US dollar equivalent.

Example 3: Intercompany with use of gross exposure

euros subsidiary has £100 million revenues and £80 million expenses, both to a third party; and
sterling subsidiary has £50 million revenues from a third party.
Scenario 1 – no hedge:

euros subsidiary CFaR is £20 million;

total CFaR is £20 million; and
EaR total is £70 million in revenues.

Scenario 2 – euros subsidiary hedges the net £20 million:

euros subsidiary CFaR is zero;
total CFaR is zero; and
total EaR is £50 million, coming from the UK sterling subsidiary, which they are not allowed to hedge, plus €20
million due to the euro entity hedge.

Scenario 3 – euros subsidiary hedges £70 million versus US dollar as they use their right to only hedge revenues and
they have revenues of £100 million (see Figure 12.21). Keep in mind they sell £50 and buy US$70, so this trade
does not get hedge accounting:

euros subsidiary CFaR is £50 million (expenses) plus US$70 million;11

sterling subsidiary CFaR is zero;
total CFaR is £50 million (expenses); plus US$70;
euros subsidiary EaR is £50 million (expenses);
sterling subsidiary EaR is £50 million (revenues); and
total sterling EaR is now zero, dollar EaR is zero because the US$70 is not EaR, as the parent is US dollar-
functional (see Figure 12.21).

Example 4: Third party

euros subsidiary has ¥100 million expenses to a third party.

Scenario 1 – no hedge:

euros subsidiary CFaR is ¥100 million;

total CFaR is ¥100 million; and
EaR total is ¥100 million.
Scenario 2 – euros subsidiary hedges the ¥100 million (with a EURJPY trade):

euros subsidiary CFaR is zero;

total CFaR is zero; and
EaR is €100 million, by hedging the euros subsidiary has changed the risk from yen to euros (see Figure 12.22).

Now the question is: which risk situation is better? The corporation might also wish to consider forward points and
volatility.
 Of course these are just simple examples. The reality is that exposures will be more complex as they are
composed of a large number of different currencies, and therefore the above figures will look something like that in
Figure 12.23. In Figure 12.23, the efficient frontier is a curve that is calculated by iterating, trying one after another
across many different hedging scenarios. Obviously, there is a high number of possible hedging combinations. For
every possible CFaR level, we then take that hedging combination such that, at the same CFaR level, the lowest EaR
level is achieved. Those hedging scenarios in aggregate will represent the efficient frontier.
When facing a choice of hedging portfolio, the company should only consider those that lie on the efficient
frontier. For any other choice that is not part of the frontier, there is an alternative that has either lower CFaR or
lower EaR. What will the final choice of hedging mix be? This is a choice that is ultimately left to a discussion
between FX manager and CFO or treasurer. There is no wrong answer as long as the answer lies on the efficient
frontier. It is a choice. With the use of this methodology, both cashflow risk and earnings risk are under control and
the company is sure that it is hedging efficiently.

Hedging policy processes and discipline

Once the company has decided at a high level where on the efficient frontier it wants to be, the next set of decisions
falls to the FX manager. We will now discuss some of the considerations that will help them arrive at most efficient
outcome for that firm’s risk/reward profile.
Many US companies with revenues from Europe have been riding a rollercoaster since the days of the euro
overvaluation in 2008, when it touched 1.5991. By mid-2018 the level was closer to the bottom half of the range
(see Figure 12.24). Most companies that regularly hedged their euro-forecast revenues every January found
themselves, on average, hedging at lower and lower rates after 2008 and receiving lower US dollar revenues
independently of how their business in Europe was doing. Many wonder: if hedging is supposed to protect your
revenues, how come it does not really work? To answer this question, we will address the differences among a few
hedging approaches, but first of all we need to state the obvious. Not many corporates have the luxury of being able
to hedge with an infinite time horizon, so for each currency that is on a downward trend there comes a time when
you have to re-hedge and you will feel the pain. There is nothing you can do about that.
 The second thing we should do is to examine the reasons of the hedging activity. If you ask around, people will
tell you that the goal of the FX management activity is to reduce the FX risk for the company. The problem is to
define such risk in a measurable way, and then define the FX hedging activity to minimise it. The truth is many US
corporates do not try to measure their FX risk before they hedge. Furthermore, after they hedge they do not run a
comparative analysis to demonstrate quantitatively the benefits generated by their hedging programme. This has two
disadvantages. First, they are flying blind, without anybody telling them whether or not they are doing a good job of
hedging; second, their hedging activity can sometimes have a negative, unanticipated effect, in the sense that the
firm would have been exposed to lesser FX risk had the hedging programme never taken place.

Reducing FX risk
Let’s begin this analysis by exploring what it means to reduce risk for US corporates. It should come as no surprise
that not everyone is on the same page. In situations where the aim is to defend a budget rate that is set at the
beginning of the year, then the best hedging strategy is a static strategy. This consists of hedging the whole year in
January of that year, and then doing nothing for 12 months. Repeat the exercise the following year. This is
illustrated by Figure 12.25 and Table 12.1.
 In other very common situations, what the company defines as “reducing FX risk” actually means reducing the
year-on-year impact. In that case, the story is different. As the comparison is done year-on-year, the exposure needs
to be hedged exactly one year before. We then have the implementation scheme shown in Figure 12.26 and Table
12.2, which is commonly referred to as a “rolling hedge”. In the table, the columns represent the date the risk is
materialising, and the rows represent the date the risk is hedged.

The third very common situation corresponds to the company wanting to decrease the quarter-on-quarter volatility
of effective FX rates achieved. This is more difficult and requires a small digression.
 Let us examine the results of a rolling hedge on the effective rate of a EURUSD revenue exposure from the point
of view of the US parent company. Figure 12.27 shows the effective exchange rates for a company exposed to
EURUSD that does not hedge (in white) and for a similar company that hedges using 12-month forwards (①). As
highlighted by the curved segments, all changes in spot will be reflected a year later. This effectively only buys time
for the hedger, but does not substantially change the volatility of the effective exchange rate.
 A rolling hedge methodology generates an effective rate that is very similar to a delayed spot, in our example
shown above, one year later. This is because changes in spot dominate changes in the forward points. In essence, a
rolling hedging programme just delays the inevitable FX fluctuations by one year. It is not a coincidence that, in the
example above, the volatility of spot is 10.94% and the volatility of the effective hedging rate is 10.69% – ie,
roughly the same. We can conclude that a rolling hedge is great for reducing year-on-year deviations from a budget
rate, but does not really decrease the volatility of the effective rate.
So the third popular hedging methodology will actually achieve a decrease in volatility of the effective rate, and is
called “layered hedging”. In our example of quarterly hedging with a tenor of one year, here is how the hedging is
done: every quarter, four hedges are traded, each with a notional that covers 25% of the annual exposure, and with
maturities of three, six, nine and 12 months (see Figure 12.28). On the other hand, a table format might be easier to
read (see Table 12.3).

Why the complication, you may ask? The advantage is that the volatility of the effective rate decreases
substantially, as is visually evident in Figure 12.29. In numbers, the volatility of spot is 10.94%, as we mentioned
before, while now the volatility of the effective hedging rate is just 2.43%. In exchange for a heavier hedging
activity, the company significantly reduces the volatility due to FX rates. The idea is simple and corresponds to the
old principle of averaging the entry point.
Essential tools to design the best hedge
Assume that after closer examination, all questions about the hedging methodology have been answered in a way
that reflects the company’s objectives. Now the FX manager needs to get down to the execution specifics. What
value dates do they need to hedge, which currencies and in what proportions? This section will go into the concepts
and theoretical constructs we need to consider, and then come up for air with some simple practical examples that
we build into a realistic hedging approach.

VaR revisited
At the end of Chapter 11, we defined VaR and went through some sophisticated but practical tools to help manage a
trading portfolio. In this section, we will examine the same concept in a more intuitive way and examine how it may
apply to corporations. VaR is without doubt the most widely used financial measure of risk, and plays a fundamental
role in how banks and corporates make hedging decisions. Let’s revisit it with a brief description.

For any FX exposure, it is possible to simulate scenarios and calculate for each the amount of losses (gains) in the
exposure portfolio. For example, if the exposure is short US$8 million versus Brazilian reals with a time horizon of
one year, a histogram of the losses or gains of each scenario would look like that in Figure 12.30. Of course you will
recognise your old friend, the Gaussian curve (just to keep things simple we are using the Gaussian here, instead of
the more correct lognormal).
The black curve is the full distribution of our possible gain/losses at the end of one year. The grey part, which is
superimposed, covers only 5% of the black area: where the entire area below the curve measures 1.00, the grey area
measures 0.05. This means that there is only 5% chance that a typical scenario will end up creating a loss greater
than US$1.8 million (find that number on the horizontal axis). This is precisely the definition of VaR. In this
example, we would say that the 5% confidence interval VaR for our exposure is US$1.8 million with a time horizon
of one year.
The interesting thing is that the notion of VaR over the years has been twisted in meaning. Today, many people
think that the VaR is the worst possible loss the portfolio can have, which is not true. First, the worst thing that can
happen is that the Brazilian real devalues down to zero (think of a Zimbabwe-type collapse), and in this case the loss
would be a full US$8 million. Second, from Figure 12.30 it is obvious that there are some scenarios for which the
loss is greater US$1.8 million. Clearly, VaR is not the worst possible loss; it just tells us that, with 95% confidence,
our loss will be smaller than US$1.8 million. In other words, in 95% of the scenarios we have studied the loss is less
than US$1.8 million. As 5% means “one chance in 20”, or “once every 20 times”, you should expect to have a
yearly loss greater than US$1.8 million once every 20 years.
Here is an interesting thought. If you took the 4% worst scenarios in Figure 12.30 and assigned them a loss of
US$10 billion, that would not change the VaR by a cent.12 However, it would certainly change your attitude towards
the risk you are taking that generates this risk distribution, would it not? That is why VaR can at times be
misleading.
One more consideration on VaR: banks, hedge funds and corporations routinely aggregate exposures and compute
the VaR of portfolios. During the financial crisis of 2008, the blind use of VaR as the only de facto risk-management
tool was decried as one of the factors that caused the near-collapse of many a financial institution. Even earlier, in
1998 the use of VaR by Long-Term Capital Management gave the fund excessive confidence in its approach to the
Russian bond default, which ultimately bankrupted the fund. Despite enlisting the services of Myron Scholes and
Robert Merton, the inventors of the second-most popular formula in modern history (the most popular being E = m ·
c2), the formula to value option prices (which by now you will have learned very well from previous sections), the
fund calculated VaR based only on the last two years of data history.
Banks tend to use 99% VaR, which corresponds to 2.31 standard deviations from breaking even. Corporations
usually use 95%, which is associated with 1.645 standard deviations. The use of 97.5% is also popular,
corresponding to roughly two standard deviations. Generally, there is a portfolio effect that makes the VaR on the
portfolio smaller than the sum of the VaRs on the individual components, as discussed earlier in this book. This is
also why regulators allow banks to manage their risk on a portfolio basis. Nonetheless, many have pointed out that
this is not a hard rule. In other words, it has been discovered, to the horror of risk managers, that there could be
portfolios that have a VaR greater than the sum of the VaRs of their components (see Chapter 11, Tables 11.2 and
11.3 and the related explanation).
To conclude this second quick crash course on VaR, a final puzzle. We often hear about tail risk hedging, which
is the practice of eliminating only the risk of large adverse market moves. The whole idea is to be ready to weather
some uncertainty, but to spend a little insurance money to cover “market crash”-type events. For example, a
company with revenues in euros might want to purchase euro puts to protect itself in case the currency falls below
current level. Assume the euro is near 1.20. The problem is that euro puts are expensive, so it makes sense to
purchase protection that only kicks in if the euro falls below, say, 1.07, which is a much less likely scenario. The
idea is that the company can weather a small decline in the US dollar value of euro revenues, but not a large decline
such as would happen with EURUSD at 1.07. The 1.07 level has a market-implied probability of 5% to happen in a
year’s time, when volatility is approximately 9% and the spot and forward levels are about 1.20.
The irony of the story is that if the company decided to hedge such 1.07 risk, which corresponds to its VaR level
(5% chance, remember?), the VaR would not be reduced, despite the expenditure of option premium. How can it be
that I spend money to buy insurance to reduce my risk, and my risk is the same? I shall now leave you to the
pleasure of meditating on such a quandary.
The explanation of the paradox is that the notion of “risk” is not perfectly captured by the 5% confidence level
VaR. It is not captured by any confidence level VaR for that matter. The choice of 5% is just an arbitrary number.
Of course, a euro put struck at 1.07 would reduce our risk – everybody knows that! Everybody also knows that if
EURUSD only declines to 1.10, then the euro put struck at 1.07 is wasted and useless. However, we could not know
that beforehand. Anyway, it was still a great thing to know that we were protected against any decline of the
EURUSD exchange rate below 1.07. We need to be cognisant that the calculation of VaR is very useful, but also
that it is not possible to boil down a complex concept such as “risk” to just one number or one measure.
There is another way that a hedger can look at options as a tool to reduce VaR. Suppose the hedger is long a euros
position versus US dollars with a one-year time horizon, and that this analysis is run on the day of the highest
EURUSD closing ever: April 22, 2008, at 1.5990 (or 1.5991, this is debatable). The first impulse may be to hedge
with a euro put option, but the hedger cannot decide the strike level. A higher strike causes a greater VaR reduction
(as long as the strike is higher than the VaR level, as we have seen in detail with the above paradox). However, for a
higher strike the premium will also be higher. For each strike level, we can calculate a ratio between the reduction in
VaR and the premium as:

We can chart that ratio in Figure 12.31. To maximise the ratio between VaR reduction and the premium, your best
bet is a strike that sits about 1.15 standard deviations below current spot. With a spot reference of 1.5990, the
optimal strike is 1.4200, using a Black–Scholes pricing framework and a one-year implied volatility level of
10.32%. Again, we see that any strike below the VaR level does not reduce the VaR.

How correlation impacts on the hedging decision: A theoretical foundation

The macroeconomic environment at the time of writing is characterised by low rates and low volatility. For those
corporates that have portfolios of exposures to FX rates, this situation might seem favourable due to the fact that
hedging cost, in the form of forward points and volatility are at low levels. Nonetheless, there is a third, very
important component that needs to be accounted for: correlation. For example, the EURUSD/USDBRL correlation,
usually a number in the range of negative 60–70%, has been precipitously moving towards zero, as can be seen in
Figure 12.32 (the correlation is on the bottom panel in the figure).
 Is this good news or bad news for a corporate with an FX portfolio exposure? Consider an example such as the
one shown in Table 12.4, which represents a sample forecast revenue exposure.
The question that we will answer is whether the choice of hedging strategy will change in the presence of a shift
in correlations.
The choice of hedging strategy is not always straightforward. In a theoretical case, the FX manager might just
hedge all the exposures and reduce the VaR to zero, but things are not always that simple. The most common factors
that prevent a simple “hedge all” solution are the following.

Hedging policy: This can reduce the hedging options by forbidding the hedge of specific currencies. Some
corporates are obligated by policy to hedge all currencies no matter what. If reality makes that directive
untenable, we’ve seen companies buy ridiculously OTM options just so they can say: “Yes, we hedged.”
Option premium: In some cases, the hedging is mandated for execution via options and a budget is fixed for the
aggregate premium to be spent to buy those options.
Cost of forward points: Some firms consider the forward points a cost, and are therefore keen to reduce such cost.
In the example above, it might not be possible to hedge the Brazilian reals exposure just because the forward
points are very high due to the elevated level of interest rates in Brazil relative to developed countries. Hence,
the potential need for an alternative hedging solution.

What is the best hedge to put in place given the constraints outlined above? Where do we start? Well, let’s decide on
the most relevant questions to ask.
The first question is: “Suppose that the company in the above example decided to hedge with vanilla ATM
options. Which choice of hedged components gives you the greatest aggregate reduction in VaR for each dollar of
premium spent?”
The answer is the component with the greatest correlation with the aggregate portfolio. The respective
correlations are shown in Table 12.5.
In this case, if there is one component to hedge, it’s Brazilian real, with a 75% correlation between USDBRL and
the portfolio. If you are allowed to create a portfolio of vanillas, then the notionals on those vanillas should have the
same weightings, up to a scaling factor, as the incremental value-at-risk (iVaR).14 In our example, the weightings
that you would choose are shown in Table 12.6.
The above exercise is only correct for relatively small hedges with respect to the aggregate exposure. To make
things simple from a visual point of view, we will reduce the exposures to just the euros and the Australian dollar for
now. Figure 12.33 shows a simple example for a euros and Australian dollar portfolio.
 Keep in mind that, in some cases, the result would be different if instead of maximising the decrease in VaR we
were trying to maximise the decrease of the portfolio variance.
Given that there are many different measures of risk, let us examine briefly the differences between VaR,
portfolio variance and expected shortfall (or conditional VaR). We start with revising some definitions.

VaR is the minimum loss that affects the portfolio when limited to the set of the worst 5% of all possible outcomes.
The choice of 5% is arbitrary, and many people use 1% VaR or other choices. To be more conservative, one
would choose a low number.
In CVaR, examine the worst 5% of all possible outcomes. The VaR is the lowest loss of all of those, and the CVaR
is the average of all of those. The notions are similar, but there are at least two important differences. First, the
CVaR of a portfolio is less than the sum of the individual CVaRs of its components. You might have been under
the assumption this had to be the case for VaR as well, based on instinct. However, this is actually not the case.
Recall the example we went through in Chapter 11. Second, VaR only reflects the potential losses associated
with the most favourable 95% of possible scenarios. VaR is totally blind to the magnitude of losses in the
bottom, say, 2% of scenarios. Not so for the CVaR as it is made of an average that includes those scenarios.
The variance of the portfolio is defined in the following way:

The Greek letter sigma to the power of two here is the standard notation. The integral is taken over all possible
scenarios (sometimes this corresponds to all possible future values of the underlying, or over all possible future
returns). The function p is the probability density function: it tells us how likely a scenario is (in the case of only
one currency exposure this would then become the lognormal probability density function). The Greek letter mu
(μ) is the expected portfolio value. Keep in mind this is just the gist of it, in a rather non-rigorous mathematical
way. The variance is the square of the standard deviation. If the portfolio is composed of one currency
component, then the standard deviation (of the portfolio returns) is also called volatility. But you remembered
that, right?
Let us now look at a comparison of the three equally virtuous goals of reducing VaR, CVaR or variance. First of all,
variance is a measure of risk that does not specifically look at extreme events. You might be able to reduce the
variance of a portfolio by limiting outcomes with relatively small gains and losses, and leave extreme outcomes
unchanged. For a hedger, there might be a problem with that. Usually, people do not really think of these three
measures of risk as separate definitions, regretfully.
Now, a very important point: if you are calculating these three risk measures using a Black–Scholes model where
the returns are assumed to be lognormally distributed, then minimising variance, VaR or CVaR will give you the
same result. This is because you are using our beloved normal distribution (the bell curve), and the shape of the
extreme outcomes of the distribution is always the same.
If, on the other hand, you are using, for example, a historical distribution of returns, or maybe a market-implied
distribution derived from option prices, then the shape of the extreme outcomes in the histogram of the returns may
be more or less thick, or jagged or peculiar in many ways. It will not be a smooth and regular curve. The calculation
of the variance will be substantially blind to the particularities of the historical distribution, but the calculation of
VaR and CVaR will not. Keeping with this example, many will contend that CVaR will give you a truer picture of
the risk of extreme outcomes, while VaR will only really give you information about the risk with a 5% chance of
happening. They will claim that CVaR is a superior notion.15
There are a few considerations that apply to basket of exposures. We cannot enter into the details here, but just
want to highlight where the trouble starts. Suppose that you are considering VaR and variance for a basket of
currency exposures. Also assume that you want to move away from a standard Black–Scholes approach and
represent the fat tails of the distribution just as the market is pricing them. Each currency in the basket has its own
specific probability density function (the bell curve) that is shaped in an individual way to reflect the idiosyncrasies
of that currency market. This probability density is used to price options in that currency, and everybody is happy.
When it comes to combining currencies into a basket, some issues arise. For example, when modelling the
behaviour of the basket, assume that currency 1 moves lower and therefore, according to its own volatility smile, the
implied volatility of currency 1 rises. Currency 1 will have a specific correlation with currency 2, so maybe it will
become very likely that currency 2 also moves down in tandem with currency 1. In such a case then, the implied
volatility of currency 2 also rises. However, to really model what is typical market behaviour, we can also assume
that in the case of a drop of the two currencies, we will have contagion and, as traders use to say, “All correlations
go to 100%.” We don’t need to use drastic back-of-the-envelope estimates, but it is reasonable to assume that the
correlation should change. Here is where further refinements to the Black–Scholes model come into place, but as a
result there are now many different ways to model such change in volatility.
To give you another hint at the problems that arise when modelling a basket of currencies, let’s look at a basket of
options. These are, simply, options on the US dollar16 value of a basket of holdings of various foreign currencies.
Therefore, instead of modelling the behaviour of each currency, we just look at one asset, the dollar value of the
basket itself, and compute the value of an option on that asset. Imagine that the value of the basket falls. This can be
because the real fell, and in such a case you would use a basket volatility consistent with a fall of the real. However,
the same basket level could be the result of a fall in the peso instead of the real, and in this case we should then
reflect that the real volatility is the same but the peso (implied) volatility is now higher, in agreement with the
market-implied volatility density distribution for the peso. This will give a different implied volatility for the basket,
but which one to choose? You can see how problems multiply quickly. This concludes our little teaser about the joys
of modelling the volatility of a basket, and we can now proceed to our second question.
The second question is: “Suppose there is a fixed dollar amount that can be spent to purchase vanilla ATM option
hedges. What is the best hedging choice?”
Consider that for ATMF vanilla options the premium spent is roughly equal to:

where T is the tenor in years, σi is the implied ATMF volatility of component i and wi is the notional being hedged
for component i. We can therefore express σN (the volatility of the entire basket) in terms of the σi for i = 1, …, N –
1 using the above equation if we know what the aggregate premium is.17 Once we have decided how much we need
to spend, we need to figure out the combination of options that: (i) costs exactly that sum; and (ii) minimises the
VaR of the resulting portfolio of exposures and vanilla option hedges.
 This is not exceedingly difficult, but it does require a bit of algebra. It will be the same thing as minimising the
variance of the portfolio (see our previous comments about differences between variance and VaR; here we are
assuming a Black–Scholes world), which is expressed in matrix form by:

where wi are the notionals of the exposures, αi are the hedge notionals, σi are the volatilities of the single currencies
and ρij are the correlations between currency i and currency j. The exponent “T” denotes the transposition of a
matrix, just a technical note of no importance to our discussion. We need to solve for each i = 1, …, N – 1 and j = 1,
…, N – 1 and αN a function of α1, …, αN–1:

This equation says that if you change any of the hedging ratios by just an infinitesimal amount the variance does not
change, which is the definition of optimal point (or local minimum).18 This is a system of N – 1 equations and N – 1
unknowns. αN is not an unknown but a function of the other αi. Once this system is solved, we have the notionals αi
for i = 1, …, N – 1 that we need to use in our vanilla options, and we can get αN from the other ones thanks to the
condition that, the aggregate premium of all these vanilla options, equals the premium budget that we have
specified. Note that the system of equations is linear, so it is very easy to solve.
In our example, assuming we have a premium budget of US$600,000, it would be possible to reduce the VaR to
under US$1 million with the vanilla allocation shown in Table 12.8.

PANEL 12.1 ALGEBRAIC DETAILS

Here are all the details needed to work out the solution. Let:

C be the premium budget;

σi be the volatilities of the portfolio components;
wi be the notionals of each portfolio exposure;
αi be the notional that is hedged with an ATMF vanilla option, for each component i; and
ρij be the correlation between the i-th and the j-th component of the portfolio.

If we express αN as a function of the other αi’s, we have:

where C = cost of hedging, which is decided in advance.

Then, we can express the variance of the hedged basket of exposures as a function of the other αi’s for i =
1, …, N–1:
 We want to differentiate the variance with respect to each of the αi’s for i = 1, …,N – 1 and impose that such
derivative be zero. We can express the derivative in this way:

All the derivatives equalling zero gives rise to the linear system of N–1 equations given by coefficients:

And the constants in this system are given by:

Once this system is solved, the αi’s for i=1,…,N–1 are determined and with them αN.

The third question is: “Suppose the hedge had to be with forward contracts. Which hedge combination would give
you the highest ratio of VaR reduction to cost of the forward points, cost of carry?” The answer will depend on the
carry of each currency, of course.
In Figure 12.34 each dot represents a hedging strategy. All the strategies are based on forwards, and the only
difference is the hedging ratio for each currency. In case where there is a set spending budget for forward points,
what is the best hedging strategy that gives you the greatest reduction in VaR?
Suppose that we have a hedging budget of US$500,000. Then the best hedging combination is shown by Table
12.9.
 Such a hedging combination would reduce the VaR of the resulting exposure plus hedge portfolio to about
US$1.2 million. No other hedging combination with a cost of US$500,000 can reduce the VaR by the same amount
or more.
The fourth question is: “Suppose, due to liquidity concerns or other reasons, I can only hedge a small US dollar-
equivalent notional of my exposure. Which currency exposure should I use that notional for?”
Let’s rephrase the question. If one can only hedge US$1.00 of the exposure, then which currency should be
hedged? The answer is the currency with the highest iVaR, as that would decrease your portfolio VaR the most. This
is the correct answer only if there can be just one hedge of one currency. In other words, the dollar hedge could not
be split. Some hedgers would start hedging a small amount and then re-check which is the greatest iVaR for the new
portfolio. Every time a small hedge is traded the portfolio composition changes, and the calculation of all the iVaRs
needs to be repeated.

PANEL 12.2 INSTRUCTIONS TO CALCULATE THE ANSWER TO THE FOURTH

QUESTION
Let:

C be the cost of hedging that is budgeted (remember the cost is that of forward points);
σi be the volatilities of the portfolio components;
γi be the interest rate differential between currency i and the US$; if the rate in currency i is higher than in
US$, then γi will be positive, meaning it is a cost;
wi be the notionals of each portfolio exposure;
αi be the notional that is hedged with a forward, for each component I; and
ρij be the correlation between the i-th and the j-th component of the portfolio.

If we express αN as a function of the other αis, we have:

where C = cost of hedging, which is decided in advance.

Then, we can express the variance of the hedged basket of exposures as a function of the other αi’s for i =
1, …, N – 1:

We want to differentiate the variance with respect to each of the αis for i = 1, …, N – 1 and impose that such
derivative be zero. We can express the derivative in this way:

All the derivatives equalling zero gives rise to the linear system of N – 1 equations given by coefficients:
 And the constants in this system are given by:

Once this system is solved, the αis for i=1,…,N–1 are determined and with them αN.

Back to our exposure – Table 12.10 shows a portfolio of forecast revenues. Suppose a treasurer decides to hedge
US$10,000 equivalent of the euros exposure, and sees that the VaR of the resulting portfolio decreases. Emboldened
by this discovery, they discuss this with their boss and decide to hedge the entirety of the euros exposure. To their
great horror, they discover that this time the VaR of the resulting portfolio has actually increased. However, now it
is too late to remedy the situation as the trade has been executed and a hedging relationship has been documented for
hedge accounting purposes. How is this possible? This, and more paradoxes, will be explained in the following
pages.

To familiarise ourselves with this newly discovered abomination, let us take a look at how to visualise VaR and
baskets of exposures. We will represent a portfolio exposure as a point on the plane (see Figure 12.35). The origin of
the plane will denote the empty portfolio (no exposures).
The length of the segment joining the portfolio to the origin will represent the VaR of the portfolio. In case a
portfolio is just one exposure, this is computed approximately as 1.645 × σ × w, where sigma is the volatility of the
exposure currency and w is the US-dollar-equivalent notional of the exposure. The volatility is expressed in
percentage terms and adjusted for the tenor of the exposure. In other words, the annualised volatility is multiplied by
the square root of the tenor of the exposure, expressed in number of years. Suppose now that a multiple-component
portfolio contains UK sterling. We can draw the whole portfolio and the sterling component on the same sheet of
paper as shown in Figure 12.36.
 The black ring on the drawing directly below the grey GBP component represents the vertical projection of the
sterling exposure over the line that joins the portfolio and the origin. That projection lies on the opposite side of the
origin with respect to the portfolio, which tells us that the correlation between sterling and the whole portfolio is
negative. Furthermore, such correlation is given to us by the cosine of the angle shown with a black circle segment
in the drawing. Those of us with a strong memory of their high-school math lessons will remember that this cosine is
negative because the angle is greater than ninety degrees.
This tells us that having the sterling exposure in the portfolio is beneficial. In other words, the sterling component
arrow is pushing in a direction contrary to the arrow of the whole portfolio, thus reducing the portfolio VaR. In fact,
if we remove the sterling component (we reverse the pound arrow and add it to the portfolio), we obtain a new
portfolio that is further away from the origin, which means it has greater VaR (see Figure 12.37).
The notion of iVaR tells us by how much the portfolio VaR decreases if we decrease the sterling exposure by a
small amount, say US$1.00. In this case, we know the answer: the total VaR would actually increase. Therefore, the
iVaR is negative. This tells us that the sterling exposure is beneficial to the portfolio as it helps reduce the risk.

The iVaR is the derivative of VaR with respect to the notional of the sterling exposure, which we indicated by wi.
We have that:

Here, the Greek letter rho (ρ) indicates the correlation between exposure i and the whole portfolio. In other words,
the iVaR equals the correlation between the exposure and the portfolio, multiplied by the volatility of the exposure.
Another beautiful thing about iVaR is that the sum of all the iVaRs weighted by the notional of each exposure
equals the VaR of the portfolio. In other words:

This is very useful as it allows us to show the VaR of the portfolio as the (weighted) sum of components. If we look
back at our previous example, we have that shown in Table 12.11.
We can therefore effectively verify that the sum of the iVaRs weighted by their respective exposures equals the
portfolio VaR.
We also have another definition to introduce, namely marginal VaR (MVaR), which measures the change in
portfolio VaR if an exposure is removed completely. You might be sceptical and think there is no need for an extra
definition when we already have iVaR, but there are some examples where iVaR and MVaR do not send the same
signal. In particular cases, you might think that it is a great idea to completely hedge an exposure since the IVaR is
positive, and then find out that you actually increased the VaR of your portfolio.
 Here is an example where iVaR>0. Instinctively, you think it is a good idea to completely hedge out the exposure,
but after you do so the new portfolio is further away from the origin, which means the VaR has increased (see
Figure 12.38).
There is also a paradox here: it is impossible to have negative iVaR and negative MVaR. This is explained in
detail by Figure 12.39. This drawing expresses graphically the fact that if correl(£, portfolio)<0 then
volatility(Portfolio – £) > Volatility(portfolio). The converse is not true (see Figure 12.39).
 To come back to the case where iVaR is positive but unluckily MVaR is positive too, the FX manager is in a
quandary – it seems that it would be a good idea to hedge a small amount of sterling exposure (iVaR is positive), but
hedging all of it increases the aggregate VaR of the portfolio (MVaR is positive). What to do? It turns out that in
these cases there is an optimal amount of hedged sterling exposure that would minimise the VaR of the portfolio.
In the example presented in Figure 12.40 we have added the sterling hedge little by little to the original portfolio.
Note in this drawing the horizontal exposure (or hedge, still horizontal) is sterling and the portfolio is the grey
diagonal exposure. We can see that initially the addition of a little sterling hedge reduces the VaR of the portfolio up
until the black circle is reached, right above vertically from the origin. Past that point, any further addition of sterling
hedge will cause the portfolio VaR to increase.
Putting it all together
A simple approach
Let’s come up for air, and get back to the simplest way we can apply the above foundation. Assume the FX manager
will have forecast revenue exposures with a value date of one year, as shown by Table 12.12. Looks familiar?
The aggregate size of the exposure is US$30 million and the aggregate VaR of the portfolio of exposures is
US$1,341,986. Below are a few possible hedging decisions that the FX manager might take.
1. Hedge only G3 currencies as “those are the most important ones” and “because this is the way we do it here”. In
that case, the euros exposure would go to zero, the aggregate exposure notional would decrease to US$25 million
but the aggregate VaR of the portfolio would actually rise to US$1,359,282. This is because the euro is
negatively correlated with Brazilian real and Mexican peso. The euros exposure provides a beneficial
differentiation to the portfolio. Once the euros are hedged, this differentiation is gone.

Table 12.13 shows the correlation matrix for this portfolio of exposures.

2. Another approach might consist of looking at the currencies as having equal importance, and picking the biggest
exposure in terms of notional. Then the Australian dollar would be hedged, reducing the aggregate exposure to
US$20 million. In this case, the VaR of the portfolio goes to US$1,454,865, which is even worse than the first
choice. What a disaster!
3. A third approach recognises that emerging market currencies are more volatile, more idiosyncratic and therefore
contribute more risk. The Mexican peso exposure should be hedged as it has a notional bigger than the Brazilian
real exposure. In this scenario, the aggregate exposure would drop to US$22 million, and the aggregate VaR
would drop to US$1,078,932.
4. Finally, the Brazilian real exposure alone could be hedged, recognising that it has the highest individual VaR of
all. In such a case, the aggregate VaR would drop to US$883,296, which is the lowest that can be done by only
hedging one of the four components. Table 12.14 summarises this.

Note that the sum of the individual VaRs is US$4,363,688, but the VaR of the portfolio is just US$2,207,371, so
there is a 49% diversification effect by aggregating all the exposures and embracing a portfolio approach. In general,
FX managers should always look at FX exposures as aggregates. In particular, the best measure of FX risk is the
VaR of the aggregate portfolio of the exposures and the hedges together.
Hedge optimisation of multiple exposures, and complex correlation relationships
Now let’s get closer to a real situation, getting a bit more sophisticated and a bit more complicated. Corporations
around the world have multiple exposures in terms of revenues and expenses in different currencies. Consider the
example in Table 12.15 as we introduce opposing signs in the exposures, where a positive exposure corresponds to
revenues and a negative one to expenses, and all the exposures have a time horizon of one year.

The VaR at a 99% confidence interval is roughly US$11.25 million – ie, the probability that the aggregate value of
these exposures will lose more than US$11.25 million over the next year is 1%. Purchasing an ATM vanilla option
for each exposure would cost US$5.8 million and reduce the VaR to US$5.8 million. Wait, all the exposures are
hedged, shouldn’t VaR be zero? We’ve spent US$5.8m to hedge, and that “loss” has a probability of 100% (future-
value).
If the treasurer decided to allocate a budget of US$5 million to the purchase of ATM vanilla options, the company
could not purchase all the options needed to hedge each exposure. How do we figure out the best way to allocate the
premium we can spend? One approach could be to hedge only the four largest exposures – Australian dollar, UK
pound, euros and New Zealand dollar – for a combined premium of close to US$5 million. By doing so, the VaR on
the portfolio drops to US$7.4 million. This approach can be improved, since it does not take into account that
different exposures have different volatilities, and it also completely disregards the correlations among those
exposures. We saw in the example above how negatively correlated exposures can be important, but often it is the
positive correlations that we have to be careful about. A typical example would be a corporation with the same
amount of revenues from euros, as expenses from Danish krone. Euros and krone have a very high correlation, the
VaR of the combined portfolio is close to zero, but hedging only the euros exposure would eliminate this portfolio
effect and actually increase the VaR, thus representing a complete waste of hedging budget. Table 12.16 is the
correlation table of the exposures.
We can use an analytical tool, available on the Bloomberg Excel library, to calculate the optimal hedging portfolio
that would result in the least possible VaR given a hedging budget of US$5 million. Such a portfolio corresponds to
the hedges in Table 12.18.
The sum of the vanilla premia adds up to US$5 million. With this optimal hedging portfolio, the VaR drops to
US$5.81 million, which very interestingly matches the VaR resulting from buying a full set of vanilla options!
Recall that buying all the options would have resulted in a hedging ratio of 100% at a cost of US$5.8 million (see
Table 12.18). In other words, blindly hedging everything would achieve the same VaR as the optimal hedging
portfolio, but would require an additional US$0.8 million! This is an interesting scenario because they both have the
same VaR. In one case, fully hedged with vanillas, and in the other situation only hedged 50% of the Kiwi exposure,
saving US$800,000.
Hedge optimisation is becoming increasingly widespread in the corporate treasury space as the savings are so
tangible and visible, and the analytics software have become cheaper and easier to use. I recall, not that long ago a
Fortune 15 company wanted to run this same analysis and relied on banks to calculate the VaR numbers. However,
there was no technology widely available to optimise.
They would call a bank and ask: “Can you calculate my VaR if I hedge 40% of BRL and 25% of TRY? Sounds like
a good idea.” The bank would take a few days to come back with a VaR number, after which, the treasury would
formulate another try, once again based on hunch, and ask again. With today’s instruments one can run the numbers
and find the optimal solution without random tries in about 15 seconds.

Hedging optimisation of the forward points

Finally, let’s take a different approach. Let us assume that we will not be able to hedge with options, and instead use
forward contracts. In this exercise, we will be considering not only the correlation levels and the volatility of each
currency pair, but also include the “cost” of the forward points. Each hedge might have a cost, most of the time
quantified as the interest rate differential versus the functional currency of the hedger. To perform this analysis,
corporates often rely on banks’ quant teams to run the sophisticated models necessary to exactly determine the
optimal hedging mix that reduces the risk by the largest possible amount.
Let us go over a full example to understand the details. To make the example clearer and more poignant, we chose
a historical period when interest rate differentials were high. After the recovery from the dotcom boom and bust of
2000, in 2007–08 the global economy was expanding quickly. Investment in emerging markets was thriving and
interest rate levels were high. In fact, the Fed had been on a hiking path and the ECB famously hiked in July 2008 in
what is largely viewed today as its most egregiously wrong rate call. Commodity prices were at historical highs. US
corporates were very keen on expanding their revenue source abroad, with the best opportunities being seen in
emerging markets. All those newly generated revenues were exposed to the value of those local currencies, which
for the previous five years had been for the most part appreciating. Some corporations did not bother hedging, we
assume reasoning that appreciating currencies would continue to appreciate forever. Many corporates, however,
heeded general risk-management principles and maintained their hedging programmes. The problem they faced was
that, with a rise in interest rates, the cost of hedging, as expressed by the difference between today’s dollar value of
future revenues and the future dollar value of future revenues, was also rapidly increasing.
Imagine a corporation that forecasts the exposures in Table 12.19 as of July 31, 2008. Those revenues are
supposed to materialise a year in the future. Corporate treasury has no idea that the following week the greatest
financial crisis in the last 80 years would be unleashed. We can see all those forecast revenues in Table 12.19. The
table shows a basket of forecast revenue exposures, where the amounts are given in local currency and all
correspond to US$100 million at spot levels as per July 31, 2008.
Table 12.20 lists the parameters of our example for the calculation of the hedging costs, and the volatilities and
correlations. For the calculation (and the minimisation) of VaR, a confidence interval of 99% is used. Volatilities
and correlations are realised or historical, as per the parameters.
 If we look at Table 12.21, we can extract some further information about the exposures. We have assumed the
same US$100 million equivalent value for each of the six exposures. All exposures are revenues coming from
emerging market economies that were rapidly expanding and whose currencies were relatively liquid. As was typical
of those years, the interest rates in all of those economies were higher than in the US. This meant that the forward
rates were higher than the spot rates, which in turn meant that the US dollar value of those exposures, when
measured using the forward rate, was lower than the US dollar value of those exposures calculated using the spot
rate.
Treasury would reason:
If I do not hedge and the spot rate does not move, then at the time I realise those revenues and translate them into dollars, I will receive the
US dollar value calculated using current spot rates. If, on the other hand, I lock in the exchange rate using a forward hedge, then I will
eventually receive the US dollar value calculated using current forward rates, which is a lesser value (under my assumption that spot levels
will not move or even that the local currency will appreciate).

The difference between the two dollar values is considered to be the cost of hedging. Take, for example, the
Brazilian real exposure, which is the difference between US$100 million and US$91 million (first row of Table
12.21). This was a cost of about 9% at that time.
How does corporate treasury look at this cost? The cost is very high, no matter what. If the gross margin of the
business is lower than 9%, then hedging the FX risk will have the secondary result of wiping out the profits of the
business, so the firm might as well forget expanding in Brazil. On the other hand, those exchange rates are all very
volatile, so foregoing hedging is also quite an irresponsible choice. With hindsight, foregoing hedging in July 2008
would have had disastrous consequences as the onset of the great financial crisis caused all those six currencies to
dramatically weaken (eg, the Brazilian real weakened by about 16%). Again with hindsight, this would have made a
hedging decision look like an obvious choice, despite the high cost.
Figure 12.41 shows the losses of the six currencies in the basket in the year following July 31, 2008. Suppose that
the hedger decides to ignore costs and wants instead to just eliminate all FX risk. Therefore, they choose to hedge
each and every one of the six exposures with one-year forward contracts. In such a scenario, the cost of the hedges
would be about US$43.43 million but the VaR of the exposure would become zero. This situation is represented in
Figure 12.42 by the leftmost point, which as has a VaR of zero and hedging cost of US$43 million. In this case, the
hedging ratio of each of the exposures is 100%.

On the other hand, if the hedger completely neglects hedging, then the cost of hedging will be zero, but the VaR
will be about US$150.5 million, as represented by the top rightmost point in Figure 12.42. This is the case where the
hedging ratio of each exposure in the basket is zero. In between those two extremes many different hedging ratio
combinations are possible. Each hedging combination corresponds to the choice of six hedging ratios, one for each
of the six exposures. Each hedging combination will have a hedging cost and a VaR associated with it, and can
therefore be represented as a point in the chart.
Ideally, the hedger wants to reduce the cost of hedging and the VaR of the resulting (post-hedging) exposure at
the same time. Therefore, the hedger wants to be as much to the left and as high as possible in Figure 12.42. This
can be done, to an extent. It turns out that the hedger will not be able to go to the left of or above the black line in the
chart. The black line is the efficient frontier. Any hedging combination that lies either to the right of or below the
black line (or both) is achievable but suboptimal because the hedger can find hedging combinations lying on the
black line that have either same cost but lower VaR or same VaR but lower cost. There are no better or worse
hedging strategies when comparing different hedging combinations that sit on the efficient frontier. It will depend on
the hedger’s preference about spending less or reducing volatility.

Figure 12.42 shows every possible hedging strategy (a choice of six hedging ratios) as a dot determined by the
VaR (or CFaR, horizontal axis) and the cost of hedging (on the vertical axis). Note that it is possible for two distinct
hedging strategies to have the same VaR and the same cost of hedging, in which case they would be represented by
the same dot on the chart and the hedger would then be theoretically indifferent between the two. No dot can be
above or to the left of the black line, which is the efficient frontier.19
 Table 12.22 shows a matrix with the pairwise correlations between any two of the currencies in the basket,
calculated using historical fixings. The HVol column to the left shows the implied one-year volatilities of each
currency.
Figure 12.43 shows the risk decomposition assuming no hedges. Note that in our example we are looking at
revenue hedges, so all the exposures are in the same direction. This explains why the risk diversification is relatively
minor, standing at below US$2 million. As there are no hedges, the unhedged and the hedged CFaR columns show
the same number.
Figure 12.44 shows the same risk decomposition but this time assuming a 100% hedging ratio, which of course
results in zero net final risk; the rightmost column shows zero risk.
Now let us come to the optimisation process. The hedger can have a cost level in mind, say of US$20 million.
This is a far cry from the US$43 million that a full hedge would cost. It is actually 46% of it. The hedger could
decide to set all the hedging ratios to 46%, in which case they would achieve a hedging cost of US$20 million.
However, nobody said that each one of the six hedging ratios has to be the same. By varying the hedging ratio
combinations, the hedger can still maintain a cost of US$20 million but achieve a lower VaR. A computer model can
easily run an optimisation process, and output the optimal set of the six hedge ratios that would generate the lowest
possible VaR level under the assumption of spending US$20 million in hedging costs. This is the essence of the
optimisation process, which corresponds to a mathematically efficient deployment of hedging expenses. It also
corresponds to the much more logical task of protecting the aggregate dollar value of the six revenue streams instead
of handling each one of the six risks separately.

In a similar and equivalent approach, the hedger could decide on a specific VaR level that they are willing to
tolerate, and then optimise the process to find the hedging combination that results in the lowest possible hedging
cost under the assumption that the VaR matches the predefined level. The output of the computer optimisation
process is described in Table 12.23.
Table 12.23 shows the final result of the optimisation process. The optimal hedging combination for a hedging
cost of US$20 million is to not hedge Brazilian real at all, to hedge 16.67% of the Mexican exposure, 13.57% of the
Turkish exposure, 15.75% of the Russian exposure, to not hedge the South African exposure and to hedge 16.67%
of the Indonesian exposure. This generates hedging costs of US$20 million and a net residual VaR of US$66.327
million.
Options and forwards cannot be optimised at the same time
Right about now you are surely expecting us to bring in the options and find the ultimate solution. Alas, we first
need to describe a very important consideration about hedging with forwards and hedging with options. One might
be tempted, and in fact almost all the clients I speak to are tempted by this, to be greedy and try to optimise the
hedging strategy with a mix of forward and option contracts, combining the approaches described in the preceding
pages. This is not possible. The reason is that our options approach is mathematically sound but our approach to
hedging cost when considering forward contracts is inconsistent with financial theory. We cannot try to combine a
sound approach with an inconsistent approach.
Characterising the difference between spot and forward as hedging cost is very wrong from the point of view of
financial theory. The foreign asset is expected to depreciate by 9% in a year’s time, and therefore the 9% fewer US
dollars does not represent a loss. You cannot compare the US$100 million with the US$91 million because you
never had the US$100 million at the time of the hedging decision. It was all forecast revenues – it was not present
value, it was future value all along.
The mistake stems from the wrong assumption about the probability that the exchange rate may be equal or better
in a year’s time. According to the way the market is priced, there is a less than 50% chance of that happening. In
fact, the market is expecting the rate to be worse in a year’s time, so contemplating the spot value of the forecast
exposure is just wishful thinking. Despite this logic, corporate treasury professionals often consider that the forward
is historically a very poor indicator of future market value, and in some cases it can be argued that the current spot
rate is a better indicator – hence, the discrepancy between orthodox financial theory and corporate practice. Since
this is how many corporates think, we added the previous section to show ways to minimise the cost of the forward
points.
In summary, for options we consider the premium as the cost of hedging, but for forwards we consider as the cost
of hedging something that is not a cost from the point of view of standard financial theory – namely, the difference
between current market level and future expected market level. A very simple example will shatter any hope of
combining the option approach with the forward approach.
Imagine you hedge with a risk reversal. Remember those from earlier in the book? No? Don’t worry, we will talk
about them again in the next chapter. A risk reversal is the purchase of a put option and the sale of a call option.
They have different strikes (the put strike being the lower) but their value is the same, so when you execute a risk
reversal you do not have to pay any premium. You will therefore agree that, in a hypothetical combined theory of
option and forward hedge optimisation, the risk reversal would count as being zero cost. Now raise the put strike. To
keep the premia equal, then the strike on the call has to fall (for calls, lower strike = higher price). Continue doing
that and you will reach a point where the put strike and the call strike will touch. When that happens the strike will
necessarily be equal to the forward rate, and this comes from option theory (return to our initial chapters to check
that).
Now that the two strikes of the risk reversal are the same, our risk reversal has become the same as a forward
contract. Think about it for two minutes to be convinced of this basic identity of options theory – and you don’t even
need to know how to price options to figure it out. The risk reversal will still be zero premium and zero hedging cost
in our combined option and forward hedge optimisation theory, but now that it corresponds to a forward this means
that the forward contract is also zero cost. Hold on a minute, we have just looked at a USDBRL one-year forward
hedge that cost 9%! How can that be? The problem is that we have a notion of hedging cost for options (the
premium) and a notion of hedging cost for forwards (the forward points) and they are incompatible. The only
solution is to throw out the hope of being able to combine the two theories.

Basket options: The benefits of correlation

This section should really be part of the next chapter, where we analyse specific hedging instruments. However,
there we will restrict ourselves to the more conventional instruments and their direct offspring. Basket options feel
more natural to this discussion as, by definition, they encompass a portfolio of offsetting risks. Furthermore, the
beating heart at the centre of basket options is correlation, which we have discussed extensively in this chapter.
With that, the concept of a basket approach is the same that informs the hedging decision of option users when
instead of purchasing six independent vanilla option hedges, one per each currency exposure, they decide to
purchase one single-basket option, which is an option on the aggregate dollar value of the basket of the six revenue
streams.20 Basket options have been used extensively by many of the larger corporations for decades. They are
difficult to price and illiquid to trade because the correlations assumptions can be tricky to pin down. I recall pricing
baskets that significantly reduced the total cost of hedging, but in the end the trading desk that would have to
manage the risk threw out the mathematics and discounted the total premium based on what felt right. With the
tremendous increase of computer processing power, it has become easier for anybody to compute the cost of these
basket options, so the banks have lost their information edge in that regard. For example, a corporate treasurer as
well as a sophisticated bank structurer can now price a basket option on many option pricing systems. Figure 12.45
is a screen shot of the Bloomberg DLIB function that can instantaneously price a basket hedge. It shows an example
of pricing for a basket option with details matching the six exposures that composed our sample basket.
Let’s briefly run though the basics and a simple example.

Basket options are options on a basket of currencies against a base currency. The buyer of the basket option has the
right to buy (call) or sell (put) a predefined amount of each currency against a fixed amount of the base currency
at expiry. Basket options are typically cash-settled in the base currency.
When you buy a basket option, the strike is the base currency notional. At expiry, the decision to exercise is like
this: I choose to exchange all these currency notionals for US$X. You cannot exercise just one currency. That is
how a basket option differs from the sum of the individual strip of options. With the basket, you can no longer
“cherry pick” those currencies that you choose to exercise. For this reason, the basket will cost less than the
corresponding strip of vanillas. How much less depends on the correlation between the currencies. The greater
the correlation between the component currencies, the more the price of the basket option will tend towards the
sum of the individual vanilla strips.
Now consider the example of a European corporate that exports approximately 50% of its goods to the US, 25% to
Japan and 25% to the UK. The corporation is looking at ways to hedge against an appreciation of the euro against
the overall exposure (US dollars, Japanese yen and sterling), and is prepared to forego some flexibility in each
individual currency pair to reduce hedging costs.
The most complete hedge would be to purchase:

three-month EUR call USD put, strike: ATMF in 50%;

three-month EUR call JPY put, strike: ATMF in 25%; and
three-month EUR call GBP put, strike: ATMF in 25%.

All those option purchases will cost 1.47% of the euro notional. The benefit is that the company is fully protected at
the forward, and if the EURUSD strike is ITM while the EURJPY and EURGBP are OTM the company gets to
exercise the one option and convert the rest back at spot. This will surely result in more money than was guaranteed
by the basket option. Again, we ask ourselves that nagging question: “Is the company in business to maximise its FX
profits, in this case by paying more upfront, or should the company be content to be able to say to the analysts, ‘We
will have this minimum amount in euro revenue guaranteed’?”
Alternatively, the company can purchase a three-month euro call/basket put, whereby the basket is composed
approximately of 50% of US dollars, 25% of Japanese yen and 25% of UK sterling. This costs 1.25% euro and
represents a discount of 15% relative to the strip of options.

At expiration, the customer has the right to receive €9,993,369, or let it expire and sell all the currencies in the spot
market. They will only exercise this right if the value of the basket in euro terms, at the spot market, is less than
€9,993,369. Now you see why the base currency notional is the strike on the option?
Assume the final spot rates for each of the currency pairs in Table 12.25. Then the company would exercise its
basket. The basket option would be cash-settled, and the bank would convert all the currencies at the spot rate and
pay the company an additional €77,984.87 to make it whole.

Note that the EURJPY and EURGBP have depreciated and the euro value of the Japanese yen component of the
basket increased, but the overall value of the basket in euro terms has decreased. If the customer simply bought the
individual strips, they would have exercised the EURUSD and EURGBP options and allowed the EURJPY option to
expire worthless, capitalising on the favourable move to buy euros at the spot rate. With the basket option, the
customer must accept buying euros against selling all of the other currencies.
As you are probably tired of our saying, there are many variations to this basic example. Some of the most fun
are:

quanto baskets, where any cash gain on the basket is payed out in an alternative currency to the base currency at a
predetermined quanto factor;
average-rate basket, where rather than use final spot rates, the value of the basket in base currency terms is
calculated using the arithmetic average of a series of spot fixings for each of the component currencies; the
averaging effect will make the average rate basket cheaper than the regular basket option; and
digital basket option, which will pay out a predetermined cash amount if the value of a given basket (in base
currency terms) is either above (call) or below (put) a predetermined level at expiry.

The concept of using basket options is very simple, but many people do not understand it. Let’s take a different
approach to see why a basket is more efficient. Suppose a company buys six different vanilla options to hedge a set
of exposures. The cost is the sum of the six different premia. If it’s a dollar-based company, it is interested in
preserving (hedging) the aggregate US dollar value of the future revenue stream. Now imagine that one currency
falls and another rises, and net the US dollar value of the whole set of exposures (the basket) remains the same.
Something very interesting will happen. For the currency that has risen, the bought option, being a local currency
put, is OTM, so it will not pay at expiry. On the other hand, for the currency that fell the option will be ITM and it
will pay at maturity. Therefore, the company is left in a very interesting predicament: the value of the aggregate (the
basket) did not change but the hedge purchased paid out above and beyond.
The company did not need this money as there was nothing to protect, the basket did not lose value, and yet the
hedge paid. The reason is that it is not the perfect hedge. The fact that it paid when there was no need tells us that we
are over-hedging. We are paying more than we should for protection that is more than we need. There is nothing
wrong with that, but we should be cognisant that our employer might not be delighted to hear that we throw money
around for unneeded hedges. One treasury FX manager at a very large US software corporation told me that he used
to do it to get a little money at the end. I can second that point of view: I also like a little lobster and caviar for lunch
every day, but my employer would not be happy to find out that they are paying for it!
The solution to wasteful hedging is to find out what the perfect hedge is. If we want to protect the aggregate US
dollar value of our basket, then the underlying of our option should be the aggregate US dollar value of our basket.
This means that the hedging instrument should be a basket option. A basket option will be cheaper and won’t
wastefully pay when there is no need. A basket option is cheaper for that reason, and if you want to look at it another
way, the volatility of the basket as an underlying will be lower than the average volatility of the six currencies,
because oscillations in the value of the components will partially offset when summed up in the basket. This lower
volatility will generate a lower premium when used in a Black–Scholes formula to calculate the option premium.
Strictly speaking, you are approximating if you use Black–Scholes for a basket option because if all components are
lognormal then the basket is not lognormal. You can also call it diversification effect if you wish. The fact remains
that with the aggregate basket being less volatile, the price of the option will be lower and the hedger will save
money.
And yet, there is one very important reason why hedgers, specifically corporate hedgers, might have mixed
feelings regarding basket options: accounting! As the six different risks composing the basket are not of the same
type or nature, the accounting standards do not allow the application of hedge accounting to the hedging of the US
dollar value of a basket. This greatly reduces the appeal of a basket option to corporate treasuries. Nonetheless, the
urban legend says that in the late 1990s a very large US entertainment corporation found a very clever workaround,
which was in 2005 reported to me by the head of equity structuring at a bulge bracket bank in New York. They
would not buy a basket option, but would instead compute all the six deltas that the option would have with respect
to each and every one of the components of the basket. Then they would enter into forward trades for a notional
equal to each specific delta, thus effectively hedging each one of the six exposures, but with hedging ratios
calculated as the deltas of the phantom basket option that they never bought. They would keep calculating these
deltas periodically, adjusting the six forward hedges each time until the final maturity.
We have already spoken about this technique, called “dynamic delta replication”. In essence, the technique
replicates the result of a non-existing option. Theoretically, at maturity the aggregate payout of all the forward trades
executed during the life of the process would result in a P&L that would equal the payout of the basket option minus
the theoretical premium of the basket option which they never purchased (this is standard Black–Scholes option
theory). Of course, the assumption is also that the realised volatility of those six currency pairs during the life of the
process should be as close as possible to the volatilities used each time the phantom option deltas are calculated.

1 S. Gad, 2016, “Why Warren Buffett Doesn’t Care About Currency Movements”, TheStreet, August 3 (available at
https://realmoney.thestreet.com/articles/08/03/2016/why-warren-buffett-doesnt-care-about-currency-movements).
2 We have spoken with investors who need to hedge the EURUSD and would simply trade an ETF linked to EURUSD, and happily pay the associated
fees and tracking error. Retail investors will definitely do that, as it is easier.
3 By “sold probably for free” we mean that when the contract was negotiated they did not request a cash payment for the fact that they effectively sold a
vanilla FX option to the counterparty, and therefore they should receive cash upfront or the future value of the premium in the price of the widgets.
4 See https://www.statista.com/topics/1715/us-export/
5 S. Lipin and B. Bahree, 2000, “Air Products Take a Beating After Failed Deal for BOC”, Wall Street Journal, May 11 (available at
https://www.wsj.com/articles/SB957980505943262758).
6 Those internal brainstorming meetings have been some of the most intellectually thrilling moments of our careers. A team is put together to decide the
fate of millions and millions of dollars by gingerly discussing what terrible pitfalls the market can have in store for the client, and how to thwart them
all with the perfect hedge.
7 “Real money” is a term used in the FX markets to describe pension funds, asset managers and insurance companies. In our use here, I include
corporations as well. Typically, they are considered market participants with a longer-term view. Hedge funds and other speculators tend to focus on
the short term.
8 Functional currency is the currency in which most transactions are conducted. This determination can get more complex for companies dealing in
multiple currencies under different jurisdictions.
9 The US dollar exposure does not impact on earnings as the parent is US dollar functional. So for unhedged only the €100m matters. If fully hedged,
then the US$80m actually becomes €80m and now both the €100m and the €80m sum up into the €180m earnings exposure.
10 There will be some portfolio benefit for the aggregate of those two risks as there is some correlation between the two currencies.
11 This is calculated as: original revenue £100 minus original expenses £80 minus hedge £70 equals £50, and then there is the second leg of the hedge:
USD70.
12 It would not change any VaR relative to a confidence interval of x%, where x% is greater than 4%, but it would change any VaR calculation relative to a
confidence interval of y%, where y% is smaller than 4%.
13 C. Acerbi, C. Nordio and C. Sirtori, 2008, “Expected Shortfall as a Tool for Financial Risk Management”, Abaxbank, February 2001.
14 iVaR is incremental uncertainty added to or subtracted from a portfolio by slight changes in the purchase or sale of an instrument in the portfolio. It can
also be computed as the derivative of the portfolio VaR with respect to the notional of the portfolio component in question.
15 We do not want to bog down our reasoning too much, but imagine a distribution with 100 outcomes. Five of them are a loss of 10, 90 of the outcomes
have a loss of zero, and five of them have a gain of 10. The 5% confidence interval VaR equals a loss of 10; the CVaR is also a loss of 10. Now
consider a second distribution, equal to the previous one except that one of the five worst outcomes is changed from a loss of 10 to a loss of 1,000.
Now the VaR is the same as before, it has not changed. However, the CVaR is now equal to a loss of (1,000+10+10+10+10)/5=208. The CVaR
reflects the difference between the two distributions, but the VaR does not. This is the reason many think that CVaR is a superior measure of risk.
Nonetheless, VaR is currently much more popular.
16 Of course, one can replace “dollar” with “euro” or any other currency here.
17 The above equation is just a very handy back-of-the-envelope approximation formula that only works for ATM options.
18 Well, no, this is not a sufficient condition for finding a local minimum, but our equations are just of the second order so in this case it is sufficient. For
third order and higher it would not be sufficient.
19 The tables and chart we used here were created by our colleague Matt Gorelick and can be found in the Bloomberg XLTP database.
20 We briefly mentioned basket options earlier.
 13
You Have Options

The previous chapter examined some of the theoretical and practical considerations to hedging. Regardless of the
approach you choose, there comes a time to execute. This chapter will examine the decisions and instruments
available when the rubber meets the road. Some instruments should be familiar from previous chapters – we will
review those quickly, but focus more on new elements that would apply to corporate behaviour. We will take it up a
notch and talk about nuanced considerations when hedging forward, discuss the difference between forwards and
cross-currency swaps, and move into greater complexity when looking at forward-like strategies. Specifically, we
will consider no hedging, hedging organically, hedging with forwards, options and then all kinds of forward-like
alternatives that stretch the use of options.

HEDGING STRUCTURES FOR CORPORATE HEDGERS

Forwards
The workhorse for FX hedging by most companies is the forward contract, just as the cross-currency swap is for
asset managers. Both instruments do the same thing, more or less. We have seen that a steady stream of revenues
from abroad is often disrupted and complicated by the vagaries of the exchange rate. What may be a dependable
local business or a growing business might appear in dire straights when translated into the currency of the parent
entity. In January 2015, during the Apple Computer earnings call, the first question that analysts posed to
management was about quantifying the adverse impact of the US dollar strength on revenues from abroad, in an
aggregate annualised percentage. The answer was a huge 4%. One can appreciate why a corporate treasury may
decide to hedge. The most naïve and simplistic form of hedging is to lock-in the exchange rate today in view of the
future translation of revenues. Entering into a forward hedge with a bank accomplishes this task beautifully, and at
zero cost, or at least nominally zero cost.
Let us critique this choice a little. The good news is that the company does not have to part with any cash at the
time of hedging. The bad news, of course, is that the forward hedge will have to be settled at some point in time and,
at least mathematically, there is a 50% probability that the settlement will be adverse in the sense that the company
will have to pay some cash to the bank to settle the forward. A lot of treasury operations do not like that. However,
the ability to lock-in the exchange rate is certainly well worth the possibility to settle the hedge at a loss, right?
It is important to understand that when the forward hedge settles at a loss, this happens because the underlying
exposure – for example, the repatriation of revenues from Europe into the US – generated more dollars than
originally estimated due to exchange rate moves. This logic often fails to convince everyone. Human nature being
what it is, the prospect of a hedge that loses money is really not liked in corporate circles. Many corporates prefer
not to hedge at all – not based on the Warren Buffett theory referred to earlier, but rather to achieve three objectives:
they can blame underperformance from abroad on exchange rate movements; they will avoid any hedging losses;
and they potentially could register gains on favourable currency movements. Of course, they will have to endure
more volatility this way.
How would a forward work? A little refresher on currency forwards:

they are the most commonly used instrument for hedging foreign currency exposure;
a forward defines a fixed price to buy (or sell) a given amount of foreign currency at a specified future date;
such a “fixed price” is “derived” from that of the underlying exchange rate (spot rate) – in this sense, it can be
defined as a derivative product; and
the relationship between the forward rate and the spot rate is based on the interest differentials of the currencies
involved (interest rate parity).
Why is the currency forward so successful? Well, it is easy to understand, it eliminates uncertainty of the future
price of foreign currency, and it complies with hedge accounting rules. As a matter of fact a forward contract offers
the highest chance of qualifying for hedge-accounting treatment. In this chapter, we will examine simple
transactions and explain how they can be hedged. This process can easily be multiplied to reflect the policy and
process considerations discussed in the previous chapter.
Take a US corporation with revenue of £10 million due in one year. Recall that when UK rates are lower than US
rates, the £10 million is worth more at the one-year forward than it is at current spot rate. This is necessary so that
one is indifferent between investing in the US or the UK for one year. Nevertheless, if the spot, and consequently the
forward rate, moves while interest rates remain approximately the same, the US dollar value of those pounds will
change. Therefore, if the pound falls the translation will result in fewer dollars, and if the pound rises the US dollar
revenue will be higher (see Figure 13.1). This is a very real concern given that just before the Brexit referendum in
June 2016, sterling was closer to £1.60/US$.
What would the position look like when hedged with a forward? To hedge with a forward, treasury staff will get
online and shop for the best bid for one-year sterling. Assuming 173 points (£0.0173/US$) is the best price,1 they hit
the bid and enter into a forward transaction that obligates them to deliver £10 million to the bank in one year, and
the bank will give them exactly US$13.673 million in exchange.

What’s the resulting position? A flat line.2 That revenue stream will definitely result in US$13.673 million in
revenue, assuming the bank is in business. This should be very familiar by now if you’ve read the first 12 chapters
of this book, so let’s focus on the forward points for a bit. Many corporates will tell you that forward hedging can be
costly depending on the magnitude of the points. In other words, they assume either that one should be able to
transact at spot or maybe that spot is the most likely outcome and therefore anything locked-in, away from spot, is a
cost. In the above example, this logic will assume that hedging forward is a profit-generating endeavour. In all
fairness, this is not a crazy way to think if your positions are all marked-to-market at spot. Therefore, if the
accounting department translated that revenue the previous month at 1.35 or US$13.5 million, when you sell the
sterling forward and generate a revenue of US$13.673, you “made” US$173,000.
In the sterling example there was a modest interest rate differential. What about emerging market countries where
the differential is huge? Consider, for example, Brazil. Suppose you want to hedge your reais revenue out five years.
Stop, don’t do it! No one knows where USDBRL will be in five years, and you don’t want to pay the points.
However, suppose you are a Brazilian company that borrowed in US dollars and in fact you have five years of
cashflows that you really want to hedge because, as mentioned previously, USDBRL rates can really move
dramatically and usually the drama is to the US dollar upside. This means that all of a sudden you need to come up
with a lot more reals to pay back the very same loan. As of the time of writing, there were more than 1,100 corporate
and government bonds outstanding with a notional of US$157 billion in this situation.
How does one hedge with forwards? Pull up the forward points as we did in Figure 13.2 and start calling your
bank to ask for levels at which you can buy US dollar forwards versus Brazilian reals. On closer examination, your
jaw drops. Spot is R$3.2587/US$, but to buy dollars five years out you need to come up with R$4.494317/US$!
Even in the one year the rate is 3.398731. That’s criminal. Where you thought you could get a decent rate issuing in
US dollar bonds off the 1.6% for US debt, when you do the calculations you will end up with are the reals
government rates of 6%. Of course, now you know that it is due to interest rate parity. You can’t get a cheaper rate
without taking some FX risk.

How does one hedge this risk, especially when experience tells you that except for the few times, when the US
dollar rose dramatically, the spot rate is closer to where USDBRL will be in five years than the implied forward (see
Figure 13.3)?
 What many Brazilian companies do to hedge is enter into a short-term forward. For example, they may hedge the
entire issue, or the portion they want to hedge based on their policy, by buying dollars forward for three months.
This means they lock in 3.289995 in our pricing page. This way if the dollar rises uncontrollably, they have
something to protect themselves. However, this is only a partial solution at best. If the US dollar rises dramatically,
chances are the forward points will also blow out. In other words, their five-year forward rate is now much worse
than before. Additionally, if the interest rate curves stay as they are, over the five years of the hedging strategy the
cost will still amount to the same (assuming a straight line forward curve), but it is the reality of things that it is
much easier to take a bit of pain over five years than a lot of pain in one shot today.
Figure 13.4 shows the USDBRL spot rate going back to 2000. However, let’s examine the more recent dollar
appreciation in 2015. In June 2015, if the company bought a US dollar three-month forward at 3.231631 it would
have paid away approximately 1035 points off a spot of 3.1281. If they had chosen to lock-in the five-year rates, it
would have been at a rate of 4.867746. By September of that year, the political turmoil in the country caused the
spot rate to rise to 3.8639. Their three-month forward had just expired and, just as they were looking to get another
short-term fix, the three-month rates were 3.987516 and the five-year rate was now a whopping 6.285655. Not only
did spot move dramatically, but the forward points exploded in the three months from 1035 to 1236, in the two years
from 7310 to 9013 and in the five years from 17396 to 24217! Clearly, they could have hedged out to two years
originally and still been better off.
 Nevertheless, companies know from empirical experience that going out beyond a certain term just doesn’t make
sense. For example, Figure 13.6 plots the five-year Brazilian NDF rate (in white) against the shifted spot rate (in
grey). We took the spot rates and moved them ahead five years. Therefore, any point on the graph will have the five-
year NDF for that day, and the spot rate is the spot rate that actually materialised five years later. There was a period
in 2006–08 (and briefly in 2009) where the forward rate correctly predicted where spot would end up, but for the
most part, up to 2009, hedging to the five-year forward was a losing proposition.
 However, that’s no way to run a business. For all our analysis going forward, we will need to assume that
everything is marked-to-market at its current market value, or replacement value. This means that forward points are
neither profit nor loss (see the previous chapter for a more impassioned argument on this). As a matter of fact, our
baseline and benchmark for analysis will always be the forward. So, if the effective rate (the transaction at
settlement) is better than the forward rate today, the company made money – and lost money if it is worse. As we
know from our previous discussion, the forward points associated with a trade have mostly to do with the interest
rate differentials. In other words, there is no cost per se – it is simply the opportunity to invest in one country versus
another. Be that as it may, reducing the forward points of any transaction is important to corporates (see the detour
discussions in Chapter 12 for more on this).

Cross-currency swaps
The twin sibling to hedging with forwards is hedging with cross-currency swaps. You might recall our previous
treatment of the cross-currency basis. We have spent some time on that topic, but we want here to give a brief
refresh of the main points.
A European company has issued debt in US dollars and wants to hedge the FX risk related to their US dollar
liability. In particular, if the EURUSD exchange rate falls by the time the debt has to be repaid, the repayment would
be much more onerous in euro terms. To hedge this risk, the company can of course enter into a series of forward
contracts, one for each coupon maturity and one for the final balloon payment. This would constitute an effective
hedge.
Alternatively, the company could enter into a cross-currency swap. Assuming the US dollar debt is fixed rate and
the company wishes to hedge it with fixed euro payments, they would then enter into a fixed-to-fixed EURUSD
cross-currency swap. At each coupon or final maturity date on their issued debt, they pay the bank a fixed amount of
euros and receive an amount of dollars that corresponds exactly to the coupon (or final balloon) payment on their
issued debt. Figure 13.7 shows the pricing of a five-year fixed-to-fixed cross-currency swap where the corporation
pays euros and receives US dollars. The final balloon payment notional is calculated using the five-year EURUSD
forward rate, not the spot rate.
In principle, the company could calculate all the forward EURUSD rates using the interest rate curves in euros
and dollars. This can be done, but the resulting forward rates would not match the actual market forward rates. How
so? Remember from our discussion on the cross-currency basis, supply-and-demand issues skew the forward rates
one way or another. To rephrase, if you just use the US dollar and the euro interest rate curves, you are looking at
domestic factors only, albeit in two different countries. This will not take into consideration any supply-and-demand
factors that play in the movement of funding between Europe and the US. For that reason, an adjustment will be
applied to the coupons in the cross-currency swap, and such adjustment, which takes into account cross-border
funding demand, is called the cross-currency basis (see Chapter 4).

Figure 13.8 shows a comparison between the traded forward points in EURUSD, in the third column from the left,
and the forward points that would be implied by the two domestic interest rate curves, the third column from left for
the euro and the fourth column for the dollar. The calculated forward points are shown in the second to last column
to the right. The last column to the right shows the difference between the calculation and the actual market quotes.
Observe the five-year tenor in the last row, where the discrepancy is 0.026426, close to 2.5%. To account for the
error, we need to take the cross-currency basis into account. This can be done by using a dedicated EURUSD-basis
interest-rate curve instead of standard domestic interest-rate curves for the euro.
Vanillas
However, there is another solution if one really wants to avoid hedging losses: buying a vanilla option. This means
paying an option premium, and we know that money is sunk and gone forever. What do you get for the money? You
get the certainty that you will not have hedging losses when your exposure is realised. This is guaranteed. On the
other hand, with a forward hedge you will always be afraid that at the last minute the hedge might turn into a
liability. Not that there’s anything wrong with that. The problem is that there is no shortage of people at any
organisation who do not understand this, and who will ask you questions, probing you to see what excuse you might
have to justify the hedging losses. This is the drawback of using forwards.
Often, salespeople who pitch to corporate treasuries an option hedge are reminded that banks make a lot of money
when they sell options. This is because people usually think that the banks just keep the option premium as profit.
Many think that selling an option does not have a cost for the bank. This is an obstacle to the use of options. No one
wants to be taken for a ride. By now, having read the previous chapters, you will know better. You know the
hedging activity that the bank needs to execute to limit the risk generated by the sale of those options has a cost that
equals the premium the option buyer spends.
How would our company hedge using an option? Recall the exposure we are trying to hedge, the company is
expecting revenue (is long sterling) of £10 million in one year. The execution person in treasury would call their
bank, or most likely electronically ask for a price on a one-year sterling put/dollar call struck at the forward rate for
£10 million. The bank would use software such as the one shown in Figure 13.9 to arrive at the price of
US$460,545.50.
You will recognise the pricing screen on the left from previous chapters, and you will recognise the hockey stick
chart on the right as that of a purchase of a put. Note that if sterling rises, the buyer loses the premium but, as
sterling falls, the option kicks in after 1.3673 (the forward rate) and continues to gain in value forever or until
sterling reaches zero, whichever comes first. Recall that the latter is not what the treasurer wants – they want the
option to lose the premium as this means that the underlying position of £10 million is gaining in value.
 Let’s put all this in one graph to better understand what the treasurer’s true position is (see Figure 13.10). There
are two items to consider. One is the existing exposure, which is the revenue expected in one year. The second is the
option the treasurer purchased, which is a sterling put US dollar call K=1.3673. When we combine the two
positions, an amazing thing happens. Look at Figure 13.10 – what does the “combined exposure” look like? That’s
right, it looks like a sterling call! We will now begin to add substance and the details to the building blocks we
showed you in Chapter 6. How can it be that a put turned into a call? Follow the example for a few lines in Table
13.1.
 In fact, we can generalise further and say that any put can be turned into a call and any call into a put by adding
100% of the opposite underlying exposure. This is a very important realisation as it ties together theoretically many
of the instruments we will show you in the rest of the book, and allows us flexibility to move risk around to where it
suits the treasurer’s risk/reward profile. This realisation is also at the core of the fact that option traders do not really
care if they are long a put or a call, as long as it has the same strike, because they can just turn one into the other
with a quick forward trade.
Therefore, this treasurer has now protected the source of revenue at the forward rate, which ideally is where it was
being valued, and will benefit should sterling rise to pre-Brexit levels. Of course, the company has spent a lot of
money on premium. As we’ve said, corporations hate spending money on premium. They will often balk at the idea
of spending 3.37% (premium/US dollar notional) of revenue to hedge. In all fairness to the treasurer, for some
corporations 3.37% far exceeds their gross margin. Does it make sense to spend so much to completely eliminate
any residual risk linked to currency moves? The solution to this problem is to identify the magnitude of the currency
moves that the company can withstand without major damage to earnings. Maybe typically, the business has a
budget rate for all its exposures, and in this case the GBPUSD rate, which defines the minimum such that the UK
operations are profitable, and this might give the company some flexibility. For example, the company may have
determined that this is a viable business within 5% from the current exchange rate. In such a case, the next best
alternative to the purchase of an ATM option is the purchase of a 5% OTM option.
The purchase of OTM options has many advantages. First, the premium decreases quickly as you go further and
further OTM. For example, if implied volatility is 10% across all strikes and the tenor is one year, then an ATM
option costs about 4%, which is a lot of money. To halve that to 2%, you only need to move the strike to 5.3%
OTM. In such a scenario, the option is basically used to protect beyond what the firm can already bear by itself. In
some sense, the purchase of the option is supposed to eliminate all scenarios where currency risk inflicts on the
corporation a loss greater than 5.3%. As we continue to examine different scenarios and different corporations, we
might realise that for some of them the pain threshold is much higher. Some can withstand losses up to 10% or 15%.
An option that is 5.30% OTM has a delta (with 10% implied volatility and one-year tenor) of about 30%. In essence,
there is a 30% chance of a currency scenario inflicting a loss larger than 5.30%. Thirty percent is a very large chance
of suffering a blow that the corporation cannot afford, so the purchase of the option makes sense. As the level of
pain tolerance increases, we raise this level. If the pain threshold is 10%, then the option costs only 1%, but the delta
is about 17%, so in this case there is just a 17% chance that the option will protect the company, all the other
possible scenarios implying a loss smaller than 10%.
 Second, in some cases, the corporation will purchase an option that has a very small probability of being
exercised, maybe 5%. So, you might ask, why would they bother for such a small chance? This is a very important
point. The chance might be very small but the damage in such a scenario might be unacceptable. It might mean the
complete annihilation of gross margin from one of the foreign subsidiaries. It might mean the entire treasury team
loses their jobs. Therefore, the spending of a small budget for the purpose of defending a very unlikely but
devastating scenario makes sense. This type of hedging is usually called tail risk hedging, as it exists on the tails of
our beloved normal distribution. Almost anybody has such scenarios in their business, as they do for their personal
assets. In fact, almost everybody has home insurance, since although the probability that your house is destroyed in a
fire is very remote, the risk is unacceptable. This is also why people purchase life insurance. In many countries, car
insurance is mandatory. In Italy, car insurance is at times referred to as “furto legalizzato”, or “legal theft”, due to
the high premia and low probability of a payout. The point is that the small, infinitesimal chance of inflicting
extremely large damage to a third party is not acceptable to the community.
In previous pages, we discussed the Black–Scholes formula at length, and have acquired a deep understanding of
its implications. However, it is not perfect and needs modification to accommodate the real world. One of the most
important modifications is to allow for higher volatility of very extreme scenarios. This was a weakness of the
Black–Scholes mathematical framework. As it was originally envisioned, it would assume that those very unlikely,
extreme scenarios were too unlikely. Over the years, the financial history of the world has been accumulating
evidence that those pernicious events happen more frequently than Black–Scholes would have us believe. This has a
very poignant psychological counterpoint in the behaviour of human traders, who on average tend to expect from the
future a simple repetition of the recent past. Reality tells us that the future repeats a past that incorporates all history.
Traders have a laser-sharp, tunnel vision focus on their narrowly defined trading mandate, so it is very
understandable that they would tend to relive their last few weeks when asked about the range of future possible
market outcomes.
Be that as it may, the literature dissecting this very famous cognitive bias is very large.3 The point here is that we
believe this type of hedging makes a lot of sense, and sometimes corporations do not want to bother spending a few
basis points of the notional exposure to take care of it.
So, what would the purchase of an OTM hedge look like for our sterling revenue scenario? Let’s price up a
typical OTM option. We will use a market convention for a typical OTM option by targeting a 25% delta. Given the
volatility level for the pound at the end of 2017, this turns out to be a strike of 1.2905 and approximately 5.62%
OTM. Note the implied volatility for this OTM strike is 9.48%, as opposed to the one for the ATMF option we
priced earlier, which was 8.58% on the offer (see Figure 13.11).
Once again, we need to combine this hedge with the known exposure already on the books, our revenue due in
one year. Take a look at Figure 13.12 and Table 13.2, and let’s highlight some observations.
In the last column of Table 13.2, the maximum loss is US$973,729! What’s that all about? We’ve insisted the loss
on the option is the premium paid, which was US$205,729. Where does the extra US$768,000 come from? Recall
the discussion at the beginning of this chapter. The way we will evaluate performance is from the rate at which one
could hedge this position in the market: the forward rate. Therefore, the logic at arriving at the potential loss is: the
premium paid + the opportunity forfeited or gain locked-in between the strike and the forward. In this example, the
premium paid is US$205,729 + the treasurer could have sold forward at 1.3673 but chose the OTM strike of 1.2905,
therefore ((1.3673–1.2905) × 10 million = US$768,000.
 While the upfront premium is a lot less than the ATMF option, the total loss could be a lot more. That’s why in
Part II we said that when buying an option, the premium is not all you could lose. Is this then a bad option to buy
versus the ATM? Not necessarily. Think of it in the context of insurance premium – you have the flexibility to
absorb a large deductible should the pound fall significantly, but you are not happy to risk significantly more than
5.62% plus the premium. In all situations, the company wants sterling to appreciate. Everyone should be rooting for
the premium to be forfeited.
 The OTM hedge breaks even a lot quicker than the ATM hedge should sterling appreciate. See Figure 13.13,
where we compare all three basic hedges.
Another way to think about this choice has to do with how confident one is that the pound will appreciate. If the
treasurer was 100% sure the pound would rise, they would not hedge at all. If they were uncertain or indifferent,
they would hedge with an ATM option, and if they were relatively certain but not 100% sure – or maybe just
couldn’t stomach the consequences of being wrong – then they would buy an OTM option. If they were sure the
pound would fall, they would hedge with a forward.

Forward-like instruments
At times, it is not the will that is lacking, but rather the ability to spend money to purchase vanilla options. In many
cases, the internal hedging policy of the corporation does not allow for purchasing options and paying a premium.
At other times, there might not be a budget set aside to pay for the premium. In such instances, the corporation
cannot pay for their options but can resort to a very old tool: bartering. The strategy approach we are about to
discuss is probably the most commonly used hedging approach for all the hedgers that employ options, not to
mention the favourite of many speculators as well.
Instead of paying for an option with money, a corporation could look around and find things to part with that may
be on the shelf and no one uses, needs or will notice if they are not there. You would be surprised at the kinds of risk
odds and ends that lie around corporate cupboards. As we know, being short an option (which is the same as having
sold an option) is dangerous as the potential liability is unbounded.4 Certainly, no corporation should entertain such
an irresponsible trade. Well, not so fast. To create some intuition around this way of thinking let’s briefly chat about
spreads. Look back at the resulting hedged positions. Our treasurer makes money as the pound rises to 1.40, 1.50
and 2.00. Do they really think the pound will rise to 1.50 or 1.60 and beyond? Even if it does, wouldn’t they be
thrilled to lock in, say, 1.50 as the top? The beauty of these markets is that there is someone out there that will give
them money for that 1.50 sterling call today. So why not pocket the cash? In this case, they would earn the spread
between where their revenue is marked-to-market, 1.3673, and where they sold that pound call, 1.50. In equities, this
form of strategy is called “covered call writing”.
In this section, we will discuss a few different flavours of option bartering. Buying what you need and selling
some form of what you don’t is the basis for dozens of different structures in FX options, as well as many other asset
classes. There are three different levers that are typically pulled on the sold option:

moving the level of the in-the-moneyness to make the option more or less expensive;
changing the notional amount;
adding a barrier or other exotic feature; or
all of the above.

We have not discussed barriers or exotics yet in this book, and we do not intend to delve deeply into the topic here
but we will give you enough to make these instruments comprehensible.

Collars or risk reversals

So what option would a corporation be willing to part with? Let’s look at some potential candidates in Table 13.3.
They may be happy to sell away profitability above all of these levels. After all, the revenue is marked at 1.3673.
However, recall the other firm objective – it cannot tolerate a dramatic fall in sterling beyond somewhere around
6%. So why not sell an option such that the premium that comes in equals the premium that will go out? If they want
to buy a 25% delta put that costs US$205,729, we can ask the computer to reverse engineer an option price to find at
what strike a pound call is bid at that price? From Table 13.3, we know it will be somewhere between 1.40 and 1.45,
closer to 1.45.
Let’s take a moment to mention a theoretical shortcoming here. In Black–Scholes as originally envisioned, all
vols across term and strikes would have the same volatility. In that case, we can simply say that, barring bid/offers, a
25% delta call will be the same price as a 25% delta put. However, alas, theory fails us. Reality says that in this
instance many more market participants want sterling puts than want calls, or more appropriately, the market expects
volatility to rise should the pound start falling so sterling puts OTM will have a much higher volatility that sterling
calls.
Here is how the whole thing could work. The corporation receives from the bank a one-year 25% delta pound put
dollar call, which is the option they wanted to buy. Instead of paying cash, they will give to the bank a one-year
pound call US dollar put that has the same value. The question is, what will the strike of the pound call be? We can
iterate using our pricing software until we get close or just ask the computer to calculate it for us (see Figure 13.14).
All option pricing systems have the capacity to quickly and seamlessly calculate this strategy. Here we used the
Bloomberg page OVML.
 Let’s see what this pricing page is telling us:

the one-year sterling put is what the treasurer really needs to buy in case of disaster;
the sterling call is struck so that it will give them exactly US$205,729;
the net outlay for this strategy is US$0 – yes, corporations really like that!
note that the strike on the put is calculated using a volatility of 9.48%, but the corporation will only receive a
volatility of 8.13% for the call – that’s terrible; and
the way the volatility difference manifests itself is in the strike difference from the forward; should the pound rise,
they will benefit from 1.3673 up to 1.4294, or US$0.621/£ × £10 million = US$621,000; however, if the pound
falls as we calculated in the previous section, they stand to lose from 1.3673 to 1.2905, or US$0.798/£ × /£10
million = US$798,000 – that’s the power of the volatility difference and the market’s vol skew.

There are many situations where this can work in the firm’s favour. Suppose, for example, this company had a one-
year payable instead of a receivable. Then this strategy, appropriately reversed, would be earning vol. We will
explore a more dramatic case at the end of this section, where a corporation can take advantage of the volatility
skew.
This bartering strategy is called a “collar”, a “zero-cost collar” or a “risk reversal”. It was originally called a
“range forward”. To our best recollection, Salomon Brothers unveiled this strategy for its corporate clients in the
mid-1980s. The name is really very clever for a couple of reasons. One is that it is, in fact, for a forward-type risk
profile but lacking a single rate, like a forward, the protection is outside a specific range. Within the range, the
position is effectively unhedged. The second reason is that corporates and the general hedging public like something
that has forward in its name, as they are comfortable with the idea it has a nice wholesome ring to it.
To re-address something mentioned in earlier chapters, the risk reversal naming is really very unfortunate since
market traders also refer to the difference between the two volatilities mentioned above as a risk reversal. Therefore,
the one-year, GBPUSD risk reversal is roughly –1.34% (9.48%–8.14%). The negative sign is arbitrary, we could
just as easily have agreed to use the opposite nomenclature. As we proceed, we will try to be clear to which risk
reversal we are referring to, although it is our hope that by now it will become obvious from the discussion.
Now the question is, what are the advantages of a collar versus just a forward? If we progressively increase the
delta of the sterling put and set it closer and closer to 50%, then the two strikes will also become closer and closer,
and at the end when they touch we will have a vanilla forward contract.5 Think about that for a minute, this is
another one of the building blocks we reviewed in Chapter 6. Let’s restate that: without bid/offer spreads, if one
buys a put and sells a call struck at the forward rate, their resulting position is that of a short of the underlying
instrument.
Therefore, the two alternatives are not that different from each other. The reason why many corporations will
prefer a collar is that it allows for some participation in any pound strength. If the pound strengthens, many in the
organisation, maybe ignoring the fact that the corporation has already hedged the risk, will expect an increase of the
US dollar value of sterling-denominated revenues. It is always handy to be able to say that, yes, the dollar revenues
are up. On the other side of the equation, if the pound depreciates by some moderate amount, to a level higher than
1.2905, then generally speaking most people in the organisation would not question that the revenues from the UK
might suffer with a depreciating pound. So, this makes the situation very defensible.
 In instances of large depreciation, below 1.2905, the sterling put will be exercised and any loss below 1.2905 will
be covered by the option. What about a scenario where GBPUSD goes above 1.4294? Of course, in such a case the
bank will exercise their sterling call and any participation in pound strength above 1.4294 will be taken away from
the company. The fact is the corporation has handsomely participated from the current forward to 1.4294, so there
will be an appreciation of the extra revenue to anybody who asks.
Let’s look at Figure 13.15 and Table 13.4, which summarise the above discussion. Can you see why it’s called a
collar? Table 13.4 shows the detailed numbers behind the figure.
The philosophy behind the collar is to be hedged when it matters a lot, namely for large moves in the exchange
rate, and not to be hedged for small moves. There are also other situations that are market-dependent and can make a
collar a very attractive hedge. Let us have a look at one of them. Suppose the corporation has a one-year expense
exposure to Brazilian reals. As we discuss the favourable circumstances for this treasurer, recall the earlier
discussion of the Brazilian corporate that had the opposite risk and needed dollars to pay back their US dollar debt.
For now, let’s look at a happy case where the company is short reals and long US dollars. We price a collar as we
did before, picking a tense time: December 14, 2017 when Brazil’s crucial pension overhaul stalled. Figure 13.16
shows the pricing of a risk reversal in USDBRL where the hedger buys a US dollar put Brazilian reals call, and sells
a US dollar call Brazilian reals put. There is a very peculiar detail about this structure – can you find it?
 In this case, we priced the purchase of a US dollar put Brazilian reals call with a delta of about 35%. The
interesting detail is that this delta corresponds to the strike being equal to the spot rate. You may be unnerved that a
strike equal to the spot rate would not generate a delta equal or close to 50%. Remember that the underlying asset in
the Black–Scholes formula is not the spot rate but the forward rate. The spot rate in this case is R$3.31/US$, but the
forward rate is roughly R$3.52/US$ (we will drop the US$ signs for the rest of the discussion as you are by now
accustomed). Therefore, the strike on the US dollar put is about 21 big figures lower than the forward rate. The
strike on the US dollar call Brazilian reals put is 3.8709. You would expect this to also be 21 big figures from the
forward, which would correspond to 3.52 + 0.21 = 3.73. How come it is 3.8709 instead? As you will recall from
previous chapters, and the sterling example earlier, the implied volatility used to price the US dollar put will be
lower than the implied volatility used to price the US dollar call, the latter normally being higher due to the stronger
demand for the weaker side of the market. In this case, the volatility of the US dollar put is 13.59%, but the volatility
of the US dollar call is about 15.57%, so a difference close to 2%.
A nice exercise would be to price the US dollar call with the same volatility as we price the US dollar put. If you
do that you will see that the price of the US dollar call goes from 3.56% to 2.75% (running all our numbers at mid-
market for simplicity). This means that the difference between those volatilities, also known as the “risk reversal” or
the skew, gives you an advantage. To compare, the difference between the spot of 3.31 and the forward of 3.52 gives
you an advantage of 0.21/3.31 = 6.43%.
Those two advantages combine to create a structure where the hedger receives an ATM spot option and gives in
exchange a 27% delta option. If you compare possible scenarios based on the current spot market levels, you can
verbalise the situation by saying that: the hedger has zero downside and a potential gain of (3.8709 – 3.31)/3.31 =
17%. Not a bad hedging structure! Of course, this is not totally true for a hedger because they will be marking their
book at the forward, not at the spot level.
In summary, we have seen how in some specific market situations the hedger can use both the forward market and
the options market to their advantage in crafting hedging structures. Most corporate hedgers would not possess
either the required market expertise or the time to acquire such expertise, and therefore it will be the job of the
bank’s structurer to bring those market quirks to the attention of those who can take advantage of them.

Participating forwards
This structure is very popular and marries different attributes of forward and option hedges to get the best of both
worlds. The essence of this trade was explained to me by a veteran salesperson, Mark, who started the business in
the trading pits of Chicago. Legend has it that he used to play bridge socially; he was tall, had a deep voice and was
quick with mathematics. One day, one of his bridge companions offered him a job at the exchange to trade FX
options. The way Mark saw it, if you compare options with the two alternatives of not hedging or hedging with a
forward, the option is never the best performer of the three. If spot rises, then the best would have been not to hedge.
If spot falls, then a forward hedge would have been better than an option, the difference being that a purchased
option implies paying the premium. Not hedging being too risky, Mark proposed creating a blend of an option and a
forward. You might think that the best blend would be to hedge half the exposure with a forward and half with an
option. Yes, good try, but your strategy would require the payment of the premium on the “half” option bought, and
nobody wants to pay premium. However, there is a better way, which is called a participating forward. For a US
corporation with a euro expense exposure, this consists of the purchase of a euro call US dollar put, and in the sale
of a euro put US dollar call in a notional equal to, say, 60% of the exposure. The strike on both options is the same,
and is calculated in such a way that the premium on the euro call is equal to the premium on the euro put, and
therefore the whole structure can be transacted at zero cost. In other words, no cash changes hands on the trade date.
The previous strategy, the collar, had two option legs that allowed the hedger to move the strike of the protection
according to their risk/reward preferences; and derive the second leg from market pricing, such that the costs are
equal, while holding the notional amounts equal. For the participating forward: start with the same protection levels
consistent with our hedging policy; then strike the second leg, the sold euro put, at the same strike. Something has to
give, so vary the notional so that the premiums are equal.
You must be thinking: what nonsense, how can the notional be less and the premiums equal? What form of
chicanery is this? It will all make sense when you observe what transpires. The strike is above the forward rate, so
that the euro call is OTM and the euro put is ITM. Then you might think that the euro put should be more expensive,
but remember that the notional on the euro put is smaller, so things can be arranged as to perfectly obtain a zero
aggregate premium for the structure (see Figure 13.17).
Let us look at a concrete example, priced on the eve of the famous Bernanke’s “taper tantrum” of May 5, 2013.
The spot level at the time was 1.3077, but the strike on both options is 1.3360, with the forward very close to spot, at
1.3119. Therefore, the strike on the hedge is not very good. The euro has to rise by (1.3077–1.3360)/1.3077 = 2.16%
before the hedge starts protecting the exposure. However, even worse, if spot does not move at all, the euro put US
dollar call will be ITM for the bank, which means a loss of 2.16% × 60% = 1.296%. That does not sound like a good
hedging structure. But look at it this way. First, no premium is paid for a full hedge on the euro upside, which is
admittedly 2.16% OTM, but fully protects in case of a euro rise larger than 2.16%. When we say “fully”, we mean
that any 1% rise of the euro after an initial 2.16% rise is fully matched by a gain of 1% in the payout of the euro call.
Also, look on the bright side in case the euro falls. It is true that in such a case the bank will exercise the euro put US
dollar call, but keep in mind that the notional is only six million. In other words, 60% of the theoretical gains due to
the fall of the euro are passed onto the bank, which means the hedger gets to keep the remaining 40%. In other
words, if the euro falls the hedger will buy in the spot market (at a lower rate) €4 million, and will buy from the
bank at 1.3360 (off-market, a losing trade) the remaining €6 million as the settlement on the euro put.
 This trade has a few very appealing characteristics. It is zero cost, and provides full hedge protection after an
initial, small loss in the case of euro appreciation. For euro weakening and consequent questions from senior
management, the corporation is able to state that such euro weakening has resulted in a decreased US dollar value of
the euro-denominated expenses. Everybody is happy and there is no regret. These attributes make the participating
forward a much better alternative than many other structures, including a half-forward, half-option combination.
Also, bear in mind that the 60% level on the euro call notional can be decided by the hedger. If less protection is
chosen, then there is more participation in euro weakness, and if more protection is needed, the strike on both
options would move lower, getting closer and closer to the forward rate.
The participating forward also represents an excellent opportunity to examine option pricing practices in more
detail. The way we have described the structure, the hedger would theoretically trade on the ask for the euro call and
the bid for the euro put. However, this pricing technique would be unfair, as it would cost the hedger 1 + 0.6 = 1.6
bid-mid spreads, based on the asymmetry of the notionals. The same structure can also be priced with the hedger
buying €10 million forward, and the hedger buying 40% of a 1.3360 euro put US dollar call. Think this through, and
you will convince yourself that this gives exactly the same payout at expiry. But in this second pricing technique, the
volatility spread is only levied on the ask of the 40% euro put, for a total of 0.4 bid-mid spreads – ie, much cheaper.
Just a little detail for the technically oriented – the reason we mention this here is that many corporates, thinking
they are checking the bank price, would price this up in Bloomberg or some other system, and then the bank
salesperson would say, often out of ignorance, use this vol for the offer and this for the bid. The client would then
get a much worse price than using our latter pricing methodology. Think of the fact that the buyer of the structure is
going long vega at the same strike, so you would pay the offer; you should not have to pay the offer and receive the
bid. What should matter to the pricing is the net amount bought or sold by the hedger at that strike.
Before we move on, let’s consider the same example in UK sterling and use the same protection level 1.2905 as
we did for the OTM vanilla and for the collar so that we have a basis for comparison. Remember the company is
long £10 million, expecting revenue in one year (see Figure 13.18).
Note that:

the strike we chose is sufficiently away from the forward that, to raise the premium of US$225,484 to pay for the
protection at 1.2905, the treasurer has to sell a sterling call option with a notional of only £2,362,731;
for any level above 1.2905 at expiration, the company can participate in 76% of the upside – typically, these are
structured with protection close to 35% delta and the participation turns out to be closer to 50%;
there is no upfront premium, but there can be a loss!;
the maximum loss is US$768,000 – remember from our collar discussion, the revenue is being marked at the
 forward rate of 1.3673. ((1.3673–1.2905) × 10 million = 768,000); and
the owner of the structure is long vega (volatility).

In Figure 13.19, you can see that the steepness (slope of the option lines) is different depending on the notional
versus the underlying exposure. The revenue line is 45 degrees, and so is the full protection, the put, at 1.2905.
However, the sterling call, which is only ~23% of the revenue notional, is at a much lazier angle.

Forward extras, or knock-in forwards

The final structure we will discuss as it pertains to corporate hedging is the forward extra. As with the two previous
structures, this is a classic type, and one adopted by a large percentage of the corporations that employ option
strategies. It is also a zero-cost structure in the sense that no premium changes hands at the time of trading. As
mentioned, these are also called forward-type structures in that, just like a forward, they don’t have a premium and
trade at zero cost, but also in the sense that no matter what happens the corporation will be hedged. In this case, it is
actually not true as there are some scenarios where the corporation will not be hedged, but they are actually benign
scenarios because in such instances not being hedged is a good thing. Finally, like any of these strategies, the hedger
buys full protection in, say, a put, and does something to the call they sell to adjust the risk/reward payout. In the
case of the forward extra, we change the other leg in a way that we have not discussed so far, as we have mostly
stayed away from discussing “exotic” options. We will go a bit deeper into these when we talk about how hedge
funds speculate with options in the next chapter, but for a fuller discussion you’ll have to wait for the next book!
 So, what is different about this call? The notional will be 100% of the underlying position and the strike will be
the same as the protection level, just like the participating forward. The difference is that the outcome at expiration
will be influenced by the path spot takes between now and expiration. Specifically, we enter into the design of this
option a clause that says: this option the hedger sold does not exist! However, should spot trade at or beyond a
certain level, this option magically appears (knocks in)! Well, if now I have 100% of the notional of a put and a call
at the same strike, it means I have a forward at that strike level. Therefore, the hedger goes from having free
protection to potentially being locked into a forward at a predefined level of their choice. In theory, the hedger
should be comfortable with this.
 The best way to understand this structure it is to take a realistic case and go through it. Back to our stereotypical
example of a US-based corporation with the need to hedge euro-denominated revenues forecasted to happen in one-
year’s time. The structure consists of the purchase of a slightly OTM euro put US dollar call that is a vanilla option,
and in exchange for that the corporation will sell to the bank a knock-in option, which is a euro call US dollar put
with the same strike (so the euro call is actually ITM), but with a knock-in level higher than the forward rate. The
presence of the knock-in level means that if EURUSD never trades at or above the knock-in level (also known as
“the barrier”, or “the trigger”), then the euro call US dollar put does not exist. However, if at any point before expiry
EURUSD trades at or above the barrier, then the knock-in option will instantaneously become a vanilla euro call US
dollar put, with such strike as mentioned already.
Is the corporation really parting with anything? The knock-in option does not exist unless, of course, at some time
in the future EURUSD breaches the trigger. Well, this same argument could be applied to any OTM vanilla option,
all of which have an intrinsic value of zero today, but they still have a price.6 They carry with them the possibility of
becoming an asset and will never be a liability, so their value is always positive.
Once again, we will use mid-market for all prices and pick May 1, 2013, right before Bernanke’s taper tantrum.
Figure 13.20 shows the pricing of a knock-in forward a US-based corporation could use to hedge euro-denominated
revenues.
The forward rate is 1.3110 at the time of pricing, with spot at 1.3066. The strike on both options is 1.29, so about
two big figures below the forward. This is bad news, as before the option protection starts to pay out, the corporation
will sustain a loss of two big figures. The euro call knock-in trigger is 1.4389. Let’s concentrate on the positives.

First, the whole structure is zero cost, and therefore not having a budget for option premium is not a problem.
Second, the 1.29 euro call US dollar put the corporate sells to the bank only comes into existence if EURUSD
trades at or above 1.4389. At the time of pricing, the last time EURUSD traded at 1.4389 was almost two years
before, in August 2011, right before the flare-up of the first Greek sovereign debt crisis bailout. This was also
the time that the US government debt was downgraded. Enough of ancient history, let’s look at the future.
In fact, knowing what we know now, 1.4389 will not trade during the life of this structure. The high will end up
being on March 18, 2014, at about 1.3934. Therefore, with hindsight, this structure was a winner because the
liability never materialised for the hedger.
At expiry of the structure, on May 2, 2014, EURUSD was at 1.3870. This does not only mean that the liability
never materialised, it means much more. It means that the gamble with the devil paid off. The corporation was
now able to sell euros at a much higher rate than 1.29 or 1.3110. As the liability did not exist and the asset, the
1.29 euro put US dollar call, was OTM, the corporation was effectively without a hedge or, with less extreme
language, the hedge was OTM. If we say it like that, we make it sound like a bad thing. However, the
corporation was fortunate not to have a hedge at the policy prescribed levels. Since without commitment to any
locked-in prices, at a time when the euro rose they were free to sell their euro-denominated revenues at 1.3870,
a much higher rate than 1.3110, close to 6% better. So, in this specific case the structure had worked
wonderfully.
What could have been the worst-case scenario? Had EURUSD gone to 1.45, the liability would have been
awakened into existence by the breaching of the barrier (or the knock-in level) at 1.4389. At that point, the
corporation would have been bound to sell their euros at 1.29, a full 12% worse than the then-prevailing market
rate. This would have been harsh, but remember that a vanilla forward contract at 1.3110 would have been
10.6% worse, so comparatively the corporation had only given up 1.3110 – 1.2900 = 0.0190, or 1.27% on the
worst-case protection level.
Had the EURUSD rate been getting close to the 1.4389 trigger level, 12% of the European revenues value would
have hinged on that fateful level. Many in corporate treasuries would have, or should have, lost sleep on such a
huge risk. If they did not, then we have a bigger problem. As a comparison, there are many US exporters whose
gross margin is smaller than 12%, so a clumsy decision on the part of corporate treasury could have wiped out
all of it in one swipe. This is often the great paradox of corporate treasury. Decisions taken by a small team of
just a few FX managers can have dramatically oversized consequences for thousands of employees and
shareholders. This is often difficult to grasp by the board of directors and the press, and all too often the result is
that people might be too quick to cry foul.
As an aside, when the time to expiration is short and the trigger is close to spot, it’s not just the corporate
treasurer who loses sleep. Absolutely mayhem ensues. Dogs are sleeping with cats and the deltas are well above
100% and into the thousands. See Figure 13.21 for a case we have been examining with some new market
assumptions. The date is just a day before expiration and let’s assume the spot rate is really close to the trigger.
The delta is 1,935%! The price of the euro call is US$1,348,901. It all makes sense right? If the spot rate rises
just a little bit more, the option that did not exist will now have a minimum intrinsic value of (1.4389–1.2900) ×
€10 million = US$1,489,000! The huge delta is telling the trader that to generate enough P&L variance quickly,
to go from zero to US$1.5 million to potentially zero, the trader will have to trade huge amounts of EURUSD
spot. These are really fun and interesting instruments to trade and hedge with, but one does need nerves of steel.
How do we apply this strategy to our UK sterling revenue discussed previously? We start by going to our pricing
system and asking to price a forward extra. Most systems will come up with all the necessary fields and you fill in
the blanks. Alternatively, one can price a put of your choice and then iterate the price of a knock-in call until the
prices are equal. Note the two new fields: “barrier type” and “barrier level” (see Figure 13.22).

There are two barrier types we care about here. American means that if spot at any time reaches the barrier, the
option knocks in. The other type is European, where the barrier is looked at only at expiration. If spot is above the
barrier when the option expires, then the option is awakened. Therefore, if spot happens to cross the barrier and then
retreats back below the barrier by the day the structure expires, then the option is still non-existent. Just for fun,
think the following through. Which knock-in option is more expensive – one that can come into existence any time
spot crosses the barrier (American) or one where it can only come alive on one day at expiration (European)?
What are the details? To compare the forward extra with the other instruments discussed, we will again assume
the treasurer hedged their sterling revenue at 1.2905. We know the spot to be 1.3500 and the forward 1.3673. It turns
out that the sterling call struck at 1.2905 will have a barrier that is massively away from spot, at 1.5791. Awesome!
That is a pre-Brexit level, really far away from today’s spot and forward.
Let’s examine the greeks a bit more carefully because something very spectacular happens with this type of
strategy. Figure 13.23 shows the greeks relative to the pricing of the forward extra on GBPUSD under consideration.
The first column to the left shows the greeks of the aggregate structure, while the second column shows the greeks
relative to the plain vanilla sterling put, and the last column shows the greeks relative to the knock-in option. Note
the hedger is short volatility but long theta, or time decay.
 In a typical strategy where net premium is paid, one would expect the hedger to be long gamma. which means the
market-maker trader who is their counterparty will be short gamma. The short gamma position holder will pray for
spot not to move up and down unpredictably, so that on the delta hedging the trader will not lose money selling low
and buying high. If spot starts to jump around, this will also mean that implied volatility may potentially increase,
making the option structure owned more precious. It is usually not possible to be long an option structure and hope
volatility goes down, except with our forward extra. To emphasise, the forward extra greeks are negative gamma
and, very importantly, are negative vega!
Therefore, this structure seems to give an opposite position in vega and gamma to what one would expect.
Negative vega means the owner wants volatility to fall. Stated differently, suppose the world is very tense and
consequently implied volatility is very high – in which case, most option structures will be very expensive. To hedge
with options, large premiums will need to be paid or potentially terribly unfavourable hedging levels for a collar or
participating forward. With a forward extra, the hedger is able to use the market uncertainty in their favour!
This is a very interesting and subtle observation. The hedger seems to like the fact that, if volatility falls, they will
be better off. As we have just explained, this comes from the fact that higher implied volatility levels improve the
parameters of the structure such as strike and barrier in their favour. However, this is just another figment of our
imagination. Selling volatility (ie, entering into a structure that makes you short vega) is not a good or a bad thing
per se, just as selling spot or buying spot is not good or bad per se. Often, people just imagine vividly a baseline
scenario where spot does not move and stays at 1.35 for the whole life of the trade. Human psychology comes into
play and distorts scenario probabilities. It is extremely difficult not to imagine a static market as a very likely
scenario. In this “prime” scenario, volatility is nonexistent. In our minds, we tend to give disproportionate weight to
this “base” scenario, hence the popular phrase “selling volatility”. Some traders even consider “selling volatility” to
be one of the typical types of trades that tend to make money all the time, such as the FX carry trade (sell US
dollars, buy Brazilian reals on the back of the interest rate differential), or the interest rate carry trade (buy 10-year
Treasuries at a high yield and sell two-year Treasuries at a lower yield), or the credit carry trade (buy high-yield
bonds and sell investment-grade bonds). Selling volatility almost becomes an asset class in its own right.
Now let us come back to the payout analysis of the forward extra. Take a look at Figure 13.24. The forward extra
will have two potential outcomes. The first is the dark grey line that occurs when the trigger is hit and the hedger has
a forward position at the strike level. Note that it’s below the zero line, and shows up as a loss at all levels since the
hedger could have hedged at the forward but chose to take the calculated risk. Of course, the second potential
outcome is when the risk pays out and the trigger is never reached while sterling rises! That is the light grey line.
Note further that it looks just like an option. However, it does not continue forever but only to the trigger level,
which is not shown here.
Table 13.6 has all the details, which by now you can calculate in your head.
This structure – the famous forward extra – is an interesting example of FX hedging. There is a necessary part that
is linked to performance in basic hedging objectives, but there is also a large component of risk. Risk comes in many
flavours. The forward extra really brought home the fact that, more than other structures, the performance of
financial decisions is linked to randomness and cannot be judged from one event. The world-famous racing driver
and four-time Formula 1 world champion Alain Prost famously said: “This is a business where you cannot judge a
race in isolation. You can judge a season or a career, but you cannot judge a single race.” Had 1.4389 in the euro
example, traded before May 2, 2014, would you have changed your mind about the attractiveness and comparative
advantage of forward extras? In the UK sterling example, had the pound reached 1.5591 and had the hedger been
locked into a forward at 1.2905, how well would the forward extra have been received by the CFO or the board?
FINAL THOUGHTS ON CORPORATE HEDGING STRUCTURES
We will limit our laundry list of hedging strategies to the ones described above. We’ve left out some really great
strategies such as average rate options that are used by a lot of corporations. The combination of structures one
could construct using the basic building blocks like vanilla options, forwards and barrier options are infinite. We
therefore drew an arbitrary line: it is the line that separates discovery from boredom, but of course it only exists in
our imagination. In general, these structures are of the same nature, where protection is bought and something is sold
to pay for it. We shall give you a final treat in Chapter 14, where we tease the reader by describing in excruciating
detail one of the most complex and mind-bending structures we have encountered. We think it will be good fun to
exercise your judgement in terms of risk/reward.
In closing, we stress the fundamental concept that using options does not in itself give you any edge or create
superior performance. All options and option structures are created equal, in that their net present value is equal to
zero.7 So, you can structure and trade all day long, your expected net present value dial will not move. However,
there is one thing using options allows for – it modifies the odds. Options allow changes in the risk profile of
positions in ways that will grant corporate treasury staff the ability to sleep soundly at night: they can protect the
business from the vagaries of adverse fortune.
No hedging strategy has the power to change expected performance. Hedging strategies can only change the
probability distribution of the performance, not its centre of gravity. But this is equivalent to changing the risk
profile. As a cautionary tale, what sometimes happened during our dialogues with corporate treasuries is that the FX
managers work hard to increase their understanding of different hedging structures and their fine points. Then comes
the time when they call the bank sales desk, all excited to share their latest grand idea. They have decided to embark
in a huge Monte Carlo simulation that will compare the expected performance of a bunch of hedging strategies so
that the firm will be able to choose the best one among all. The salesperson could tell them on the phone the result of
the ambitious mathematical project without running the numbers, although it may be awkward. Under the hypothesis
of no-arbitrage, all hedging strategies have the same expected performance: break even.8
For example, if they are planning to expand in Australia by building or buying major assets there, Warren
Buffett’s advice is probably best. Do not hedge the value of your plant on a monthly basis. To the extent that
corporate treasury needs to hedge translation risk, plant included of course, they can do that with average rate
options (not discussed here). If the company can issue debt in Australia to mitigate some of the Australian dollar
revenue, that’s definitely something to consider. However, if dealing with a large ongoing business, it is most likely
that the treasury will also need to use some or all of the other approaches we have discussed.

1 The difference between the forward rate and the spot rate is called “the forward points”. Often at the time of trading, the market risk is divided in spot
and forward points. Operationally, a corporation might decide to first of all eliminate the spot risk by selling sterling and buying US dollar spot. Once
the trade is executed at a rate of 1.3500 and the spot risk is eliminated, the corporation then proceeds to eliminate the forward points risk by asking
banks to quote on the forward points. The winning bank will therefore deliver the forward points to the corporation at 0.0173, the best price in the
example. The winning bank will transform the pre-existing spot trade at 1.3500 into a forward trade at 1.3673. The net result of the whole operation is
for the corporation to have a forward trade at 1.3673 and no spot trade in place, which was the goal all along.
2 As an aside, the fact that the figure is a straight line also tells us that a forward contract will have the same price no matter the level of implied volatility
for a given exchange rate. If volatility is higher, so is the chance of large losses, but by symmetry so is also the chance of large gains, and the two
impacts wash off.
3 See N. Taleb, 2007, The Black Swan (New York: Random House) or D. Kahneman, 2013, Thinking Fast and Slow (New York: Farrar, Straus and
Giroux).
4 Not always, but bear with us for sake of argument. Why not always? Can you find an example where the liability generated by having sold an option is
bounded?
5 We have already practiced this nice little exercise in one of our previous endnotes.
6 This could take us into a very interesting metaphysical discussion about whether an OTM vanilla option exists right now or does it just possesses the
potentiality to exist in the future.
7 We can already hear everybody screaming, but here comes the fine print: “after premium has been accounted for”.
8 Assuming no transaction costs. When we include transaction costs, the best strategy is no hedge.
 14
Situations Gone Mad, From the Most Complex to the
Simplest

In this chapter, we will explore two situations that in hindsight look absolutely insane. We continue with the hedging
theme of the previous chapter, and examine very complex structures that are appealing to every weakness in our
human nature and often are just not worth the risk: TARNs. We will spend some time understanding what it is that
makes complex trades interesting, before presenting two examples that should be revealing as to how they work.
The second has to do with a situation created by the Swiss central bank’s desire to manage economic conditions and
trade relationships, but in the end proved devastating for those market participants that were happy to play on train
tracks by jumping around the third rail.

COMPLEX TRADES TEND TO HAVE HIGH HEDGING COSTS

The following discussion will hopefully be clear, but be prepared to jump into nomenclature that we have not spent
time carefully building up. We introduce and analyse a specific trade that represents to some extent the tip of the
iceberg in terms of the complexity of FX trades. Such complexity initially depends on the liquidity in the market
since more complex trades generate more hedging costs. Those costs are of two types: the hedge is itself a complex
trade; and the hedge needs to be rebalanced more frequently, thereby causing traders who have to manage their risk
to pay more bid and offer spreads.
Starting with the first problem, as we have seen many times already in this book, a real market-maker in options
does not do back-to-back trades. Back-to-back means that they will sell the trade to a customer and buy it back from
another bank. In general, this might sound like a good idea on the premise that the bank will give you a better price
than you gave your client. The reality is that if you ask a price from another bank and the trade is very specific, with
particular dates and other details, it will be obvious that you have just transacted the same structure with one of your
clients. This will make your dealings with other banks very arduous. They will show a wide bid/ask spread, and in
all likelihood their price will be worse than you gave your client. Therefore, this game could work for spot or
forward trades, but will not work for exotic or bespoke option structures. At the end of the day, why should the other
bank do you a favour? They are in the market-making business, competing with you for the same client’s business.
The second problem is also a big headache. As the complexity and customisation of the structures sold increases,
the ongoing hedging cost, after the initial hedge is put in place, increases. I remember some banks had probabilistic
models of how many times a hedge would need to be adjusted to estimate the expected hedging cost. As we already
know, when the trade is sold to the client the trader will put in place an initial hedging structure, mostly composed of
vanillas, but at times requiring some one-touch or double no-touch options.
There is a cost associated with trading those in terms of bid/ask, but the real problem is whether or not the initial
hedge needs to be modified later. Let’s look at a very simple example. Suppose a trader sold a digital option. This
option is very simple, and is really the kind of bet we are used to making with our friends in casual conversation.
The bank offers the following deal: pay some sum of money upfront and, if at expiry EURUSD is at or above 1.20,
then the bank will pay the client US$1 million. Otherwise, the client gets zero and loses the upfront premium.
The bank trader could decide to just delta hedge this trade. If the delta is 35%, then they buy US$350,000 dollars
and the initial hedge is done. After a week, EURUSD has moved and now the delta is 225%. This is not a typo – in
digital options the delta can very easily be above 100%. Think of it this way: as the seller, you need to have on hand
enough dollars so that, on the next move up, you will earn enough to eventually have the US$1 million on hand if
we reach 1.20. They adjust the hedge and incur bid/ask costs. This happens continuously. Of course, they can decide
that EURUSD will instead return to the previous value, and so decide not to adjust the hedge. However, this
approach reduces the issue to playing a game of chicken with the market, and creating a speculative position instead
of hedging, right? Market-makers do not like to do that, and in the cases where they do, their management certainly
does not like it. It then follows that the delta hedge will have to be adjusted a number of times, at great cost. To
clarify, the cost we are referring to is not the cost referred to in previous chapters, that occurs when a trader is short
gamma. For that activity we assumed mid-market trading. Here we are mostly taking into account trading costs in
terms of the bid-offer spread that needs to be crossed with every hedging trade.
Given those hedging difficulties, what is the incentive for a market-maker to execute very complex or bespoke
trades? The incentive is that those trades are customised and relatively few banks can offer them. In other words,
very few banks can offer competitive prices on them. The end-game is that the salesperson will be able to extract a
larger mark-up on those trades to compensate for the potential higher cost of hedging.
To give you a clear example, consider hedging structures that can have an aggregate delta of US$40 million. A
forward trade with that size might have commanded a mark-up of five basis points, which means US$20,000. The
bespoke structures commanded a mark-up at times equal to 1.6%, which is 32 times larger, equal to US$640,000.
The incentive for the market-maker and the salesperson – in other words, for the bank – is clear.

CREATING ATTRACTIVE TRADES

What is the incentive for corporate clients to participate? As we will see, the trade looks very attractive. It is logical
to assume that a trade, with the mark-up described above, has a very low probability of turning out favourably for
the client, as the higher the mark-up in a trade the lower the probability that the trade will outperform any
benchmark, like a simple forward hedge.1 You might be surprised to know that many clients executing the trades
described below for years, and always outperformed the benchmark. This can often be a win–win situation.
There are a few classic methods to make a trade look good. Those are due to cognitive biases that good
salespeople and structurers learned to use masterfully. We defer to Daniel Kahneman, Amos Tversky, Dan Ariely
and all the other behavioural psychologists, and refer the curious reader to their books, or maybe to the excellent
account by Michael Lewis.2 Here are some intuitive examples.

Suppose a trade outperforms if the market does not change or move. Spot does not move and the hedger makes
money! This is one of the most powerful ways to make a trade attractive.
The second way is for the trade to be short vega, or sell volatility. This is very similar to the first trick, but not
necessarily the same.
A third way is to use the carry. Typically, a trade that sells a low-yielding currency and buys a high-yielding
currency.
Another little trick might be to hide unpleasant payouts for scenarios where the exchange rate has never historically
been. A lot of people who claim long experience in the market will very often dismiss spot scenarios that they
have never experienced.

The best way a smart salesperson can make a trade look good is to do their homework. Before they spend hours on
the computer to find the best structure to pitch, they spend a lot of time on the telephone with their client to
understand their market view. They understand where the client thinks the rate will be in the short, medium and long
term. What they see is the ceiling to any possible rally and the floor to any possible fall. They ask them why they
think so, and why this time is the same or different than in the past. Once they understand all that, the structurer can
create a trade that pays out handsomely in exactly the market scenario the client has described, and the favourable
payout is financed by taking on risk in the scenarios the client does not think are possible. The client will not care
that much about all those ugly losses for the simple reason that they don’t believe those scenarios will happen. So
much for us playing amateur psychologists.
The point is that the more detail the client reveals about their favourite market scenario, the more mathematically
unlikely it becomes. For example, a statement such as “within six months the EURUSD will reach 1.30” may have a
probability of 30%. However, if I add other provisions, such as: “the EURUSD will plunge to 1.10 within three
months, after which time, it will reach 1.30 within six months”, it could have a probability of 5%. The latter
represents a much better opportunity of creating a bespoke trade that realised such a detailed market view because it
will be much less expensive.3

WHAT CONSTITUTES GOOD PERFORMANCE?

After having spoken a little about the incentives for corporate clients to transact esoteric bespoke structures, it
should be clear that the whole game will depend on how the performance of corporate treasury is evaluated. Every
FX manager at a corporation wants to have a positive review at the end of the year. They will not be paid based on
their effective hedging ratio, so strictly beating a benchmark is not exactly their main objective. Keep in mind that,
even if they should, many corporations do not have a benchmark, so it may be easy to claim a positive performance
based on considerations crafted after the fact. The more important elements for the internal functioning of corporate
treasury are processes. To some extent, this is the single most fundamental difference between corporate treasury
and a hedge fund. For a hedge fund, the market changes continuously, trade ideas last a month and opportunities for
high performance need to be invented every day. For a corporate treasury team, the opposite is true. Best practices
and streamlined processes are the most important objectives. They seek continuity, elimination of errors, strict rules
that last for 10 years, and idiot-proof systems to execute, confirm and settle. Ideally, they would love for their entire
workflow to become completely mechanistic. However, this creates a contradiction with another goal they have:
each FX manager in a corporation wants to leave a lasting mark on the department before migrating to other
departments, and eventually to the CFO suite. Some of the typical ways to leave that mark include introducing
Straight Through Procesing (STP) for confirming trades, or changing the hedging policy to include new accounting
approaches, or establishing risk calculations based on VaR methodology. Of course, they can also include exotic,
bespoke option structures that improve the exchange ratio on some select FX exposures. Here is where our
structures come in. They allow, if everything goes well and the market view is correct, large improvements in the
hedging rate. Improvements that would not be imaginable using forward hedges.
Let us now briefly consider the attractiveness of exotic bespoke option structures for hedge funds. This is a rather
obvious point to make. The single most important factor for a hedge fund is leverage. Bespoke structures permit
leverage increases to unprecedented levels. Why do hedge funds need leverage? Hedge funds are generally
constrained by their credit. They depend on the banks to deal. Every FX position includes some degree of credit
usage. The exception to that rule is a vanilla option, which allows a hedge fund to take a large position with virtually
no credit usage, because after the premium has been paid to the bank, the bank has no exposure to the hedge fund.
The bank will never have a loss if the hedge fund goes bust.
What is true for vanilla options is even more true for some exotic options. Think, for example, of a worst-of-three
option. This is an option structure whose payout is the least of the payouts of three specific vanilla options. Those
could be, for example:

an ATMF three-month € call US$ put;

a 25-delta three-month US$ call ¥ put; or
a 15-delta three-month £ put USD call.

If any of those three is OTM at expiry, then the worst-of option pays zero since the least of the three payouts is zero.
However, suppose all three are ITM. The payout is the worst of the three. Worst-of options are generally very cheap,
of the order of a few dozen basis points, depending on the moneyness of the vanillas and the tenor. The payouts can
be much larger. Therefore, they offer tremendous leverage and hedge funds love them.
There is a final reason why hedge funds really like bespoke structures. As you might know, one of the favourite
types of trade for hedge funds is a relative value trade (there will be more detail on relative value in Chapter 15). By
this, we mean a trade that depends on the relative price of two or more instruments. Typically, if those are
historically correlated in a particular way a hedge fund may look for deviations from that pattern. If their spread
widens relative to history and expectations, the hedge fund may bet that the spread will return to its historical levels.
This is the type of trade that Long-Term Capital Management (LTCM) entered into back in 1998, and they are also
known as mean reversion trades. A bespoke structure can easily incorporate a spread or even the relationship among
three or more instruments.

TWO TRADERS, TWO VERY DIFFERENT APPROACHES

For this and the next chapter, we want to highlight a fundamentally different approach to risk that some traders take.
To do that we will caricature two different market players that will be attracted to different risk/reward outcomes.
Let’s introduce our protagonists that will guide us through this. The first trader is Christine. Christine is cognisant of
her sophisticated understanding of the market. She understands relationships and will try to make money by
exploiting relative value trades, specifically correlations between asset prices that have deviated from normal.
Christine will position her portfolio and wait for the right time, when differences in specific assets deviate from what
is natural.
Christine’s nemesis, Bonny, is a different type of trader. We’ve actually met Bonny previously but we just did not
name her. Bonny is the trader who wanted to have the ability to make 1% in a great day at the office. Bonny does
not think it is possible to make a lot of money using relative value trading, and besides she doesn’t have the patience
for such finesse. For her, the only way is to find the right trade, make a bold bet, take a lot of risk and strategically
pile on it as much capital as possible. Bonny told me that a few times each year she would be aware that the size of
her position, if a stop-loss level was hit, would cause her to lose her job. However, she was happy with the risk. I
once asked her why she would ever use options. She told me that she likes to use options because they eliminate the
need of a stop-loss.
Bonny was always very animated (some would say loud) and opinionated. It was a great perk of the job to be
sitting within earshot of Bonny, as you would get exposure to all sorts of market or economic theories. Note another
difference between Christine and Bonny. Christine was always asking why people traded such and such security,
and continuously tried to question her own theories, and also she was always ready to give up on a trade on the
premise that she did not have a clear understanding of market moves. She would often say that other traders must
know more about it, and therefore she could not trade that instrument. On the other hand, Bonny very often thought
that the whole market was wrong, and that is was just a matter of logic to realise prices were going to head in the
opposite direction very soon. She had the courage of her opinions.
The LTCM spread trades are the type of trades that Christine likes and that Bonny loathes. As markets evolved
and liquidity increased, at the end of the 1990s Wall Street was swarming with a new breed of trader and structurer
coming from Ivy League schools, with degrees in applied mathematics, financial mathematics or statistics. All those
quant types were eager to apply advanced analytics to calculate the price and manage the risk of bespoke structures.
Many thought that the logical evolution of the market imposed more and more competition among the banks,
increasing liquidity and, inevitably, further mathematical complexity. Hence, the proliferation of complex structures
over the life of two market cycles, first during the dot-com bubble that burst in 2000, and then during the happy
years leading to the financial crisis of 2008.
Is the party over for complex mathematical trades? Many think it is, due to the stringent capital requirements
imposed on banks with new regulations intended to address the issues that gave rise to the great financial crisis. The
main ones are Basel III and Dodd–Frank, which are trying to force capital adequacy and transparency. For example,
you may recall the 1.6% mark-up we mentioned above for a complex structure. Dodd–Frank requires, at least in
principle, that before the transaction is executed the client be given the mid-market level of the trade. This would
reveal to the client that trading the bespoke structure, instead of a simple vanilla forward, they would have to part
with 1.6% of the principal on the trade. It is difficult to imagine any FX manager being at ease with an email sitting
in their inbox spelling out that they decided to spend US$650,000 of the firm’s money because of their market view.
This is just one example of the powerful incentives of the new regulation, which is pushing everybody towards
simpler products. Having said that, political winds are already changing and the bank lobbyists are working
overtime. It is very easy to imagine a repeal of the Dodd–Frank legislation. Add to that a very low interest rate
environment that encourages leverage to get a decent return, and you can see why we make no predictions on
whether complexity will decrease or increase over the next few years.

STRUCTURES GONE MAD: TARNS

The following discussion on TARNs serves to help us describe the culture and drivers for complex derivative trades.
As technology allows us to calculate and manipulate more and more complex scenarios, human nature is sometimes
ill suited to resist the siren’s call. During historical periods with low yields, the hunt for returns leads everyone to
think of more clever ways to make one dollar feel like one hundred. TARNs are one of the most alluring and easiest
to significantly leverage, all the while leaving market participants feeling that it’s as normal as the day is long.
Dodd–Frank and regulatory requirements following massive financial catastrophes have quelled the demand.
However, regulatory ebb-and-flow cycles are gradually relaxing those rules, with the effect that banks are jumping
in headfirst once again. After all, it all turned out very nicely for them the last time these instruments caused a global
calamity. We will explore how complex trades arise, how it can sometimes work out wonderfully for everyone, and
then provide an example in which, maybe not so much.

TARNS look really good

Having given you the backstory that drives bespoke trades and explained the motivation behind mathematical
complexity, a description of our sample trade now follows. Let’s price it as of November 28, 2011. This is a real
example of a real price that was put together by a major US bank for a real client on exactly that day. The date is not
very important, but it corresponds to a period when these types of trades were in decline, and the Dodd–Frank Act
was not yet implemented. Think of the decline of those trades as the decline of the musical form of the fugue in
Germany, in the 18th century. The thing is now clearly out of fashion, and counts fewer and fewer fans, but you
cannot count it out yet as among those few fans you can find very talented musicians – Johann Sebastian Bach for
one.
Before we dive in, here is the punchline: the hedger is short US dollars. They need to buy US dollars and sell
Canadian dollars between June 2012 and year-end 2012. This can be due, for example, to revenues from a US
company’s Canadian subsidiary, or payables to a US vendor from a Canadian company. They can enter into
forwards today at C$1.0390/US$,4 but want to see if they can do better by hedging with a TARN. The deal the client
makes with their bank is to buy US dollars at a phenomenal rate way below the trade date market level. In our
example, that rate is C$0.9585/US$. However, should Canadian dollars appreciate (ie, should the exchange rate fall)
past C$0.9585/US$, they instead would potentially have to buy US dollars at 0.9785 in double the amount. Before
you get self righteous about speculative activity, it can all be rationalised. For example, “Well, if I end up with more
US dollars than I need in 2012, I’ll have expenses in 2013, so I’ll just roll it out to then” or, “I am only hedging half
of my exposure, so if I end up buying double the dollars, then it just means I will be 100% hedged”. Think of how
many of us react to a sale. It’s such a great price I should just buy extra. We will talk more about the rationalisation
of this strategy and the massive bankruptcies it caused later in this section.
Let’s return to our previous discussion about understanding the client’s view. This client would never imagine
Canadian dollars appreciating significantly beyond (which means below) 1.00 or being able to buy US dollars
around such historically low rates! Be that as it may, here are the details of the trade. Be warned this will make for
very dry reading, but bear with us until we give you all the elements you need to know. If you invest the time to
understand, the magic will appear in front of your eyes. If you just want to skip to the end, the whole story will
sound very boring.5

The structure has a schedule of fixings that starts on June 4, 2012, and runs weekly for 30 fixings until December
24, 2012.
The spot level in USDCAD at the time of pricing was 1.0350, and just for reference, the forward to December 24,
2012, was 1.0390.
At each fixing:
if the fixing is at or above 0.9585, then the client buys US$1.5 million, sells Canadian dollars, at 0.9585;
if the fixing is below 0.9585 but at or above 0.9385, then no cashflow takes place; and
if the fixing is below 0.9385, then the client buys US$3.0 million, sells Canadian dollars, at 0.9785.

Hold on, we are not done! Every time the client buys US dollars at a rate that is below market, which is the same as
saying every time the fix is above 0.9585, the client will have a gain. This gain will accumulate over the life of the
trade. For example, if the first fixing is 1.00, then there is a gain of 1.00–0.9585 = 0.0415. Suppose now the second
fixing is 1.01, then the gain will be 0.0515, and the accumulation will be 0.0415+0.0515=0.0930. And so on and so
forth. If at any time the accumulation reaches 0.35, then the structure terminates and there are no more cashflows in
the future.
Not done yet! Right after fixings 10 and 20, if the structure has not capped yet, meaning if the 0.35 threshold has
never been reached, then the accumulation is reset to zero. This structure is zero cost, in other words no premium
changes hands at trade date!
You can skip the rest of this chapter if your head is spinning. If you decide you want to keep reading, the best
thing would be to re-read the description of the trade a couple of times to think through what could happen in a
couple of examples. By now, you get the essence of complex structures. One can add all kinds of bells and whistles
so that any crazy or sane idea about what will happen in the market or the kind of risk the client wants to take or sell
can be accommodated. In the following paragraphs, we will analyse this structure and try to understand its
attractiveness and its disadvantages.
Here is the final outcome of the story. Had the client executed this structure, all the fixings in the schedule would
have ended up higher than 0.9585, with the structure being capped out and the client gaining 0.35 big figures,
equivalent to C$1,500,000 × 0.35 = C$525,000. This turned out to be a good structure in that particular moment in
time. By the way, typically, the way this works is the client would transact in the spot market for the US dollars they
need and the bank would write them a cheque for the accumulated profit, or they would write the bank a cheque for
the loss between 0.9585 and the fixing rate below there. This is known as cash settlement, but in the next example it
will actually be a physical settlement. They will exchange US dollars for Mexican pesos – get excited!
 Figure 14.1 shows that for the entire duration of the schedule of the TARN structure, the spot rate never reached
the 0.9585 barrier. This means that this structure would have been a winning trade for the client. Table 14.1 shows
the fixings included in the trade’s schedule.

Early termination of the hedge. Have a look at the “cumulative gain” column. It reaches 0.35, actually 0.3991, on
July 6. At that point, the structure would have terminated. With spot and all the forward rates for all the dates in
the schedule much higher than 0.9585, the structure seems very attractive. However, what happens when the
trade reached its cap? The hedge is over.
Do it again. The client can just trade again a similar structure. As long as the market keeps behaving as it did, the
client will keep outperforming the spot level on every hedging date. However, there is one downside to this
thinking – the exchange rate went to 1.10 at the beginning of 2014. Repeating the same structure with the new
market levels as of 2014, and assuming comparable market parameters such as the implied volatility, the
corresponding hedging rate for the new structure could have been 1.0235, much higher than 0.9585 but still
0.0765 better than 2014 spot levels. This is the magic of the trade – the ability of the client to benchmark to the
forward rate (or the spot rate) at the moment of trading, thus comparing to spot of 1.0350, and to benchmark
against the future spot (ie, 1.10) in case the structure caps out and needs to be replaced. This is a little bit of a
flaw in our perception, and again works because of cognitive biases in our brain (see Kahneman, Tversky and
Ariely for more details). A more logical argument would be to consider that in case of re-pricing of the trade
with 1.0235 being the new strike rate, then there is a loss of 0.0650 – from a more benign 0.9585 to 1.0235.
Here is the deep attraction to this trade. Although the bank made a lot of money in creating and marketing the
structure, the client also had a huge advantage. As we have pointed out, this was a client who had traded those
structures repeatedly, and over the years continued to outperform the market. How was that possible? Did a
humble corporate treasury know USDCAD better than all the wolves of Wall Street combined? We have three
answers to this question.
Yes, the humble corporate treasury knew better, or got lucky. USDCAD never ever traded at 0.9385 or below,
so maybe this treasurer played the odds and won. Once you win big on the first trade, it gets easier to do
another. If the second trade wins, now there’s a winning streak going and lots of incentive to keep trying
these structures.
This is the second hedging strategy we examined that actually gets the hedger short vega. While hedging the
client is selling volatility! The exotic option trader at the bank was happy to create an exotic structure where
the client sells USDCAD volatility to the exotic options desk. At that time, it so happened that the exotic
 option trader knew that their vanilla option trader, and colleague, wanted to buy volatility. He tried to take
advantage of that axe. In essence, the corporate treasury was exploiting a market need/imbalance.
We can’t prove this, but we’d wager that speculative interests that put on a similar trade probably overtraded
and did not wait until expiration. Those hedge funds speculating in the same market tended to overlay too
many trades, some of which made money but most that lost money. Could hedge funds have identified and
exploited this imbalance faster and more effectively? The fact is that selling volatility is not easy. It’s a
better trade in the long run, but for sure in the short run there can be a great deal of P&L volatility. This is
because selling is usually done by selling vanilla options, and each of these that terminates ITM will cause a
leveraged loss to the hedge fund.
On the other hand, the corporation has a real need to buy US dollars and sell Canadian dollars, so any technical loss
is easily incorporated in the effective exchange rate. This is another reason corporations are not driven to
maximise FX profits.

Here is a quick summary of how a corporate treasury has a competitive advantage in its ability to spin any scenario
in a positive light. We already saw what actually happened during the fixing period to our specific trade, but what
else could have happened?
Suppose that spot goes to 0.94. At that rate, the structure mandates no cashflows. This is again good for the client,
because then they can just hedge by purchasing US dollars versus Canadian dollars at 0.94 in the spot market or by
trading a corresponding forward. Is this good or bad news? Well, by definition this is much better than 1.0350,
which was the spot reference at the time of trading, so the client should be very happy that the market moved their
way.
What happens if spot goes below 0.9385? Well, in this case, the client is buying dollars at 0.9785, which again is
much better than the spot reference of 1.0350 at the time of trading. They have the option of spinning the level
relative to where the original spot was, once again. So they should be happy. It is true that they will now be buying
double amount, three million versus one and a half, but this structure works very well if the client is actually
hedging an exposure of three million per week or more. In that case, they can claim a hedging ratio of 50% for those
weeks when they buy 1.5 million, and a hedging ratio of 100% for those weeks when they are forced to buy three
million, as mentioned above. Once again we have some sort of magical arbitrage that allows the client to justify with
a positive note what is happening. If they buy US$1.5 million, they claim they were hedging half; if they buy three
million, they claim they were hedging all of the exposure. Keep in mind ongoing exposures are not as constant or
timely as we describe in our example, nor does corporate policy require that 100% of exposure be hedged. Usually
there is wide leeway for the treasury department to make that determination.
If the spot rate falls to, say, 0.91, then they would be buying dollars at 0.9785, which is 0.0685 higher than the
spot rate. Crucially, in that case there is no cap on losses, which means that they could be doing that up to 30 times
for a total loss versus spot of 30 × 0.0685 = 2.055 corresponding to US$3 million × 2.055 = C$6,165,000. This is
just a crazy scenario where USDCAD plummets to 0.91, but it serves to illustrate the asymmetry that will allow the
trade to be zero cost. In fact, USDCAD never went to 0.91, at least not since the birth of the FX market, and never
after this trade was priced, so this fear turned out to be unfounded. However, the computer that was pricing this trade
did not know that USDCAD never went to 0.91, and was certainly very far from being able to predict the future. The
computer thought it was possible. There was a definite scenario where USDCAD reaches 0.91. In this scenario, the
bank stood to make a lot of money, much more that any scenario where the client made money. This asymmetry
allowed the computer to declare that the trade was an even exchange, that there was no cost to the hedger.
Here’s the crazy part. The exchange of a very likely small gain for the client with a very unlikely catastrophic loss
for the client was an even exchange. This again is a little piece of magic where Dan Kahneman is looking over the
shoulders of the structurer designing the trades, and is holding the invisible strings that make all the numbers dance
together on a huge Excel spreadsheet. At the end, the client likes the trade and thinks it will make money, while the
computer likes the trade because it thinks the trade’s cost of hedging is negative. In other words, hedging the trade
will generate money for the bank. How much money? Well, we said that already: about 1.6%. The ability of a good
structurer is to make the numbers dance until a wide gap opens between what the computer thinks the trade is worth
and what the client thinks the trade is worth. The name of that gap is markup.

Not all stories have a happy ending

The above example worked out wonderfully for all concerned. However, by the end the risk most likely ended up
with a professional risk shop and they knew what they were getting into. Now we will tell a tale of a situation that
did not end up so well.
After the North American Free Trade Agreement (NAFTA) was enacted, many US companies, eager to take
advantage of cheaper Mexican wages and duty-free and tariff-free importing of raw materials in special economic
zones, built thousands of factories at or near the US–Mexican border. They were called maquiladoras. They caused
massive Mexican migration from the south of Mexico to the north by poor southerners that wanted a job at these
factories. By 2004, maquiladoras accounted for more than 50% of US–Mexico trade. As an aside, in the early to
mid-2000s, I had occasion to be near Tijuana many times visiting the homes of some of those workers. After
corporations discovered that Asian workers were willing to work for far less than Mexicans, they abandoned these
factories and went to Malaysia, India, Pakistan and, particularly, to China’s special economic areas. This left many
thousands of these migrant workers stuck in the hilly dust bowl that is the US–Mexican border, impoverished and
trapped. Entire families living in deplorable conditions in old trailers and cardboard boxes, without electricity or
sanitation, and no resources for them to even return to their previous life in the south.
Many of those US companies discovered that for the first time they had to worry about a Mexican peso
appreciation. They needed pesos to pay their Mexican employees. The Mexican economy was looking good versus
the US, and the peso was expected to strengthen by many economists. The general macro consensus from banks was
that steady foreign direct investment (FDI) flows, a widening current account deficit, record energy prices for oil
exports and a weaker dollar may lend momentum for a stronger peso. Figure 14.2 shows the headlines from summer
2008.
 This is the story of one such US company that was doing great in this milieu. It decided that it needed to plan to
get ahead of any peso appreciation, so called its bank to see where it could buy pesos in exchange for US dollars
well into the future. The bank recommended a TARN. By August 2008, USDMXN was hitting recent year lows,
meaning that the Mexican peso was hitting highs (remember the quotation convention). The concern that it would
get a lot stronger intensified. So the company pulled the trigger with spot at 9.8580 on August 4, 2008.
The TARN we have chosen to highlight here has slightly different bells and whistles from the one above so that
we can show you the infinite versatility that structured derivatives have to create something that will absolutely
appeal to any market view of the future. We will go through this structure quicker than the previous trade, since by
now you are all probably well-versed in basic TARN construction.
Sample term conditions offered to a client to buy pesos are shown below.
Trade summary:

the client will sell US dollars against Mexican pesos on a weekly basis for 26 weeks for a notional of US$1 million
at 10.30 (strike);
the strategy is terminated early if the accumulated profit (calculated as described below) is equal to a predefined
target profit; the trade is knocked-out for all subsequent months;
the strategy is extended for an additional 26 weeks if, on a predetermined date, USDMXN is above the extension
barrier (calculated as described below); and
this extended fixing schedule will still be subject to the same early termination condition if the accumulated profit is
equal to a predefined target profit.

1. Trade date: August 4, 2008; extension date: January 30, 2009.

2. Fixing dates: August 8, 2008, August 15, 2008, August 22, 2008, … January 30, 2009.
3. Extension dates: January 31, 2009… July 31, 2009.
4. Spot rate: 9.8580 Ps per US$.
5. Strike: 10.30 Ps per US$.
6. Reference notional: US$1,000,000 US$/week.
7. The reference notional is the same if the USDMXN weekly fixing is above or below 10.30.
8. Weekly fixing rate: the official fix published at or around 11:00am New York time.
9. Weekly value: Max [Strike – Fixing, 0].
10. Cumulative value (CV): the CV at the weekly fixing date is defined as the sum of the weekly values up to
 and including that week.
11. TARN KO event: a TARN KO event is deemed to have occurred on the first week such that: CV ≥ 6.00 Ps
per US$.
12. Extension condition: on January 30, 2009, if the fixing is above 10.30 and has not already triggered the
TARN KO event, client will be extended into the same trade for an additional 26 weekly fixings.
13. Note that if the trade is extended, whatever CV has been accumulated up to that point will carry over into the
new weekly fixings.
14. On every weekly expiry date, one of the following three scenarios will determine the payments made
associated with that fixing:
a. if a TARN KO event has not occurred on this weekly expiry date or any previous expiry date, the client
will sell the amount of US dollars equal to the notional per month, against Mexican pesos at the strike
price, on the delivery date for that month;
b. on the weekly expiry date that a TARN KO event first occurs, an adjusted strike is calculated in such a
way that the CV at that month using this adjusted strike is exactly 6.00 – then the client will sell the
amount of US dollars equal to the notional per month, against Mexican pesos, at the conversion rate
equal to the adjusted strike, on the delivery date for that month; and
c. on the weekly expiry date that a TARN KO event first occurs, the trade is knocked-out for all subsequent
months.
15. Initial premium: zero upfront premium.

Got that? Wow! The differences between this structure and the Canadian one are numerous. However, the one we
want you to note is that instead of selling some greater amount above the strike, the client is obligated to keep
buying pesos for another 26 weeks! That kind of commitment at 10.30 is worth a lot to the bank. The forward rate
out to July 31, 2009 a year before was 10.40. Guess what? The first item that left a lasting impression on the client
was also the last item: “zero upfront premium”. The second item to impress the client was that spot is 9.8580, the
forward out to January 30, 2009 is 10.12 and the bank will let you buy pesos at 10.30! What can possibly go wrong?
 Look at Table 14.2. In the beginning, this strategy looked fine. But, alas, as September rolled around and Lehman
roiled the world, the US dollar was picking up steam. In October, the world changed and not in their favour. The
unthinkable happened: USDMXN skyrocketed. These are the first few fixings.
Now this company had to buy pesos at 10.30 while the spot rate was considerably cheaper (see Figure 14.3). It
can soothe its concerns with the thought that it still needed pesos and, in fact, 10.30 is still better than what it could
have achieved in August. However, it has a commitment it may not have fully thought through. If in January 2009
the level is above 10.30, it has to keep buying pesos at 10.30 for another half a year! The reality is that buying pesos
at 10.30 is now problematic – especially when you consider that its business is upended by the economic crisis
engulfing the world, and particularly emerging markets such as Mexico. The mark-to-market loss from the first
negative observation on September 5, 2008, until July 31, 2009, on US$1 million per week is almost US$11 million
dollars! It tried to buy the structure back in mid-October 2008 but, as we said earlier, buying TARN makes one short
volatility. As volatility rises, the structure becomes prohibitively expensive. Volatility for six-month USDMXN on
August 4, 2008 was 6.6%, but on October 17, 2008, it was 30%! The company waited until the extension date,
hoping for some sanity to return to the markets, but to no avail. The forward rate on January 30, 2009 out to July 31,
2009 was 14.0807! As we can see from this small example, the magnitude of losses become exponential and there is
no fire exit! Unfortunately the company, along with many others, never survived this hedge. The hedge sank the
company.

TARNs made the whole world cry

This story played out across the entire world, and especially in emerging market currencies. Billions upon billions
were lost by non-financial companies. There is a plethora of literature written since 2008 that documents and
analyses the havoc TARNs and their close cousin, the knock-in and knock-out (KIKO), caused to financial and non-
financial institutions, here, we refer the reader to the IMF working paper by Randall Dodd.6 Using this research, we
can highlight some of the damage done as well as critique how these instruments were used and abused.
According to Dodd, and quoting the Bloomberg wire from April 2009, “The direct cost to non-financial firms of
these derivatives losses, based on the sum of national estimates, is US$530 billion. Possibly 50,000 firms in at least
12 economies have suffered derivatives losses.”
To give a more palpable sense, below are just a few examples taken from Dodd’s paper:

571 Korean small and medium-sized enterprises lost an estimated aggregate US$2 billion;
in Indonesia, roughly 10% percent of exporters were involved, and they lost at least US$3 billion;
Sri Lanka’s publicly owned Ceylon Petroleum Company lost US$0.6 billion;
China’s Citic Pacific suffered US$2.4 billion in losses;
in Malaysia, PCCS Group Berhad had large losses, and Japan’s major food importer, Saizeriya, had multi-billion
dollar damages;
India’s Axis Bank is being sued by its customers, who lost over US$3 billion on foreign currency derivatives;
exporters in Brazil and Mexico experienced even larger losses, with the Brazilian authorities estimating that its non-
financial sector lost US$28 billion;
Sadia, Brazil’s second largest food processor, lost US$417 million in closing out its derivatives positions;
Aracruz Papel e Celulose SA, paper and pulp, lost US$2.3 billion;
Banco Itau Holding Financeira SA executive director Sergio Werlang estimated US$24 billion in derivative losses;
Controladora Comercial Mexicana SAB, a major supermarket, filed for bankruptcy on October 9, 2008, after losses
estimated at US$2.2 billion;
Banorte, Mexico’s fifth largest bank, announced it would set aside Ps1.1 billion (US$74.5 million at the time) to
provision against crisis losses to Controladora Comercial Mexicana;
Gruma, a tortilla and corn flour producer, lost US$684 million as of October 2008 (the firm owns 9% of Banorte);
maturities of exposures extend to 2010 and 2011;
Cemex, the world’s third-largest cement maker, lost US$711 million on currency and equity derivatives; and
Poland had a similar experience, with losses estimated at US$5 billion.

With losses being so huge and widespread, everyone that was touched by these transactions was affected. Banks in
Korea and other countries went bankrupt, and litigation and recrimination was global.7 Entire economies were
impacted. As the losses accrued and it became clear that they could not be understood, let alone contained, everyone
rushed for the exits from emerging market currencies, forcing even greater depreciation. Dodd (2009) claims that
these instruments are simply not appropriate for hedging as they do not match the existing risks of non-financial
institutions. Further, he says “the instruments are not appropriate because the firms are not capable of absorbing the
potential losses that can occur from a ‘geared’ or double-size notional principal used to calculate losses on the
downside”. Finally, he points out that should these firms want to speculate, and such instruments are not good at
speculation.
We do not believe the discussion of appropriateness can be contained so crisply. Starting with the last point,
speculation in the FX markets is not constrained to the movement of the spot and forward markets. As we learned in
previous parts of this book, risk in FX is multidimensional and one can engage in any combination of its basic
components. For example, if a non-financial institution wanted to leverage their natural exposure to go short
volatility and short spot, a TARN may in fact be a great way to do that. With regard to Dodd’s other observations,
maximum losses can be contained easily by the type of characteristics of the TARN. The problem became severe
when companies being helped by their bankers started to turn tentative exposure into concrete risk so that they could
monetise it upfront. In one case I recall, a midwest manufacturing firm was sure USDCAD would not appreciate
past 1.10. When it did, the treasurer was desperate to find ways to keep the charade going. He came back to the
bank, saying he wanted to hedge out the company revenue for the next two years, and then three years, and
eventually he hedged away five years of future revenue projections! They sold volatility and took a spot position on
fiction. They could not realistically project revenues out to 12 months, let alone five years. They started making
long-term plans and the bank, which coincidentally had lent them money, was happy to take these plans and agree
that they had a good idea of what five-year revenue would be, mostly because the FX area made hundreds of
thousand on every transaction.
Many companies cried foul, and insisted the banks took advantage of them by not disclosing the true risk inherent
in these instruments. The truth is everyone knew what they were getting into but the times were so heady that no one
imagined what the downside could truly be. We close this section the way Dodd did, by quoting the lobbyist in the
film Thank You For Smoking: “We sell cigarettes. And they’re cool and available and ‘addictive’. The job is almost
done for us!”

MARKET SITUATIONS GONE MAD: THE CASE OF EUROS VERSUS SWISS FRANCS
Christine would like TARNS, as they appeal to the relative value hemisphere of the brain. TARNS are complex and
super tough to price and risk manage. Risk sneaks up on you, and once you’re in trouble you sink fast. However,
crazy situations can arise from conditions that are very straightforward, where the risks are known and everyone
understands the game they are playing. This is the case we will discuss now, one that would appeal to Bonny.
Consider the predicament of FX spot market-maker traders. How do traders project their short-term view of the
exchange rate on the two-way prices they make? If the trader thinks the rate will rise, then their quote will be very
high, both bid and ask, so that the client who wants to buy will not buy at such a high ask, or will buy from another
bank. The risk is that if this trader is making a high price to a client who wants to sell, they may end up being hit on
their bid and have to buy. Then again if the trader is right, that the price will rise, they will still make money when
the rate rises. In a case when the trader bought from a client and the rate falls, then the trader will wait for the rate to
come back and hopefully rise again. Let’s examine what happens when a trader is into a trade that is on the back of a
definitive point of view as to where the market is headed. However, this trader (as with all traders) is under risk-loss
constraints that limit the potential losses they can incur. Two factors will be important in such a situation.
First, the common and appropriate thing to do for the trader is to set a stop-loss level. In the trader’s mind, if the
rate goes as low as the stop-loss level, then there is no more hope for the rate to turn around and rise, and so they
will finally stop their loss and sell, thus closing out their position at a loss and moving onto the next trade. The
ability to recognise one’s mistakes, take the loss and move on is alien to human psychology. Traders become
attached to their position, and hang onto the false hope that a miracle will cause the rate to turn around and start
climbing. It is considered a key trait of a good trader to have the ability to stop-loss early to avoid increasing losses,
and the ability to decide on a stop-loss level and not change it when the market hits that level.8
The second factor in the management of a losing position, with the trader long and an asset that is falling in price,
is the risk limit the bank has given the trader. If the risk limit level is US$1 million, then the trader is not allowed to
have a P&L in the trading book that accumulates a loss greater than US$1 million at any time. If this happens, then
the risk manager will force the trader to close their positions at a loss. The trader might still foolishly cling to the
hope of a rebound in the asset price, however unlikely. But the risk manager will overrule sentiment and force the
trader to close the position.

Easy money
One example of this situation happened to many hedge funds that traded on credit with banks via prime broker
arrangements at the beginning of 2015. The story goes back a few years to 2011. The Swiss National Bank (SNB),
the central bank for Switzerland, is one of the most effective central banks at coordinating national policy to benefit
Swiss companies. In 2011, the world was still reeling from the 2008 great financial collapse. Global wealth was
looking for a safe haven. Whether it’s WWII or any other calamity, the Swiss keep themselves at enough of a
distance to be the place where the rich feel safe to invest. That was the situation in 2011, when money rushed into
the Swiss franc. At that point, the franc appreciated significantly.
The most relevant counterpart to the Swiss franc is the euro. The way the currency is quoted, as the y-axis goes
lower the Swiss franc is appreciating. Therefore, from Figure 14.4 one can see the franc appreciating from the
Sfr1.55/€ to below Sfr1.10/€ by 2011 (again, due to market convention, when the exchange rate falls the Swiss franc
strengthens). As the appreciation accelerated in the summer of 2011, the SNB was very concerned that the Swiss
export sector would be hurt. Swiss exports account for almost 70% of the Swiss economic activity. They initially
increased the supply of Swiss francs in the market to meet the increased demand. This means they sold francs and
bought euros and dollars. However, if a central bank creates more of their currency they increase the chance that at
some point in the future they will create a very inflationary environment. In September 2011, the SNB announced its
policy to sell Swiss francs and buy euros and dollars so that the exchange rate relative to the euro would remain
around 1.20. By the end of September 2011, the rate fell to 1.20 and stayed there. During this period, the SNB
accumulated massive amounts of dollars in their foreign reserve account. Traders could trade around the predictable
intervention and make money every time, guaranteed by the SNB. Nice gig!
 Specifically, by gambling on the SNB keeping the Swiss franc artificially weak, as the bank promised, hedge
funds consistently bought euros and sold Swiss francs at any rate below or close to 1.20. The idea was that a long
euro position against the Swiss franc was basically defended by the central bank. If the exchange rate threatened to
cross 1.20, the central bank would intervene. For some hedge funds, this was a one-way bet. Another way to play
this game was to sell euro puts Swiss franc calls with a strike below 1.20. By the way, these options were sold for
peanuts. EURCHF three-month volatility went from 23% in 2011 to 2.3% in November 2012 – talk about picking
up pennies on a railroad track.9 Think of our discussions in Part II of this book about implied volatility: a 2%
implied volatility basically says that there is almost perfect predictability of where this FX rate is going and at what
speed. With such an abnormal market, EURCHF options almost stopped trading altogether.
 The other element to consider is the tight range that was maintained by the flow into Swiss francs and the SNB’s
intervention. Figure 14.5 shows the EURCHF spot rate from autumn 2011 to early January 2015.

The pressure cooker blows its lid

If a trader hit the perfect high and the perfect low on €1 million, they would earn Sfr57,250. And that is if the trader
was perfect and kept the position for a year. On average, the gains from each round trip were miniscule and the
option premium earned was tiny. How does one make real money when the sure trade has tiny payouts? Leverage it
up massively! The powder keg is full and primed for an explosion. This situation could not be sustained.
The Economist summarised the reasons nicely:
The SNB suddenly dropped the cap last week for several reasons. First, many Swiss are angry that the SNB has built up such large foreign-
exchange reserves. Printing all those francs, they say, will eventually lead to hyperinflation.10

The Federal Reserve’s policy of quantitative easing (QE) was now becoming a central bank phenomenon around the
world, and the SNB would be impacted by the upcoming ECB’s decision to introduce QE.
The Economist continues:
But there is also a third reason behind the SNB’s decision. During 2014 the euro depreciated against other major currencies. As a result, the
franc (being pegged to the euro) has depreciated too: in 2014 it lost about 12% of its value against the dollar and 10% against the rupee
(though it appreciated against both currencies following the SNB’s decision). A cheaper franc boosts exports to America and India, which
together make up about 20% of Swiss exports. If the Swiss franc is not so overvalued, the SNB argues, then it has no reason to continue
trying to weaken it.

When the SNB without warning stopped defending the 1.20 floor on January 7, 2015, the EURCHF exchange rate
briefly plummeted to a low of 0.95 in a matter of seconds. According to the stop-loss rules, many banks called
margin on their hedge funds. The losses on those positions, at least on paper, had exceeded the risk limits and the
banks were therefore asking the clients to post collateral. Those hedge funds that did not have sufficient cash on
hand to post collateral saw their positions closed out at a loss of, in some cases, 30%. Figure 14.6 shows the
EURCHF exchange rate on January 15, 2015, and the 40% drop in a matter of minutes and a partial recovery by the
end of the day.
Most of those same hedge funds had stop-loss orders in place just in case the peg broke. In essence, they placed
orders with the banks to sell back the euros and buy back the francs at rates just below the peg, rates such as 1.1990.
The problem with stop-loss orders is slippage. Slippage is not a problem under normal circumstances and in liquid
currencies. Slippage in contrived situations with pressure building up over a number of years is another matter
altogether.
Let’s break this down. In a falling market, one can easily get their bid hit and buy. Suppose the market trades at
1.1990 and the bank wants to buy at that rate, but before it can do it the rate trades at 1.1985. That is no problem at
all – you will always find somebody to buy from at 1.1990 when the market is at 1.1985. The opposite is not true,
sadly. A stop-loss order at 1.1990 is triggered when the market trades there, but the next price might be 1.1985. The
bank would execute your sell order at 1.1985 for lack of bids at 1.1990. The 0.0005 extra loss is called the slippage.
As it’s clear from the figures above, there was no orderly drop where reasonable slippage could be explained.
Dramatically enough, many of those same hedge funds got their stop-loss trades executed with slippage of 0.15 or
even more. That is a loss of about 15%, which could bankrupt the fund if the position was leveraged.

Some of these funds did in fact go bankrupt due to the losses from that night. Lawsuits ensued. In court
proceedings, some of them argued that the banks should have waited until the EURCHF rate, as it was “obvious” it
would, recovered from the initial panic selling. Sound risk management does not rely on the “obvious” prediction of
a rate turning around, nor do collateral agreements. It has to be noted that the fall of the EURCHF peg (really just a
verbal promise from the SNB to the press, so not a peg in the technical sense) represented a seismic event in the
markets, one the likes of which only come around once in a decade. Additionally, central banks hate being gamed.
Whether it’s the SNB, the Fed or the Bank of Japan, central bankers love to come in and punish speculators for
using their intervention as a guaranteed way to make money.
The consequences were not constrained to the hedge funds speculating. Brokers that gave credit and leverage to
them to trade suffered as well. The consequences for the biggest players of this game were deadly:
FXCM Inc., which handled a record US$1.4 trillion of trades by individuals last quarter, said clients owe US$225 million on their accounts
after the Swiss National Bank’s decision to abandon the franc’s cap against the euro roiled markets worldwide. Global Brokers NZ Ltd said
losses from the franc’s surge are forcing it to shut down. IG Group Holdings Plc estimated an impact of as much as £30 million (US$45.5
million) and Swissquote Group Holdings SA set aside Sfr25 million (US$28.4 million).11
In the UK, retail broker Alpari Ltd said it had entered insolvency. ‘Where a client cannot cover this loss, it is passed on to us. This has forced
Alpari (UK) Limited to confirm…that it has entered into insolvency’, the firm said. Fellow UK broker IG Group Plc said it was facing a
negative impact of up to £30 million (US$45.7 million) after the ‘sudden and extreme movement’ in the franc.12

These dealers would allow massive leverage for FX trades, to the tune of 100 or 500 to one in some cases. We’ve
spoken extensively elsewhere in this book about the effects of leverage, so we will not re-analyse them here, but
suffice to say that it’s great when a levered position moves in your favour, but sheer hell when it moves against you
– recall our TARN discussion.

SUMMARY OF EXTRAORDINARY SITUATIONS

In this chapter, we’ve explored extraordinary situations that came about from apparently very different conditions.
However, there are some obvious commonalities. Volatile situations arise when there is opportunity for great profit
but the commensurate risk is not properly accounted for. In our two examples, the repeated dipping of toes in shark-
infested waters that did not result in lost toes, led smart market players to decide the waters were safe for daily
swims. The trades were incredibly appealing and our human nature convinced the participants that the unthinkable
could not happen. In the end, the market truth is that trades that allow for extraordinary, consistent profits simply
must include in their make-up the potential for unique, extraordinary calamities.

1 Take a simple example with a forward trade in EURUSD where the client is selling euros. Suppose the mid-market quote is 1.20, and the salesperson
takes one big figure in mark-up. The rate is now 1.19. At maturity this trade could be ITM if, for example, EURUSD ends up at 1.17. Now, suppose
the mark-up is bigger and the trade with the client is at 1.15; here, the 1.17 scenario would no longer work since the client would lose money on the
trade. If the mark-up brings the rate to 1.10, now to be ITM the EURUSD needs to be below 1.10. The more money taken in mark-up, the more
unlikely the client is to make money. This is a very widespread phenomenon in finance. In times of favourable markets (especially in the equity
markets), as every client is making money no matter what stocks they pick, large mark-ups are easy. Clients do not take brokers to court after they
become millionaires in the stock market. That is also why financial lawsuits tend to increase in number when the market turns.
2 M. Lewis, 2016, The Undoing Project: A Friendship That Changed Our Minds (New York: W. W. Norton & Company).
3 The counterpart in real life is the celebrated experiment ran by Kahnemann and Tversky, where people estimate that the number of words ending in
“ing” is larger than the number of words ending in “g”.
4 We took great care to be clear on how the rate is quoted because the Canadian dollar versus the US dollar is often quoted by retail outlets the opposite
way from the professional financial markets. Furthermore, when it is right around one to one, it gets real confusing and very difficult to figure out
whether the US dollar or the Canadian dollar is stronger. To be clear, we are quoting Canadian dollars that buy US$1, so if the number rises then the
US dollar becomes stronger.
5 Sophisticated pleasures are often acquired tastes. It is said that when the sultan went to Vienna for a diplomatic visit, the emperor invited him for a
night of entertainment to hear a Mozart opera. At the end of the opera, they asked the sultan what piece he had liked the best. Without hesitation, he
said it was the first piece, but when they played the overture again for him he said that no, that was not the first piece. After some embarrassing back-
and-forth, it eventually became clear that what the sultan had meant by the first piece was actually the orchestra tuning before the performance.
6 See R. Dodd, 2009, “Exotic Derivatives Losses in Emerging Markets: Questions of Suitability, Concerns for Stability”, IMF working paper WP/09
(available at http://www.financialpolicy.org/kiko.pdf).
7 In some cases, most of these transactions were marketed and sold locally by foreign global banks. Local governments were happy to put pressure on
the judicial system to skew the lawsuits in favour of the local corporations that had entered into these “hedging” transactions. The argument was that
corporate treasury did not understand the trade at the time of execution, or that the trade had been mis-represented by the bank and, therefore, the
trade should be torn up and the corporate losses pardoned.
8 Human psychological biases affecting traders is a fascinating subject of behavioural finance, which is now being studied very extensively. The initial
work that generated a lot of those groundbreaking ideas on human irrational behaviour was carried out by Kahneman, Ariely and Tversky, as referred
to earlier.
9 The options market in EURCHF at some point basically ceased to exist due to this aberration.
10 The Economist, 2015, “Why the Swiss Unpegged the Franc”, January 18 (available at https://www.economist.com/blogs/economist-
explains/2015/01/economist-explains-13).
11 https://www.forbes.com/sites/timworstall/2015/01/16/currency-brokers-fall-over-likedominoes-after-snb-decison-on-swiss-franc/#2e319c972acd
12 ibid.
 15
Speculators and Hedge Funds: How Do Portfolio
Managers Make Money?

The following discussion will explore if there is something special about those people who take risk, and are able to
generate a profit from the markets better than most of us. What makes one person an expert? Why should I pay you
to manage my money? What does it mean to have a talent for speculating or creating a market edge? We arrive at a
rather controversial conclusion.

DO SUCCESSFUL PORTFOLIO MANAGERS HAVE TALENT?

There is a collective mythology around hedge fund performance and the individuals that create wealth. The reason it
is more myth than reality is that performance is very difficult to measure consistently across time and circumstance.
Very often, great performance is a result of the frothy times the trader is living through.
How does a portfolio manager (PM) make brilliant trading decisions in the first place? Is it instinct or a set of
conscious or subconscious rules? On the surface, a PM will receive massive amounts of information and analysis
that could not be assimilated by one human being. Even deciding what is relevant often becomes overwhelming.
This is much more relevant and overwhelming in FX where information about a currency involves information
about multiple countries and the state of the global economy. When trading mid-cap stocks, on the other hand, the
breadth and depth of information are radically different. Therefore, we are left wondering: is there a mystical trait
that some have that allows them to arrive at the right decision consistently, or is it a method or edge that can be
applied independently of the individual? Maybe it is just random luck.
Fundamentally, we believe there exists an edge or an evolving set of edges – a consciously articulated set of rules
or market conditions that yield good returns over prolonged periods of time. Where these rules cannot be articulated,
it is easy to claim a hidden, divinely inspired talent. We believe it just means we haven’t been able to isolate that
edge. Edges span the spectrum of legality and morality. The most common include: inside information, yield-
producing trades that persist for a time before they snap back and destroy all in their path. Finally, the evolving use
of modelling to consume an ocean of data, before digesting it and spitting out a single number that is the ultimate
answer to life, the universe and everything.
Are edges really fair advantages? There are some edges we all can agree are immoral and should be illegal, but it
wasn’t always this way. Is it meritocratic for one person to make money at the expense of another or society because
their machine is physically closer to the exchange? What about if they put together a quantitative model that predicts
FX direction for a time? Our perception of what edge is fair has evolved over the decades, and has been greatly
affected by the global financial crisis, social movements such as Occupy Wall Street (OWS) and the very blatantly
illegal activity of some of our most vaunted institutions.

Our modern mythology about PMs

Very often, a hedge fund trader (more appropriately, a portfolio manager)1 is portrayed as a modern knight in
shining armour, someone who achieves fame and wealth by the strength of their intellect, courage and bravery.
Films such as Wall Street and The Big Short have created a distorted image, a false parody that altered reality for
millions of college graduates who decided they were interested in the industry.2 The attraction of the myth is
obvious, even though the technical details and the job specifics are very sketchy and the subject raises endless
philosophical debates. In other words, everybody wants to hang out with the PM who shows up at the party driving a
Ferrari, but few really know how to go about making that kind of money.
There are a couple of elements that make it mythical. First, it is the amount of money and the speed with which
you can accumulate wealth that is alluring. This is not your “father knows best” traditional US career. It is not the
family company that was run prudently for 30 years, until it expanded into a successful public company. PMs can
reach millionaire status within a few years of starting in the business. Second, there is the perception that it is
accessible to all. Anyone can potentially figure it out – you just need to work out how to get the job, but then
actually doing the job will be the easy part. There is a sense that this is a latent talent within us, and all one needs is
the forum.
My favourite comparison is to brain surgeons. It is clear to everybody who watched ER on television that, to
become a brain surgeon, you need to study very hard for at least 10 years. However, if you start to binge watch
financial daytime television and analyse the market views of all the pundits and all the PMs that are interviewed, the
feeling starts to mount that you too could come up with the same sound bites and three-second explanations of why
you knew for a fact the market would go up today. Not just in the equity markets: you can be Billy Ray Valentine,
played by Eddie Murphy in Trading Places, and figure out orange-juice futures. This is peculiar to finance, and
separates it from almost all other human endeavours. The feeling that the explanations and the knowledge they have
is not much more than what you have, or what you can quickly pick up. This myth has myriad candidates for our
idealising fantasies – we have created a class of aspiring portfolio managers who are attracted to the trappings of
hedge fund lore. They are smart and they want to prove it.
True critical evaluation of these success stories and rankings is not easy. We use total wealth accumulation as the
ultimate arbiter as to whether one is a brilliant speculator, but in our heart of hearts we know that is not an accurate
reflection of speculative prowess or smarts. After all, where you end up has a lot to do about where you started. In
practical terms, it is devilishly difficult to compare performance. If a brain surgeon studied and practiced, their
success ratio will likely be high. However, we can also easily observe that if they don’t kill people on the operating
table, then generally they have a high success ratio. Not so in finance, where the difference between a star PM and a
failure is a 50.2% success ratio versus a 49.8% success ratio. Let me give you another illuminating example.
Consider professional tennis, where the difference between the number one in the world and number 50 is that the
number one wins 53.4% and the number 50 wins 52.9% of second serve points, respectively. The difference is as
minuscule as the difference between the correct predictions of a star PM and yourself. But if you watch the number
one tennis player in the world, you know immediately that you cannot hit the ball as hard, you cannot run around the
court as fast and you are absolutely convinced that the professional player would wipe the court with you.
Not so in finance, where an in-depth discussion about trade ideas can be a very frustrating experience. Ask a star
PM why they bought energy stocks today and the explanation is almost never impressive. The more you listen, the
more you are convinced you could have come up with the same. Famously, somebody met George Soros at the gym
and asked him what he thought of the stock market. Soros launched into a lengthy and complicated explanation of
why it would go up. Essentially, all the indicators and analysis he examined were consistently bullish. A few months
later, the same person met Soros again and told him that, following the advice, he had invested in stocks but stocks
had only gone down. Soros laughed, and said that the next day he had changed his mind, shorted stocks across the
board and made an absolute fortune.
A couple of comments are in order here. The first is that PMs (or George Soros, in this case) have a clear interest
in obfuscating and using verbal smoke and mirrors. This is especially true when being interviewed on television, lest
they give away the goose that lays the golden egg, or sometimes they want the general investing public to come in
and take a bad position off their hands. Second, a PM may have the same information, but the PM will claim a talent
advantage. This means that their decisions, relying on exactly the same information that is available to you, will be
better based on market experience or pure talent. What this means is not so clear. The rest of this section will be
dedicated to figuring out what part is talent and what part is luck.
The corollary to the first myth is the idea that the hedge fund world is a forum for pure meritocracy, and that this
meritocracy just lifts the best and the brightest. For example, one of the reasons many hedge fund traders move from
the sell-side to a hedge fund is that, as a trader for a hedge fund, one doesn’t need to be diplomatic or politically
correct in their dealing with fellow workers (or clients for that matter). One can sell, raise money and receive a great
bonus based solely on their annualised returns. The returns will speak for the individual.3 The accumulated returns
year after year will be the proof of pedigree. No need to manage a team or be voted into office, one can just work by
oneself in perfect solitude and one’s worth will be revealed by performance. For all the Americans who feel their
potential in life was unfulfilled because they have not been able to play the political game to the advantage of their
corporate careers, this is a new and refreshing gospel. This is their chance to prove their worth. This has a big retail
application – even if you do not get hired by a hedge fund, anybody can open an electronic trading account and start
their trading career with a few hundred dollars, and then there is no politics: either you make money or you don’t.
This meritocratic dream thrived, especially during the dot.com bubble at the end of the last century. Thousand of
people left their jobs to become day traders thanks to the incredibly cheap execution platforms that were heralded by
the new Internet era. Despite what all of them thought, it was not their magic touch that made money. Any trade
would make money in such a frothy market. Reality came back to bite them with a vengeance with the market crash
of 2000 and 2001. The new obsession with cryptocurrencies has a similar feel. Although we are convinced the
cryptocurrency phenomenon will evolve in a way that will transform our world as the dot.com era did, the truth is no
one really knows the dimensions it will take. This current crypto-based attempt to break away from the daily grind
of the machine, feels even more extreme than the dot.com craze, and it is being experienced by a broader swathe of
the global society. Millions of teenagers all around the world are trading in hundreds of thousands of dollars in
cryptocurrencies. Coinbase has more users in certain months than Charles Schwab has accounts.

HOw DO PORTFOLIO MANAGERS MAKE MONEY?

Let’s start simply. How is the PM supposed to know that the euro will appreciate today against the US dollar but fall
against the yen? They can access massive amounts of data, analytics and opinions, gladly offered up by the sell-side
and their analysts. They can superimpose this information on their experience and gut feel. Is that all they need to be
successful?
First, let’s examine this issue from a philosophical perspective. Many PMs, when asked, will say they have
experience in the markets and their experience tells them. Some will say they have intuition and their intuition tells
them. Many, especially short-term FX traders, claim that it is not possible to design any rules to predict the market,
as the market is unpredictable. This last claim evokes most palpably what is often embraced in films and books –
knowledge of the next market move comes to PMs just as inspiration comes to artists, or just as Jedi Knights are
supposed to “feel the force” and be able to fight while blindfolded. Let us try to clarify this paradox a little. The
British philosopher Gilbert Ryle postulated two kinds of knowledge. His most famous example is about riding a
bike. We know that a bike is composed of many parts, including pedals and wheels. We also know about the
distribution of weights, centre of gravity and momentum from our physics studies. This type of knowledge is called
“knowing that” by Ryles. Does it help us in learning how to ride a bike? Not one bit.
So how do kids learn how to ride a bike? They try, they tinker and experiment, they fall, and finally they learn.
This second type of knowledge is called “knowing how”. Any of us would have a very hard time explaining how we
balance on a bike, but most of us can easily ride a bike. This second type of knowledge is invoked by PMs when
they describe the way they trade, the way they “feel the market”, the way they try to portray their knowledge as
mystical, as a gift that cannot be shared. This way of thinking just “feels right” and “makes sense”, and
coincidentally is a great marketing tool to justify high total compensation for PMs. Despite PMs not being able to
explain their knowledge, just as we are not able to explain how we are able to ride a bike, there is no mystery in
riding a bike – and we do it following a very simple algorithm, even if we will never be able to describe it. The same
is true of PMs: “knowing how” is still knowledge gained through an algorithm. The unprecedented advances in AI,
and the refinement of the application of different AI techniques to different problems, leaves little doubt that we will
be able to describe any algorithm that just feels magical when humans do it. As far as figuring out the algorithm of
how to ride a bike, google “robotics and Boston Dynamics” and you will be convinced.
Lest you think that discovery is restricted to the realm where laws of physics are prevalent but human judgement
is absent, consider medical diagnosis, especially in the detection of cancer via the visual examination of scans. It
used to be thought that only humans with lots of experience could successfully detect cancer, but recent advances in
neural network computing have changed this preconceived notion. Funnily enough, a successful neural network
computer is also unable to explain what rules it follows to detect cancer; nonetheless, very few would argue that
such a computer is not using an algorithm. We have to mention here the work of Lew Goldman in the 1960s, and
what is generally known as the “expert problem”. To summarise, Goldman examined the decision-making process
of highly specialised professionals such as radiologists. It was believed at that time, and still is by most people to
this day, that professionals use a highly complex logic to make medical decisions. That this logic is not replicable by
a simple algorithm, and that many years of experience are needed to reach comparable performance levels. Goldberg
debunked all that, showing that professionals’ decision-making processes could easily be captured by a simple
algorithm, as long as the algorithm is fed data about the professional’s past decisions and their input data. Goldberg
also showed that professionals can easily render a different judgement when fed with the same data, which means
their decision-making is not very precise or reliable. Finally, he proved that the performance, or the quality of
decisions, of experts with years of experience could be matched by experts with no experience.
This is very sobering news for those who justify their compensation by being experts, especially those who are
interviewed on television and pontificate in sound bites. In finance, the picture is even more disheartening than the
one Goldberg paints. For a deeper dive into the “expert problem”, we refer the reader to the works of Nassim Taleb,
one of the pioneers of skeptical empiricism in the financial markets. We have mentioned this before, but finance is
the only profession where a total rookie can beat the gurus. While one is absolutely sure they will be happier if a
surgeon operates on you instead of a plumber (and if a plumber fixes your sink rather than a brain surgeon), the
picture is much less certain when it comes to investing. Suppose, on any given day, you had to invest a dollar in a
stock. One can ask the plumber, a hedge fund PM or just invest in an index. Which one of the three alternatives
would you choose? There is hardly any evidence, after so many decades of quantitative academic research, that the
PM is the best choice.
To our human nature, it sounds great that there are talents that are God-given and cannot be taught; it is
convenient and opportune to claim that certain people should be paid more if they are able to do something that
other people will never be able to do; it feels great to believe one is a special individual that can “feel the market”. In
the end, it is still an algorithm that lets you ride your bike. To be fair, some people cannot ride a bike, no matter how
hard they try. Let’s assume that this is due to fear of falling and hurting themselves. They may have a lower risk
profile than others. Now we may be hitting on the essence of what is special about PMs. Maybe we can say that
successful PMs are more willing to give in to the algorithm and potentially fall, and that is what makes them special.
It may seem unfair that a PM would profit from a trade idea that had been created by their analyst, for example. In
this instance, the PM may not even possess the algorithmic gene, but still made the most money. However, the
essence of capitalism is not really knowledge, it is more about risk. All those armies of analysts on the sell-side do
not lose any money and do not lose their jobs if their trade ideas lose money, but the PM will. As a friend that used
to trade oil told me, “any idiot can make money, the genius is in not losing money”.

Algos hit FX traders very hard

In the FX markets, the discovery that most of the trading and market judgement can be done by computers has been
devastating to a huge swathe of our friends and ex-colleagues. At least one generation has made a living for the last
40 years convinced they had a special talent. There are exogenous factors that make it more difficult for FX PMs,
such as increased competition with the increasing number of hedge funds being created. Competition decreases the
returns that are chased by more and more traders. In some markets, liquidity decreased dramatically after the great
financial crisis, raising trading costs and wiping out potential profits. This made it easier to prove the old,
academically researched, claim that for all the prestige and the allure, investing in hedge funds is a losing
proposition. This is particularly true in equities when comparing investments in low-fee index funds and ETFs.
However, it all comes back to the fact that computer programs can perform the most popular classical trading
strategies better and faster, squeezing human traders out of jobs.
A company we know needed to hire an FX market expert. From a vast pool, one type of applicant stood out:
former FX spot market-maker traders. These were former traders, with 20 years’ experience on average, who had
been trading very liquid currency pairs such as EURUSD, USDJPY or GBPUSD. Their trading book had been able
to produce, on average, eight-figure revenues per year, and most of them had senior titles. They believed that their
natural talent had produced that money and that, given another similar job, they would be able to continue to
produce at those levels. They did not believe that in the current market the advantage has moved to computers, and
had an unshakable belief in a trader’s ability to “feel the market” and guess its direction, an ability that a computer
could never possess. They also did not believe that the stricter trading regulations would prevent them from ever
reproducing the type of returns they had enjoyed in past markets. Regretfully, most of them had been let go from
their respective banks as a consequence of the secular trend of shrinking trading floors, or at least had been replaced
by more junior (and cheaper) traders – to the delight of Lew Goldberg, in what was called the “juniorisation of Wall
Street”.
The main problem when interviewing those former traders was that they had been slaving away for 20 years
staring at two numbers, the bid and the offer. For 20 years, they had done nothing other than watch the screen. When
asked how they would decide whether to sell or buy, they were not able to explain why, or to give any rules. In their
opinion, it was all very “perceived” or “felt”. They did not deeply appreciate that their bank’s large presence in the
market gave them the advantage of seeing all sorts of trade flows that their competitors and even the hedge funds
could not see. Was there any fundamental economic reasoning at play? No, they did not really follow central bank
policy. Was there some algo at play? No, none of them knew how to code. Were they snatching information from
the options market? No, they did not know options theory. Did they have experience in selling, marketing or
pitching products? No, they rarely met clients. Did they like to speak in public? Not really, most were just happy to
hide behind a couple of computer screens.
As the interviews progressed, it became apparent it would have been very difficult to make use of the skills that
those former traders have. It is extremely hard to prove that they were able to outperform on their trading job
irrespective of randomness and irrespective of the clear advantage that trading at a bank as a market-maker confers.
This makes it problematic to convince anybody that, on hiring them today as traders, they would again be
consistently able to produce eight-figure revenues. With hindsight, those coveted seats on trading floors look more
and more to have been like a sort of lottery. If you were lucky enough to be given the EURUSD trading book to
manage, you were probably going to make eight-figure revenues out of it.
When it all comes down to whether some people have a special gift to make money trading or this is it just
another dream, we suspect it is the latter. However, a lot of myths are very difficult to disprove. Alice Stewart’s
crusade to raise awareness of X-ray dangers took her many years, and resulted in her being ostracised by the medical
community. However, today it seems to us that the issue is very clear-cut, and that X-rays are indeed dangerous, but
at that time in the 1950s, the idea just did not sound right. We collectively love a good story, we love to see a
winner, we all grew up with this idea of knights in shining armour.

Do PMs process the deluge of information they receive?

One would expect that the person who will decide if the firm is buying or selling the euro will be the same person
who will process all market information, but that would be wrong. As a matter of routine, the portfolio manager will
receive and read reports, emails and presentations that the sell-side and internal research team put together. This
material usually consists of trade ideas. Analysts or salespeople compete for PM mindspace to show how to think
about economic data, transaction flows, political machinations and commonsense to achieve a particular conclusion.
All banks compete to provide their buy-side clients with the best trade ideas, and therefore the portfolio manager is
by definition flooded with emails and PDF attachments. For decades, the largest FX market players did not just
compete on price, but leveraged the fact they had a large market share and could transform these flows into the best
trade ideas.
Some of these banks employed journalists to regularly send out on their own password-protected Internet website
short “market colour” explanations. The current regulatory environment and a few billions in fines have put the lid
on this practice. The regulations are such that banks are forbidden from giving their clients any piece of information
that is generated as a consequence of their trading flows. Therefore, the famous sentence: “hedge funds are buying
the euro”, or worse: “a west coast pension fund is selling the yen”, have become very problematic. This puts the cap
on the paranormal abilities of execution traders in terms of decoding market moves based on “market colour”
provided by sell-side coverage. What is left is the deluge of emails and attachments containing trade ideas based on
numerical research. And here the portfolio manager comes into the picture, not as the person who produces the
information, nor as the creator of market intelligence, but as the person who provides the decision about which one
of the scores of trade ideas deserves to be executed.
A few years ago, I worked at one of the largest banks in the world, with a very large market share in FX. This
bank maintained a full global team that was in charge of research and “thought leadership”. Team members prided
themselves on their PhDs and their academic output. They regularly referred to themselves as the “pre-eminent
research outfit on the Street”. Their job was to engage the buy-side, mostly PMs and other senior executives, and
present them with the deeply insightful results of their market insight, trading schemes and understanding of global
currency flows. The hope was that the clients would use the knowledge thus gleaned to make money and reward the
bank with business, ideally with some insensitivity to price. The global head of the team – let us call her Jeja – was a
very smart lady who spent her time flying first class from continent to continent, visiting the best of the best clients.
Typically, she was paraded around by an entourage of adoring salespeople.
One of my reports, let us call her Anna, had recently been transferred from New York to Singapore and, being
still relatively junior, had been tasked to coordinate Jeja’s visits and scheduling. During a client tour in Australia on
the way to a prime customer, Anna was diligently carrying all the PowerPoint printouts for the meeting, when she
suddenly stopped and said: “Jeja, I have bad news – for some reason, all these printouts are exactly the presentation
we used when you visited last time.” Jeja looked at the printouts and shrugged, saying: “Yes, that’s fine.” Anna
continued, her eyes full of terror: “But, Jeja, you don’t understand, this is the same client I took you to visit three
months ago!” Again, Jeja, imperturbable, said: “Yes, that is fine.” It turned out it really was fine – the meeting was
conducted in keeping with the highest standards of professionalism, and the client did not reveal in any way whether
or not they had noticed the material and the presentation deck were the same as on the previous visit. Unluckily for
Jeja however, Anna travelled to London for an internal meeting and related the story to Jeja’s boss, who was not a
very patient man.4
So, we are left to ponder, especially in FX markets, how much of the posturing around great ideas being
important, and consumed and digested by PMs, is actually valuable? Jeja was probably being paid a seven-figure
salary. Does that bank or any sell-side really feel that it is getting its money’s worth from this type of an approach?
The fact is the new MiFID rules on research will test these relationships like never before by forcing banks to
separately charge their clients for research. How many of those will accept to pay?
PMs gather all the information about their market of interest before making an investment decision. The question
we’ve been struggling with is: does the PM take the investing decision from a set of rules inside their head based on
randomness or based on instinct? Let’s agree that instinct is that innate ability to make the right decision based on
intangibles. To us, this just means randomness. If there are no rules, then the decision is random. Instinct is often
used as a code word designed to convince that the ability to trade is in their DNA but not in yours, and it cannot be
taught. We are not claiming that after examining every PM we will discover that their decisions are either
completely based on randomness or all come from a rule-based methodology. We are claiming that the sum of their
random decisions plus their decisions based on rules add up to 100% of their decisions. They might have screened
the market and isolated two stocks based on a few rules, and then invested in only one of those two stocks based on
a coin toss. So this is a combination of rules and randomness. We are
claiming that all their decisions come from combinations of rules and random draws, and there is nothing else.
Words like “talent”, “feeling”, “inspiration”, “feel the force” have no meaning.
The problematic issue is that individual savers, who could become individual investors, decide to delegate
investing decisions to hedge funds in the belief that talent and instinct exist. The academic community has been
trying for decades to study the ability for hedge funds to outperform the market. In aggregate, they did not find
superior talent, or instinct, or superior performance.5 The words of Lorenzo Da Ponte, Mozart’s librettist, come to
mind, from Così fan tutte:
E’ la fede degli amanti come l’araba fenice
Che ci sia ciascun lo dice
Dove sia nessun lo sa

Loosely translated for the benefit of those of us who are not opera lovers (or are not Italian, or are not lovers): “The
faith of lovers is like the phoenix. Everybody says it exists, but nobody knows where it is.” In the same fashion,
everybody is convinced that raw natural financial talent exists, but nobody knows where it is. By the way, Da Ponte
lifted those verses from Pietro Metastasio.6

TO MAKE MONEY, GET AN EDGE

The bottom line is: anything that is not algorithmic is random, there is no third way. Either the market behaviour is
determined by an algorithm, no matter how complex, or it is random, or it could be a combination of the two. This
has deep implications for the very existence of PMs. Is there a sophisticated computer program inside their head –
call it their intuitions and experience – that we can claim is another way of saying “a coin toss, or a combination of
the two”? In this section, we will postulate that the algorithmic part is actually not such a big mystery. We do not
always know what it is, and it most definitely changes and evolves over time, careers and individual situations, but
it’s really just an old-fashioned edge.
Let’s explore further the idea that successful PMs make money because of one of two explanations: either they
follow some sort of rules, possibly very complicated ones, or they are lucky. By “innate talent”, we take to mean
that there are a set of rules that consciously or subconsciously are followed that give the trader an edge. To move the
discussion forward and unravel the mystery, let’s examine a few great PMs whose performance has defied
understanding, hedge fund owners that have outperformed the market for year after year. The list contains the usual
names, in equities, such as Warren Buffett, Steven A. Cohen, Alan Howard, Cliff Asness, Jim Simons and Victor
Niederhoffer, and famous FX names such as George Soros, Paul Tudor Jones, John Taylor, Louis Moore Bacon, Jim
Leitner and Phil Gellos. How can we explain their success?
Luck cannot be the answer. Nobody could be that lucky consistently over a career that spans a few decades; well,
maybe, but it has a very, very low probability. It is possible that a couple of skills can come together in one
individual. It may be that a few good trading rules that work very well for a few years can then be leveraged by an
ability to use one’s network, accumulated wealth and market sway to one’s advantage. This will replenish a stable of
trading ideas over time, allow quicker and better access to information, allow them to attract skilled people that can
get even better information and use it more efficiently. This can create a virtuous cycle that, when coupled with
managerial ability, can lead to decades of success.
Imagine a good trader who has a good run for a few years thanks to a favourable market and maybe a niche
trading idea. After those initial years, the trader has many contacts and can poll their community to gather other
good ideas, which they did not create. The trend becomes your friend, so every year there will be just a few trades
that work very well. After a few more years in the business, their trading record starts to look very solid, and the
assets under management soar. The time comes to switch skillsets and continue to make money but using a different
method. At this point, there is enough money to go around that our PM can hire traders and relinquish the role of
foot soldier, sifting information, crafting trading strategies and executing trades. Our PM now has their firm and
hires and fires traders, and because of their reputation young traders flock to them – and would work for free if they
could. Here is when managerial ability becomes the main engine of the operation. The ability to spot smart and
hungry people, the ability to fire them when their little money-making machines stop working, the ability to network
and raise assets, and the ability to be ruthless. A bit cynical, maybe, but an extremely plausible scenario.

Creating an edge with privileged information

Let’s refine the idea of what an edge is, and point out the most obvious ones. A PM may have different edges and
the cleverness to know when each of them will stop working. Such a PM would know why they are making money
and be prepared to change should a new edge be required. The core skill would then be the ability to design or find a
constant stream of edges. There are a large number of rule-based methodologies that will work at any given time,
although they regularly go out of fashion and need to be replaced. We will take a whimsical look at these, to give
you a short tasting menu and a feel of the incredibly large variety of ways to make a buck in FX in the hedge fund
industry.
In the most glaring of cases, made famous by films and criminal investigations, especially after 2008, PMs can
gain access to nonpublic information. We hope it does not come as a surprise, but this is illegal. However, it is a
very effective way to produce a positive P&L. In the simplest incarnation, you just need to receive nonpublic
information from inside a publicly traded corporation and know in advance of the market whether or not the next
earnings release will be a positive or a negative surprise. Place your trades accordingly, just as if they let you bet at
the racetrack after the race was over. Similar to this behaviour, but not illegal, is to become a politician, serve on the
relevant congressional committees, sponsor legislation favouring specific companies or industries, and trade on them
just before it becomes public and before it becomes law. We’ve seen some glaring examples of this in the current
crop of political leaders. Be warned, it is a significant investment in time and effort to become a politician at that
level, not to mention that you have to sell your soul! In FX, it is a lot more difficult to get inside information that’s a
sure thing because of the size of the market. Nevertheless, the concentration of volume in a few big banks and the
dominance at times of a few big speculators have made inside information a valuable money-maker over the years.
The new regulatory scrutiny has greatly reduced the ability for the sell-side to peddle this information and, as MifID
takes effect, it threatens to reduce it even more.
So much for obtaining specific information that is potentially illegal on specific situations. What if the job bathed
you in privileged legal information that gave the trader or the salesperson an edge? As mentioned in previous parts
of this book, the probability of consistently being profitable by trading or earning a bigger sales margin were
massively more likely at one of the five top banks in FX than if one speculated for a small mid-western bank. The
big global bank saw all the flows that mattered. One did not have to take the chance and risk falling off the bike.
You got training wheels if you were the USDDEM dealer at Deutsche Bank or, later on, the EURUSD trader at
Citibank. Clients needed to deal with the sales desk at JPMorgan because, in addition to borrowing money from
them, they had the best information on where the orders were. As the banks significantly cut back their spot dealing
desks and these dealers were forced to look for other positions, it was a very cruel reality that hit them – that, in fact,
they did not have an innate ability to read the market without the training wheels.

Invest in a sure thing

A second very simple way to generate a P&L is the carry trade. Buy an asset that has a very high yield and finance
the purchase with borrowing something that demands a lower interest. For example, in FX a long position in the
Brazilian real, or any other high-yielding currency, financed with borrowing in a currency with a low interest rate
such as the US dollar – or, even better, currencies with negative interest rates such as the euro or the yen. We have
discussed many times, particularly when talking about the risk reversal, that the risky asset in the USDBRL
relationship is the Brazilian real. Therefore, the real is subject to grinding higher but also subject to disastrous
crashes, the magnitude of which few PMs can survive.
In other words, a PM can use the carry trade as a way to consistently generate gains, with the risk that if a crash
comes their trading book will have to close and they will lose their job. This is a famous trap for any hedge fund that
is under enormous pressure to generate constant positive returns. They enter into these trades with outsized
downsides to alleviate the constant threat of asset redemption. A carry trade strategy, which is in essence the same
thing as buying high-yield bonds or buying subprime mortgage-backed securities, will be able to generate those
regular returns for a time. When entering into slightly less risky and, therefore, less profitable, sure things, many
hedge funds will lever the strategy by buying on margin or entering into structured trades that take advantage of the
rate differential. As the game continues, the carry trade makes the PM look like a hero, but there is a reason for that
– it is a game where the beginning is very easy and everybody can be a winner. The difficult part comes later, when
the market inflates. Sooner or later everybody will figure out that this particular market or position is doing very
well this year. They will all be under tremendous pressure to enter into the same position, and so the positioning in
the market will become more and more extended. Then the real crucial part will start. Everybody knows this cannot
go on forever, but they cannot leave the party lest they give up on their chance to produce those juicy returns.7 At
some completely unpredictable moment in time, enough money will leave the trade that the market will correct.
For another example of a sure way to make money, see the EURCHF discussion in the previous chapter. In some
cases, central banks can be the providers of guaranteed profits. For three years, the SNB kept EURCHF artificially
above 1.20 in the face of huge market pressure, for that exchange rate to decline due to euro selling in the middle of
the various European sovereign credit crises. There was a large community of PMs who were buying EURCHF
every time it got very close to or at 1.20, because they could not lose money. When the SNB finally let the floor go
in January 2015, EURCHF plummeted more than 30% in the space of a few minutes without anybody being able to
stop-loss, and many hedge funds went bankrupt.
Here, we want to highlight two items mentioned earlier in this book and this chapter. One is that the speculators
that got eventually crushed by the oncoming train always got better jobs as soon as the train wreck was cleared.
They had developed a reputation as big deal traders who made money consistently. Second, this highlights the issue
that in fact there is no credible way to track long-term real performance. The only measurable performance,
particularly for sell-side traders that we know personally, is that they made a ton of money for themselves. In good
years, they got paid very handsomely, and when they got run over, no one told them they had to give back the
previous year’s bonus. New compensation rules are supposed to temper this behaviour, but there is a huge
infrastructure in place at the individual and institutional level to privatise gains and socialise losses.

Follow the crowd

A less reliable but effective way to make money for a while is to follow the crowd. There are large crowds and
crowded movements such as the dot.com and cryptocurrency crazes we referred to earlier, or simply being long the
most basic stock market index from 2008 to 2018. In FX, it gets a bit more difficult but it can be done. If a PM has
enough connections, it is easy to keep talking to a lot of other PMs or asset managers, and even without knowing
anything about the market buy when others buy. If the PMs contacts are part of the smart crowd, those who create
trends instead of following them, then our PM will be buying in the early stages of trends and selling in the early
stages of trend ends. A variation of this approach is to know who has bought before a fall in prices, how much they
have bought and how close they are to their stop-loss.

Predictive models
One of the most popular ways to make a living, albeit one of the most difficult, is to design a quantitative model to
predict exchange rates. This is the classic expert figure – generally, somebody with quantitative background,
possibly in economics, at times with nothing at all to do with finance, exploiting the rebel or the outsider branding.
The variations on this theme are absolutely limitless, and arguably any of the methods explained above and below,
in some form or other, fall into this category.
The classic underpinning of the quantitative model exercise is the concept that the path of future price movements
is hidden in the collection of historical data. Since the human brain cannot accurately process and appropriately
weigh all the secrets of the past, a computer can help. The more information it is fed about the economy, market
forces, supply and demand, hunt for yield and other assorted topics, the better the design and outcome of a model
that gives us the future. Generally, in its most humble incarnation, it is an Excel spreadsheet. Up until recently, the
individual’s infinite wisdom, encapsulated in a set of equations or rules that, when fed with the appropriate (usually
publicly available) data, will generate a prediction about the direction of the market. Such an exercise typically takes
daily data and works on a relatively low frequency, with trading happening maybe daily or weekly, and with an
average position holding period of a few months.
Sometimes the model comes from a bank. It is used to generate discussions with buy-side clients, to offer them
trade ideas, and to give their clients access to the experts who have been able to design such a successful model.
They are a starting point for hedge fund traders to tweak with their own special sauce. They rarely make money as
is. When the models are created and managed within a hedge fund or asset manager, then the story might be
different. Over the years, there has been a number of legendary models that have been very successful. However, no
model works forever and under all conditions. Spectacular failures have happened, above all the spectacular
implosion of Long-Term Capital Management (LTCM) in 1998. As mentioned, this hedge fund had been operating
with much success for a few years, employing, among others, 1997 Nobel laureates Myron Scholes and Robert
Merton, when a series of highly leveraged fixed income positions led it to bankruptcy. The fund had been so
successful and the positions grown so massive that the size and leverage of its positions necessitated a Fed-
coordinated rescue by a syndicate of 16 financial institutions. LTCM’s strategies looked pure genius and
mathematically rational, but as John Maynard Keynes supposedly said, “the market can remain irrational longer than
you can remain solvent”. As an aside, Bear Stearns – who were closest to LTCM – was the only firm that refused to
be part of the bailout.8 The Fed never forgot. In 2008, they got even by refusing to salvage Bear Stearns’
shareholders. They handed over to JPMorgan and Jaime Diamond all the Bear Stearns crown jewels for the support
they received a decade before.
Thousands of books have been written on financial models that help predict future market movement. At best, we
will skim the surface. The simplest and most intuitive is the economic model. A team of economics PhDs design few
simple algebraic rules that take economics variables as inputs. Those would include GDP, inflation, unemployment,
interest rates and others, combine them mathematically and output the future return of one or more securities. Most
commonly, they would predict exchange rates or interest rates paths, but possibly also equity indexes and the like.
One of the problems that these models encounter is the lag. For example, it is very difficult to estimate how long it
takes hyperinflation to weaken a currency. Another problem is that they classically disregard market positioning.
Hyperinflation could potentially take a currency very quickly to its knees, except where market participants are
currently mostly short that currency, so that there is nowhere to go for the market but higher, as you need fresh
sellers to move the price down. However, econometric models such as those described above have a great
advantage: they sound good. They are usually designed by very serious people, are easy to defend as it stands to
reason that economic variables govern financial asset prices, and they are easy to explain to any board of directors
who will be very quick to agree that, if GDP growth exceeds expectations, then the currency should appreciate. It
sounds very logical but, then again, look at what happened in Japan.
Enter the need to include market positioning and quicker reaction time into the models. If you want to include
market positioning, then your issue will be where to get that information, as the FX market has historically been
very opaque since trade information is not public. The Dodd–Frank Act has changed part of that. Option and NDF
trades, where either counterparty is a US entity, by law is now public information. Nonetheless, the information
becomes public with a delay, which diminishes its value.
Figure 15.1 shows the Bloomberg SDRV screen that analyses data received from the DTCC. It has cleared trades
in FX NDFs, options as well as fixed income. It is incomplete and delayed, but revolutionary for the FX markets!
Never before Dodd–Frank did the market have any sense of what is trading unless they had the inside track from the
big banks or waited for the Bank of International Settlements survey, which is published every three years. This data
is pure gold for these models.
The logical evolution goes from models that are completely dependent on public economic and flow data to
models that incorporate high-frequency trading, pricing and positioning data. The economists are being replaced by
traders who have experience in how the market reacts to any sort of new information, and are able to figure out
simple but street-smart ways to exploit opportunities.
 The third evolution is born from the realisation that there is only a finite number of trade ideas (also known as
“old dog rules”) that traders can come up with for trading. The reason is that humans have this innate desire for a
rationale to be behind what they discover. We have a need for a concept to be behind the existence of an edge. This
makes a lot of sense, but is also invalidated by a key development in computer science that changed the landscape of
model trading forever: pattern recognition. A typical example of computer pattern recognition is when a computer
program is fed YouTube videos for a sufficiently long period of time, and ultimately becomes able to identify cats.
What the computer is doing is trying to find patterns, or instances of similarity, without being told what to look for.
Let us consider an example in the markets. I can easily instruct a computer program to calculate correlations
between a very large number of assets. If two assets are highly correlated and one rises while the second does not,
then sell the first and buy the second. However, this is not an example of pattern recognition because I have told the
computer what the pattern is. A better example would be when the computer figures out all by itself that the stock
market moves after non-farm payrolls releases are much smaller when the initial public offering (IPO) calendar is
particularly large for the following quarter. This is much more difficult, and I certainly did not tell the computer
anything about this pattern. The computer does not know why this rule works. Actually, I don’t think there is a clear
reason why it should. We can conjecture that the market move should be larger due to large positioning ahead of the
IPOs, or that traders are paying more attention to IPOs and therefore are not trading around the non-farm payrolls
release. But, then again, equity traders usually don’t trade bonds so the link in itself is suspicious. In other words, it
would not be very difficult to come up with some sort of logical explanation, to make a strong case.
The main concept behind all this is that now we have trading models that have no rational justification by an
expert. Trading ideas are being discovered automatically by computer, and no explanation is necessary (or maybe
even possible).9 The challenge that remains is how the computer will know when the pattern has ceased to be
effective – how to stop trading that particular model, and concentrate the investment in the other two hundred
models that the computer has discovered. Additionally, we refer back to our original mythical perceptions of the
lone gallant speculators – we realise that there is no place for them in this world. AI requires massive investment in
computers, data and programmers, not guys that can feel the market.

ARE EDGES ETHICAL?

The laundry list of edges we have reviewed superficially, hopefully dispels the idea that market speculators are
divinely endowed with special talents. However, we need to acknowledge the reality: to make money, PMs need to
invest repeatedly, every day, to control costs, manage profit and stop-loss levels, decide how much to invest for each
trade, and so on. Even if a random person gets the market direction right today, it is unlikely they will be able to
manage all those details.
When the broad public does come to understand some speculators make money by exploiting an edge, the
reaction is often mixed and evolving over time. Back in the 19th century, the Rothschilds complained that the spread
of fast communication had made it impossible to make money any more. They used to be the only ones to
communicate via telegraph and therefore arbitrage the price differences between the Philadelphia and New York
stock exchanges. There is a very powerful point being made in that famous remark that “one cannot make money
any more”, about the belief that their speed advantage, their edge, was an integral and sacrosanct part of how to
make money trading. In fact, it was the very essence of how to make money. They thought that the absence of the
edge would put an end to the money-making! They thought that the mere act of owning knowledge should be
enough to generate money. Clearly, in their time, and especially when they had enough wealth to yield power, this
was an acceptable edge. In the current financial and regulatory framework, we consider it immoral and illegal to use
exclusive information to trade public companies. It is just catching up to the FX market. The creation of an even
playing field in terms of knowledge seems to be a societal goal.
However, this knowledge issue morphs and has a metastasis as our technology and trading modes are
revolutionised. Consider a group of PMs who set up a computer network with communication cables so fast between
Chicago and New York that, even if all the banks use computers to weed out arbitrage, minute changes in prices that
happen in Chicago will be known by the PMs in New York before anybody else. Those PMs will be able to buy in
New York before the price in New York aligns with Chicago. This is familiar to those who read Michael Lewis’
book, Flash Boys, where the author explains in detail the entire high-frequency trading industry. The “flash boys”
industry is predicated on the ability to be the first to know about price changes that happen in one location, and be
fast enough to exploit that knowledge before it becomes common.
Once recognised and exposed, this modern manifestation of the knowledge edge is suppressed. Regulation is
introduced to enforce the concept of public information and outlaw insider information. As a society, we’ve decided
that having an edge of speed is a bad thing for capitalism and the proper functioning of financial markets. As a
matter of fact, when an edge becomes that obvious, regulators generally try to eliminate it. This is a very pivotal
change in the concept of trading – it means that whenever you are able to explain in a simple fashion why you are
making money, this can be construed as an unfair advantage and might be challenged by regulators. The financial
regulatory community has arrived at a very romantic way to look at things. Our lawmakers are saying that they want
traders to only make money with their inexplicable God-given talent, with their inspiration and their intuition. Any
simple edge will easily be the subject of popular ire. How do we reconcile that with our thesis that all successful
speculative money-making is based either on an algorithm or luck?
In truth, how edges are perceived has gotten a dramatic black eye since the late 2000s. Key to this attitude
surfacing for the general public has been the very novel (certainly for the US) and pivotal popular uprising of
Occupy Wall Street (OWS). Without passing any judgement on OWS, its vilification of Wall Street types and their
use of privileged edges to amass massive wealth has had a profound effect on the public psyche and subsequent
regulations. The OWS movement was born as a consequence of the GFC of 2008, and was composed of a
spontaneous popular coming together at Zuccotti Park in lower Manhattan. OWS sought above all to address income
inequality, corruption and the undue influence of banks on government. By the way, decades ago I worked across
the street from the park, and spent many lunch hours watching the Wall Street types play chess with random people
that would come there to hang and play. It was called Liberty Plaza Park at the time and it always surprised me that
the OWS crowd did not use its original name – I assumed it would have resonated more with their cause.
It is clear that OWS was a very powerful source of inspiration that changed the public’s attitude towards banks.
Banks were more likely to be seen as negative forces within society, and as agents of theft. OWS was very
successful on this point, and as the public’s attitude changed, so did the government’s. Very important legislation
was passed as a consequence of the new mood. Some might argue that the winds of change were already afoot, as
the all-important Dodd–Frank Act was passed in 2010, but its implementation is certainly contemporaneous or
subsequent to OWS. Apart from the legislative activity, enforcement agencies started many more investigations,
levied many more settlements from banks, and concentrated on the accountability of individuals. In particular, very
onerous monetary settlements put pressure on the large banks, but did not substantially alter the picture of their
bottom line. On the other hand, personal accountability had profound consequences on the appeal of Wall Street
careers.
OWS criticised jobs where it seemed very easy to make a lot of money simply because there was an inherent
edge. They highlighted the restrictive club of white, educated men that were able to protect their very high salaries at
the expense of the rest of the population – famously, the “99%”. Part of OWS’s contention was that behaviour such
as collusion or outrageous fees are not beneficial to society. Typical examples cited were the packaging of hundreds
of residential bonds in structures that received favourable credit ratings as a consequence of colluding behaviour by
banks and rating agencies. Once a correction in the real estate market began, the high leverage and the high credit
rating of those securities ensured that a credit bubble of unimaginable proportions would burst – which was
obviously later called the GFC of 2008.
However, investigation of collusionary activity did not stop with the banks and S&P. The regulators discovered
collusion in the manipulation of interest rate and FX fixings. We have addressed these in the first and second parts
of this book, but in summary the cashflow of many financial instruments depend on the daily value of specific
interest rates or exchange rates, which are fixed by a panel of expert banks every day. The very same banks own
large amounts of securities whose cashflows depend on the fixings. It is not a big stretch of the imagination to see
that they had a very strong incentive to manipulate the fixings in a way that made the related cashflows more
favourable to them. The fact that for decades nobody questioned the intrinsic conflict of interest is maybe more
interesting than the obvious finding that, over time, many traders had taken advantage of it.10
Even more interesting is what they said the moment they were on the stand. You will not be surprised to hear that
many invoked that trickiest of concepts, the edge. Many of those traders claimed that they were taught how to
collude, and that all their colleagues and bosses did it, and that it was an integral part of how their trading desk made
money. In their view, manipulating the fixings was an integral part of the functioning of a bank as much as it was to
borrow from depositors at one rate but to charge a higher rate when lending. This activity of course is considered a
perfectly ethical and sound way to make a buck, and is in fact a foundational concept of western banking. I must
confess, as someone that benefited from that culture, it did not seem wrong at the time and we certainly did not see
the broader damage to society. Listen to the defence arguments in the court documents we cited earlier in this book,
and you will see we are not alone.
It remains to be seen if the concept of outlawing each and every edge is realistic. Remember that the very essence
of finance is to eliminate fat cats and cartels by allowing the free flow of capital. If you discover some inefficiency
but the government sees it as an edge and forbids you from arbitraging it, fat cats will continue to have a great time.

1 A portfolio manager (PM) decides what positions to take, and the trader is in charge of executing the trades in the most efficient way.
2 Urban legend tells of numerous fans meeting Michael Douglas, the main character, or Oliver Stone, the director, of Wall Street, and telling them they
chose a career in finance after watching the film.
3 As a side note, PMs are very often hired on the basis of their multi-year returns, but these returns are all but impossible to verify. There is no public
record and past employers are not obliged to disclose anything.
4 Remember the famous words: “The emperor is not as forgiving as I am.”
5 See S. Lack, 2012, The Hedge Fund Mirage: The Illusion of Big Money and Why It’s Too Good to Be True (Hoboken, NJ: Wiley).
6 About two hundred years later, Pablo Picasso would famously quip: “Good artists copy, great artists steal”.
7 A very simple example is the stellar performance of Venezuelan bonds in 2015 and 2016. Funds that did not purchase those bonds due to their very
high risk heavily trailed their competitors in performance, and potentially lost assets due to that. However, Venezuela technically defaulted at the end
of 2017.
8 R. Lowenstein, 2008, “Long-Term Capital Management: It’s a Short-term Memory”, New York Times, September 7, 2008 (available at
http://www.nytimes.com/2008/09/07/business/worldbusiness/07iht-07ltcm.15941880.html).
9 Once again, after machine learning algorithms have identified a pattern, it is not possible to examine the computer memory and figure out how the
computer arrived at that pattern.
10 As another famous sultan once said after refusing an invitation to the racetrack: “I know already that some horses are faster than others, and I do not
care to find out which ones.”
 16
Speculating and Hedging: The Fundamental Differences

The previous chapter delved into how speculators and PMs are perceived by the market, and the most common
approaches they use to differentiate the way they make money. In this second chapter on speculators, we examine
the difference between hedging and speculating, and take a deeper dive into relative value trades. We then raise the
bar and look at how a PM would speculate across many currencies at the same time – what criteria would be used to
judge performance and what would be the most efficient way to allocate capital. Finally, we present a cookbook of
different speculative FX option structures that can fit any market view. However, we don’t examine those in detail
like we did for corporate hedgers, as that would require another book.

HOW IS SPECULATING DIFFERENT FROM HEDGING?

What is the difference between trading by corporations and trading by speculators? The first fundamental difference
is that hedging is a problem with a solution. The hedger expresses an objective, which could be to reduce the
volatility of earnings year-on-year, or maybe to defend a budget rate. Whatever the objective, it is very narrowly
defined and its solution can sometimes be expressed by a formula. The way to minimise that risk has a mathematical
solution, which then becomes the optimal hedging strategy. Corporate treasury can figure exactly how to hedge, and
anything other than that mathematical certainty is suboptimal. In that sense, corporate hedging is well-suited as a
scientific, intellectual pursuit of absolute and unshakable truth.
Not so for speculators: they are in pursuit of profits, and can trade any instruments they want. They have a wide
array of instruments to choose from and a wide mandate in terms of the markets where they can trade. A PM is
therefore always confronted with difficult choices. Many markets can be attractive, at least on paper, and the PM has
to pick which one to trade in. There is no right or wrong choice at the start. This is why PMs are bombarded with
trade ideas from the sell-side, their own research and strategy teams, and their traders. At the beginning, this
sounded like a puzzle to me – how can analysing the markets and compiling trade ideas not be the job that is paid
the most money? But, no, the PM is the job that is paid the most money.
The function and figure of the PM was tough to figure out. Once the analyst had calculated that USDBRL was
supposed to collapse after the election, then all the work was done. However, with all this mountain of research and
trade recommendations, there has to be somebody who actually takes ownership of the P&L and pulls the trigger. It
sounds paradoxical that producing the information and acting on it would be two separate functions and two
different people. The PM receives all the information and has to make a final call on where the capital will be
deployed. The brainy analyst is not the best person for that task. This will forever be confusing for me, as my
scientific sensibility always thought that the money should belong to the person who provided the intelligence, just
as the loot should be given to the ironsmith who made the sword. But I was wrong, the treasure will always belong
to the knight who took the sword and went to get the treasure and risked their life. Let’s see how the new world,
with AI systems taking over the analysts and algos taking over the execution, will change that dynamic.
The second element that separates speculators from corporate hedgers is timing and flexibility. When a corporate
hedges, a series of trades are executed that under any market condition protect the corporation from the predefined
risk. The corporate knows that executing the plan will have the intended consequence to reduce that risk. It may be
complex or expensive, but the result is guaranteed. Not so for a PM. When shorting the dollar, the PM is not sure
whether the real will rise. For this reason, the PM is allowed to change their mind at any time and adjust the
execution details of the trade midway. All this is possible, and might even constitute the core of making money.
For example, suppose that after purchasing a US dollar put Brazilian real call, some news comes out that is US
dollar bullish. Right away, the PM decides to delta hedge the option by buying US dollars and selling Brazilian
reals. As spot rises, the option loses money, but the delta hedge makes money, thus immunising the PMs’ position
(for a more detailed discussion of how this works, see Part II of this book). After a rising trend for the morning, the
market quiets down and realises that the news was mostly known already, and that there is not much of a follow
through. At this point, the PM’s market view might come back in play and the PM unwinds the hedge.
Let’s try another more complex iteration. Suppose that the Braizilian real bullish PM had not bought a US dollar
put Brazilian real call, but rather sold a US dollar call Brazilian real put. The PM pocketed a lot of premium, but
now the game is to defend that premium from the possibility of a US dollar rise. Suppose, as in the previous
example, the US dollar starts to rise and the PM, panicking as the option becomes ITM, should delta the trade by
purchasing USDBRL. Recall from Part II that when delta hedging a negative gamma position, every hedging trade is
a losing proposition. However, there is another possibility – do nothing. Be strong in your conviction of a higher
Brazilian real. No delta hedge means no locked-in loss.
The point is clear: there are many possible scenarios, and at any given time the market view is reassessed. The
PM can even have multiple market views for different time horizons. They may even buy USDBRL as much as sell
USDBRL. The delta hedging activity gets mixed up with the original bearish USDBRL option position. It is obvious
that in such a complicated sequence of events the final P&L of the trade will be generated by many transactions, and
by many decisions taken during the entire life of the trade. This is very different from the buy-and-hold stance of a
typical corporate treasury. Many traders will tell you that the position they take early in the morning is irrelevant to
the P&L they had at the end of the day. That depends, mostly, on how they traded during the day. For example,
some traders used to start a day by having one of their trading books sell a vanilla option to another one of their
books. During the day they would try to delta hedge as much as possible the book that bought the option, and be as
stoic as possible and trade the other book as little as possible. To the non-initiated, this might smell of a self-
aggrandising comedy of wasted time, but often, both books made money.
The consequences are important. First, dynamic continuous trading includes options but also spot or forwards. For
this reason, it is much more difficult to explain why a trade made or lost money. This freedom to tweak the trade or
react to live developments in the markets makes speculative trading a psychological game. The typical circumstance
is when a PM is long and the market is falling. It might be easy to just close the position and take a loss, but it might
also be a great idea to wait for a change in momentum and for the market to start rising again, making the trade a
winner. The ability of the human mind to rationalise which of those two alternatives will be the better one in each
situation is limited, and emotions always muddle the picture. Stories of losing or winning trades are always
accompanied by silly pearls of wisdom, pronouncements better suited to fortune cookies. Typical ones include
advice on how to always cut your losers (losing trades) fast but ride the winners. Relating this to behavioural
economics, the human tendency is to feel the pain from losing trades more than the pleasure from winning trades.
This pushes us to keep delaying the inevitable closing of losing trades until losses are large, and to hurry the
pleasure to close a winning trade at a profit by exiting the position while the market is still moving in our favour. All
this is part of classic trading lore, again going back to the days of the celebrated 1987 film Wall Street by Oliver
Stone.
The third difference between the trading activity of a speculator and that of a corporate treasury practitioner is the
presence of relative value trades. A corporation almost always is exposed to the market price of some financial
instrument. This could be the cost of an expenditure – the price goes up, the corporation is worse off, the price goes
down, the corporation benefits. Speculators, Christine (described in Chapter 15) in particular, have the luxury to
enter more nuanced trades where they can customise the financial variables that determine whether their trades are
winners or losers. For example, a PM might buy two-year Treasury bonds and sell 10-year Treasury bonds in an
equal amount. This is a classic “relative value” trade, or spread trade.1 The P&L of the trade will not depend on the
level of the two-year price but on the spread between the price of the two-year and price of the 10-year. A speculator
may anticipate a change in the correlation between US dollars, euros and Japanese yen, and trade EURUSD and
USDJPY.

WHY BOTHER WITH OPTIONS?

There is a great deal to talk about when it comes to relative trades, which we will cover in the next section.
However, before that, it is important to see how a speculator may experience speculating with options versus spot. A
PM will need a market view to start trading. Assume they are bullish on the Brazilian real with a three-month time
horizon. This is convenient since USDBRL (“dollar Brazil” in market parlance) is volatile. The rate is a reflection of
the performance of the Brazilian economy, it is subject to capital controls, and its trading is subject to liquidity
shocks. The first possibility is, given the market view, for the PM to just sell US dollars and buy Brazilian reals with
a maturity of three months. To make the example interesting, consider the strategic time period preceding the second
election of Dilma Rousseff to the presidency on October 26, 2014.
Why would the PM want to do anything other than a forward trade? One consideration will be credit. Hedge
funds tend to use leverage to amplify any positive returns they might have, and to use leverage they need lots of
credit, usually extended to them by the prime brokerage group of a large bank. The sale of US dollars via a forward
trade is generally a very credit-intensive trade, even for just a three-month tenor. Replacing the forward trade with
the purchase of a US dollar put Brazilian real call reduces the credit to almost zero. There is some exposure for the
bank from the trade date until the hedge fund pays the premium on the option, generally two business days later.
After that, the option is an asset for the hedge fund and a liability for the bank, and therefore there is no credit risk
for the bank.
Now let’s look at some general considerations. Brazil is an emerging markets country, and as such conforms to
the general observation that it is a high interest rate economy. As explained in preceding chapters, this means there
is an advantage in buying Brazilian reals and selling US dollars: the interest rate differential. If you look at Figure
16.1, you will see that on October 21, 2014, USDBRL spot was 2.4832 but the three-month forward rate was
2.5451. This means an annualised interest rate differential of about 10.25%. In Figure 16.1, looking at the “3M” row
showing data for the three-month maturity, we can see that the US interest rate was at 23bp (annualised) while the
corresponding Brazilian rate sat at 10.91%, and therefore the interest rate differential was roughly 10.25%.

Another consideration, and a very important one in this case, is the volatile history of USDBRL spot. Figure 16.2
shows the spot history of the USDBRL exchange rate in the five months leading to the 2014 Brazilian presidential
election. Note the 8.89% decline. The figure shows the recent spot context for the PM who wants to short USDBRL.
In the previous four months, the real lost 8.89%. A similar decline at the time this trade is being considered would
take the spot level from 2.4832 to 2.7052. A short USDBRL three-month forward position at 2.5451 would lose
6.29%. This worries the PM.
 The environment for hedge funds has changed significantly with the 2008 financial crisis. There has been a
reduction in their risk appetite, and most importantly in their investors’ patience with a negative mark-to-market
position until the trade turns positive. Investors threaten withdrawal of capital at every opportunity. This means that
it is really important to have positive returns for a fund as a whole every time returns are disclosed. In addition,
banks that allow hedge funds to take levered positions by lending to them through their prime brokerage arms are
extremely stingy with their balance sheet. Recall in Part 2 the discussion on the cross-currency basis and the
important role of “renting the balance sheet” in allowing the basis arbitrage to close. A loss of 6.29% might simply
be too big for the PM. Prior to the changes brought on after 2008, the PM may have been able to have a mark-to-
market of negative 6.29% on a position, but would be allowed to keep the position and stay in the trade.
The changing dynamics of how risk is perceived can be seen from an encounter I was privy to. A portfolio
manager initiated a position in USDCNH (dollar China) of just US$20 million. She was called into the risk
manager’s office and told the position took her over her risk limit. She protested, saying that the volatility of the
currency pair was small. The risk officer asked her a question: “Am I confusing this with another currency, or is
China that currency where they have a meeting each day and just decide what the exchange rate should be?” The
PM protested, that was the case, but the proof of the low risk was in the fact that realised volatility was low. The risk
manager did not change his mind. As long as the exchange rate was arbitrarily decided and not freely floating, the
position had to be reduced. We highlight two observations from this story. The first is that the risk tolerance of
hedge funds has decreased substantially. The second is that before 2008, the PM would have screamed at the risk
officer, called a meeting with a manager and, in all likelihood, would have been able to keep the position. Those
days are gone. Risk managers hold much more power than ever before, which comes from the fact that investors are
hypersensitive to risk and ready to pull assets. Many PMs now complain that it is not possible to make money under
these conditions – or rather that only algorithms can make money in these conditions. Ah, and one more thing, the
PM in this anecdote was actually the real Bonny in our story.
Typically, algorithms have a much shorter time horizon for trading, and can therefore initiate and close hundreds
of trades each day. In that scenario, algorithms can afford to make less money on each individual position, as their
daily P&L is the aggregation of possibly hundreds of trades. Bonny claimed that, to be successful, one should be
able to leverage one’s position in such a way that on “a really good day”, one would make 1% of assets under
management. Suppose that the asset traded has a volatility of 10%, and that a very good day means a day that
happens once every quarter, or approximately once every 65 trading days. This means that a very good day has a z-
score (number of standard deviations from average) equal to 2.16 standard deviations (the probability of 1/65,
corresponds to this number of standard deviations), which multiplied by 10% (the volatility) and divided by the
square root of 255 (to de-annualise and make the number a daily move) gives 1.35%. In other words, if the asset
value moves by 1.35% the trader should be able to make 1%. This seems very plausible, but then the trader would
have to have invested 1%/1.35% = 74% of their trading capital in that one asset. First of all, 74% of the capital in
one trade is a large risk, and, second, this creates a risk concentration that will increase the VaR of the trading book,
which could hit the PM’s risk limit. More common weights for a trading portfolio would be below 10%.
The above argument may be compelling as to why a PM might see an option trade as a very attractive proposition
when trying to avoid huge stop-losses in case the market view is not correct. However, why not use stop-loss orders?
The PM can place stop-loss orders very close to the current market level and never lose a massive amount. This is
good in theory, but not so much in practice. First, recall the case of the EURCHF in Chapter 14. Although that is an
extreme case, slippage in USDBRL is notoriously high. Second, placing a stop-loss order 20 basis points from the
current level and a takeaway profit of 2% away may seem like a strategy that cannot lose, and which in the long run
will give you sure profits. Sadly, many PMs have found out that is not so. At the time of writing, I was talking to a
trader who had just been fired. His market view was correct 70% of the time, but then he got stopped out 50% of the
time, leaving him with a 35% success ratio and eventually no job.
That is not to say traders want a damping of volatility. What traders and PMs fear is a complacent and somnolent
market where spot does not move much at all. However, sudden bursts of volatility that can move rates by half a
percentage at any time, triggering conservative stop-losses and make a loser of any strategy. This contrasts with
Bonny’s desire to take large bets that can really impact the performance for the year (propensity to have few large
winners), and the desire of the hedge fund management to allocate strict risk limits to each PM.
We continue our inquiry by analysing more of the details. Enter into a bullish Brazilian reals view utilising a US
dollar put Brazilian real call vanilla option. The price for such an option is shown below. Figure 16.3 provides a
comparison between two trades that a PM bullish Brazilian reals versus US dollars with a time horizon of three
months might have executed on the eve of the Brazilian presidential election. The first trade is just a forward and the
second is a plain vanilla option. Time of execution is a few days before the vote, on October 21, 2014. The vote
happened the following weekend.

One of the advantages of the option is that in a worst case scenario the buyer loses the premium, which is 3.86%
of the US dollar notional. In USDBRL, on the eve of the presidential election, that does not sound like a bad
proposition. The other advantage of the option is that even if USDBRL rises, there is no stop-loss level to worry
about. The stop-loss is given at the beginning by paying a premium. After that painful moment, the exchange rate
can gyrate as much as it pleases, but the PM will never be stopped out of the trade, and the possibility of unlimited
riches is always there. Additionally, options can be used to multiply the exposure. Suppose that the PM’s risk limit
on a forward trade (in the case of USDBRL, it would be an NDF) might only allow them to initiate a US$10 million
position, as the risk limit is 8% of the notional. Note that the option above costs about 3.86%, so in such a case the
PM would be able to purchase an option with a notional of US$20 million, magnifying the firepower.

RELATIVE-VALUE TRADES
What is the advantage of relative value trades? There are various theories and no clear answer. In this section, we
will refer back to our two very different traders, Christine and Bonny. There are three characteristics of the relative
value trade we need to acknowledge.
Relative value trades can get overcrowded by the number of traders and by algorithms. Traders who believe this
will tell you the following.

Trends are killed by the desire or need to take profit early. It is very difficult to turn a profit in such a market unless
you have deep pockets, very low latency execution technology and/or the best algorithms. One-man shops
cannot hope to compete in this highly professional and industrialised market.
The only way to make money is to exploit cognitive (Kahneman) bias – in other words, to exploit collective fictions
that traders and programmers create to justify crowded trades. However, again remember that computers do not
suffer from cognitive bias, so in this area human traders definitely do not have an advantage over algos.

When a trade becomes crowded, people will tend to enter into the trade from any open door. A typical example is
when corporate bonds become popular, and prices rise and become stretched so that PMs start looking for something
similar but not as expensive. They start buying high-yield bonds, thus compressing spreads to levels that are
unreasonable. Eventually, it all comes to tears when the high-yield bonds start falling in value and corporate bonds
continue to rise, thus proving that it was a very wrong decision to try and get a better entry level with the purchase
of high-yield bonds. Another textbook example is the carry trade we’ve talked about. In USDBRL the carry might
be 5% per annum, so a trader starts by going long Brazil, and the trade starts making money. However, as the real
rises and approaches new highs, the trader is reluctant to enter into new trades, the entry point is now more
expensive. Then they decide to buy the Mexican peso instead of the Brazilian real. Now the carry is only 2.5%, but
the trade is still profitable as the peso also gains. This continues with them now buying the Australian dollar, with
the carry down to 1%, until some specific developments in Australia or Mexico cause a market correction. What
would Christine do in those markets? How would she take advantage of other people’s foolishness? In the first
example, she would sell expensive high-grade bonds and buy corporate bonds, while in the second she would sell
expensive Australian dollars and buy Brazilian reals.
There is an important advantage to relative value trades. Many PMs are constrained by the VaR of their portfolio,
not by any limit to the notionals of their trades. Suppose, from the above example, our trader bought reals and sold
pesos. The correlation between the two currencies will significantly reduce the VaR of their portfolio relative to the
sum of the individual VaRs. This is a big plus for the trader, as they would now be able to enter into larger trades.
Famously, Long Term Capital Management (LTCM) made it their speciality to exploit very small yield discrepancy
on bond spreads because they were able to enter the trades in very large notionals.
Finally, the last key characteristic of relative value trades we want to highlight is that they tend to bet on a
reversion to the mean. There is no real logical argument that can explain this. There is no reason for financial
quantities to revert to the mean. The fact remains that, in the particular case of spread trades, something is different.
In situations where the spread between two correlated prices is moving away from a historical, rational relationship
to very unusual levels, at some point this will reverse itself. How does this happen? Market participants decide to
buy one asset and have not given much thought to buying the other. They are just focused on one trade and, because
the trade is working, they likely have capital and risk limitations and will not feel the need to explore variations.
Consider this example. A trader is bullish on the real; they bought it, and the real started to appreciate. Their first
thought would be to buy more, not necessarily to buy the peso. This is where the relative value trader, Christine,
comes in. She would spend a lot of time monitoring spreads, and would notice that, with respect to the rise in the
real, the peso now looks undervalued. Relative value traders would sell the real and buy the peso with the view that
the temporarily severed correlation between the two would be reasserted, thus netting a gain to the relative value
trade. This is the opposite example to our first USDBRL USDMXN example. In this case, Christine would be the
first to move on to the peso. In our first case, she would be the last.
Risk reversal relative value
Now let’s examine a specific FX options application to relative value trades. Suppose our traders are considering a
risk reversal (collar) in USDJPY in the aftermath of the Japanese earthquake on Friday, March 11, 2011. On
Wednesday, March 16, the three-month 25 delta USDJPY risk reversal, as quoted in volatility differential terms, is –
1.8850%. For those of you that skipped the definition in Part 2 of the book, the risk reversal expresses the difference
between the implied volatility used to price a three-month 25 delta US dollar call Japanese yen put minus the
implied volatility used to price a three-month 25 delta US dollar put Japanese yen call (there is also a 10-delta
analogous definition). As we know, historically, the weaker and higher-yielding side of the market in USDJPY is the
US dollars, so usually this risk reversal is negative, meaning that yen calls are more expensive than yen puts.
Figure 16.4 shows spot (in white) and the 25 delta three-month risk reversal in USDJPY during the year before
March 16, 2011, a few days after the earthquake of March 11. Christine decides to fade the move in the risk reversal,
gambling on a reversal (pun intended) to the recent mean of about –0.60%. Note that the move in the risk reversal
follows a similar move in the spot level: in USDJPY they tend to move together and in the same direction. You
might also ask why USDJPY fell in the aftermath of the earthquake, signifying a strengthening in the currency of an
earthquake-stricken economy. Surely the earthquake was not a blessing for the Japanese economy? Of course not,
but consider that Japan has a high rate of foreign investment,2 and a part of those yen invested abroad had to be
repatriated, or were about to be repatriated, because they were needed for the reconstruction or just in general for
liquidity, thus creating demand for yen.
 Figure 16.5 shows the pricing of a three-month 25 delta risk reversal in USDJPY as of March 16, 2011. In the
figure, the investor is selling the US dollar put and buying the US dollar call, betting that the volatility of the US
dollar put will decrease relative to the volatility of the US dollar call.
You can see the full price. The risk reversal quote is now about –1.8% mid, with the US dollar put being quoted at
~15.37% mid and the US dollar call at ~13.11% bid, 13.84% offer. Suppose that the market changes instantaneously
and the risk reversal comes back to its pre-earthquake level of –0.50%. How much money would the PM make if the
notionals on both options are US$10 million? Suppose that it is the US dollar call volatility that moves from its
current level. We know that it is 15.37%–0.50% = 14.87%.
Figure 16.6 shows the previous risk reversal in USDJPY priced with the same details, including the same strikes,
but this time the implied volatility on the bought US dollar call is 14.87%, thus making the risk reversal quote about
–0.50%, the risk reversal medium-term average at that time. The value of the structure then moves from zero to
about US$30,800. This is assuming no passage of time and all other market parameters unchanged – not a realistic
scenario but a base scenario nonetheless.
 You might remark that 31bp, corresponding to US$30,800 on a US$10 million notional, is hardly worth all the
work, but remember that leverage will be possible here as we are talking of a relative value trade. It turns out that
such a move would represent about 10 standard deviations if we look at the history of the risk reversal. If –0.50% –
(–1.80%) = 1.3% is 10 standard deviations,3 and that corresponds to US$30,800, so one standard deviation is about
US$3,080, and the VaR4 is about 1.64 × US$3,080 = US$5,051, or, annualised, US$20,204.
To compare, calculate the VaR of a directional trade, such as buying USDJPY, can be of the order of 16.4%
annualised (assuming annualised volatility of about 10%, to make the numbers easy), which corresponds to
US$410,000 in three months (16.4%/4*10 million=US$410,000). If Christine has a VaR trading limit of
US$410,000, then with a directional trade she is only allowed to take a position of US$10 million, but with a risk
reversal the VaR is only US$20,204, so the notional she can trade is much much larger, equal to US$202 million.
Now you can see the advantage of the relative value trade. If, with a notional of US$10 million and a one standard
deviation move, the PM expects to make a P&L of US$3K, with a notional of US$202 million the P&L would be
US$60,000.5
For Bonny, who put on a directional trade in spot USDJPY, this would correspond to a very large move of 4.1%
in the exchange rate (again, over three months). As he has mentioned repeatedly, the problem with this comparison
is that in the remote past those large moves were possible, even in the span of a few days. However, in the current
market, with the large participation of algorithmic trading, the incentive for market participants to take profit earlier
instead of waiting for a 4.1% move has changed the game. The 4.1% move in spot will probably take much longer
and be accompanied by a number of retracements that could potentially hit a stop-loss and end Bonny’s trade. The
risk reversal, on the other hand, is something that is not subject to rapid-fire trading or the scrutiny of algos (not yet).
Figure 16.7 shows USDJPY spot in white and the three-month 25-delta risk reversal indicated by the ①. After the
trade date of March 16, the risk reversal reached a trough of –2.35% on the next day. At this point, the trade was in a
losing position but, by April 6, the risk reversal was at –0.68%, and the trade was ITM.
Delta hedging required
As you will note having read Part 2 of this book, this is an oversimplification of the relative value trade. Buying a
call and selling a put is a version of entering into an outright forward position. Recall that if the strikes on both are at
the forward rate, the total position is long the underlying and equivalent to a forward. As the strikes move away
from the forward, the position is still going long but the delta becomes less than 100% of the notional. Therefore,
this position can be both a relative value and a directional trade, up to a point. Look at what happened next (see
Figure 16.7). Initially, the trade lost money, as the risk reversal dipped more and spot dipped as well. However, by
April 6, the risk reversal was up to –0.68% and spot was at 85.54, both favourable moves to the trade. At that point,
the trade was worth US$36,000.
Figure 16.8 shows the market value of US$36,000 for the risk reversal trade as valued on April 6, 2011. This
change in value is due to both the change in spot and the change in the risk reversal. Note the deltas are now 2.5%
for the US dollar put (which is about worthless at this point) and 68% for the US dollar call. This trade is now a poor
reflection of the one Christine imagined due to the huge change in the deltas.
 The first thing to realise then is that this type of relative trade absolutely requires delta hedging to be a relative
trade. Since the trade is predicated on the risk reversal as a volatility spread, then logic imposes that the PM should
have mitigated the sensitivity of the trade to spot, since they did not have a strong spot market view. In fact, spot
moves are so large and frequent with respect to the moves in the risk reversal that they can easily wipe out any P&L
dependent on the risk reversal.
Assume Christine dynamically delta hedged from the beginning. This makes our analysis much harder, since to
execute a perfect textbook delta hedge the PM would have to constantly adjust the delta of the position, meaning at
every instant. At the limit, the frequency of the adjustment would be infinite and the bid/ask spread cost to enter and
then reverse such hedging trades an infinite number of times would also be infinite: this is why such a strategy is not
desirable. The PM will have to come up with a hedging strategy in terms of what would trigger a re-balancing of the
delta hedge. It could be done at fixed time intervals, or each time spot moves by more than a given threshold, or
each time delta moves by more than a given threshold, and so on in an infinite permutation of strategies (see Part II
for a more detailed discussion on delta hedging).
This is where, for a PM, the concept of backtesting becomes very fuzzy and almost useless. The corporate
treasurer can backtest a buy-and-hold strategy very easily, but any backtest of a delta hedged strategy is very
difficult as it is not easy to assume that any specific rule-based strategy would have been followed. For example, in
times of extreme market volatility, it is likely that the delta rebalancing would have been executed sparingly.
Imagine the delta jumping around by double-digit percentages one way and the other. To be fair, this flexibility in
the decision on when to delta hedge is giving the PM a possible hedge. For example, if spot rises then the position
becomes long gamma (see Chapter 13 on the greeks for the definition and preliminary consideration about gamma).
In such a situation, the PM hopes that the market becomes very volatile, and tries to execute delta hedge adjustments
as frequently as possible. The reason is that when the PM is long gamma, every delta hedging trade is a take-profit
trade. If spot rises then the PM has to sell spot. If spot falls then the PM has to buy spot. This buy low, sell high
activity is lucrative.
However, if spot falls to levels close to the strike of the US dollar put, then the position will be short gamma and
every delta hedging adjustment will be a stop-loss trade. If spot rises the PM has to buy spot, and if spot falls the PM
has so sell spot. This means that the PM is playing a game of chicken with the market. Every time spot moves, the
PM will try to resist recognising the pain, and will try not to adjust the delta hedge and pray that the move will
retrace. If the move retraces to the previous spot level, then the market is back to where it started and no delta hedge
adjustment is needed at all. If in fact spot does not revert and continues to move against the position, this is a very
dangerous game and sooner or later they will need to decide to throw in the towel and adjust the delta. Suppose spot
drops 20bp and the delta adjustment is to sell US$5,000. The PM does not think the move will stick. The market
continues to move and the PM does not do anything until the move is 100bp. At this point the PM should sell about
US$25,000, which is a much larger amount and at a much lower rate, so you can easily see that by playing chicken
with the market the PM can start losing P&L very fast. On the other hand, the PM cannot continue to adjust the delta
hedge every five basis points move, because then they would be incurring trading costs (bid/ask spread). As you can
see, what started as a pure relative value trade has fast become a glorified spot trading strategy. And this is just the
spot part of the deal.
In the above example in USDJPY, we looked at the value of the trade on April 6, which is not the expiry date of
the trade. However, April 6 happens to be the date that the risk reversal went back to relatively standard historical
levels, and it is therefore natural to assume that the PM might have started to think about taking the position off at a
gain. Keep in mind that in a scenario where the PM diligently delta hedges the trade, all the P&L of the combined
trades executed should then come from the change in implied volatility.6 All the P&L due to changes in spot should
have been neutralised by the delta hedging activity. Why should the PM take the trade off, or why should she keep it
on? One consideration would be the market view. The trade was put on in the first place because the risk reversal
had moved from historically average values to an extreme. Now that the risk reversal had come back, there is no
reason to believe that it would keep moving in the same direction. Therefore, the maximum P&L of the trade, at
least as its original rationale is concerned, was achieved.7
Now that we have examined a pure risk reversal trade, we want to show you a variation that uses more advanced
risk management techniques. Remember from previous chapters that in USDBRL the weak side is Brazilian reals.
This means that Brazilian reals domestic interest rates are higher. The corollary in terms of the risk reversal is that
US dollar calls are more expensive than US dollar puts – ie, they are quoted using a higher implied volatility. As
mentioned, speculators are willing to lend reals and borrow dollars, pocketing the interest rate differential, making
money for running the “credit risk” between the two countries and praying the FX rate doesn’t kill them. That is
called the carry trade. In volatility terms, speculators would put the same rationale in place by selling the US dollar
call and buying the US dollar put, betting that the persistent fears of Brazilian real weakness are unfounded. A
variation on the same theme would be to just sell the US dollar call and purchase a USDBRL straddle, on a smaller
notional, making sure that the initial price of the structure is close to zero.

What is the idea behind the trade? If spot does not move much the PM is long a straddle, which means long
gamma, and therefore the PM can hope to pocket money by delta hedging. If spot declines, then the PM could just
stop delta hedging as the US dollar put in the straddle will be ITM and provide the P&L on the position. But what
happens if spot rises? In such a case, the PM will buy back the sold US dollar call and sell another, again a 25 delta,
but this can only be done at a hefty cost, about 1% of the notional.
In Figure 16.9, the speculator sells a 25 delta US dollar call Brazilian real put, and buys a straddle. All expiries are
three months. The notional on the straddle is US$2 million per leg, and the notional on the 25 delta option is
determined in such a way that the aggregate structure is worth zero. This structure can work well if the market never
goes into a panic during its existence. The PM can potentially be long gamma all the time. In fact, in the scenario
where spot rises, the idea of purchasing back the sold option and selling another with a higher strike is meant to keep
the overall position long gamma, at the expense of the extra cost. In this example, it is clear that the trade execution
depends on continuous delta hedging, and is also contingent on a constant adjustment of the volatility and gamma
positions, in addition to the delta.

Volatility relative trades

After exploring the different risk approaches traders take into relative trades, we want to conclude with a discussion
of the main area of relative trading that FX option traders engage in. We will survey the typical monitoring systems
that traders use to research relative value opportunities, and make them a bit less artisanal. In the 1990s, a couple of
banks invested heavily in computer systems and developed expertise in trading volatility inconsistencies across the
curve. For example, they would price 15 delta options faster and tighter than anyone else in huge amounts. They
would then trade 25 and 10 deltas around their theoretical pricing engines, and do this until they realised the inherent
profit. This is a big simplification of the process, but you get the idea.
This can be profitable endeavour since each volatility surface has to mathematically hold together. Further
discussion of these topics is beyond the scope of this book, but suffice to say the volatility skew is so finely
calibrated as to prevent arbitrage. This deep understanding of the surface, combined with their huge balance sheet,
allowed these banks to take on a lot of risk. It made them great market-makers, which increased their reputation and
fattened their wallets, and allowed them to take on more risk, which enhanced their reputation, which brought more
volume, which allowed them to see more inconsistencies first, which made them a lot money, which made them
great market-makers. You get the cycle. In the early 2000s, many more banks got into the game and technological
innovations, usually a combination of internal proprietary software and Excel, started to automate the spotting of
these opportunities. Today, the mathematical knowledge involved in this game has been largely commoditised, to
the point where very few banks will tell you that they are able to make money in the FX options market because of
the quality of their in-house pricing software.
With that, let’s briefly explain further what we mean by relative value opportunities in the volatility space. They
can broadly be classified into two categories. The first has to do with a single currency volatility surface, which has
a few dimensions.

The first step is to understand the skew the relationships of implied volatilities between different OTM strikes and
between OTM and ATM strikes.
The second step is to understand how the term structure of volatility for that currency pair will move. Think of a
very simple example: there is an election on the horizon, in three months, and an extreme candidate may win.
The vol curve has crushed the one and two month but has a spike in the three month out to six, but then it looks
normal out to one year. The risk reversal is skyrocketing to the currency puts in the three month, but going the
other way for other periods. After the trader examines history, they believe this is unreasonable: they buy the two
and 12 month, and sell the three month as one relative value play.
Lastly, the trader can compare implied volatility anywhere on the curve to realised volatility, and decide whether
this implied vol will be higher or lower than future realised volatility.

The same analysis can be used across different currencies. The opportunity may be stated simply as “three-month
EURUSD versus three-month USDJPY is not in line with three-month EURJPY”. This activity involves arbitraging
a mathematical relationship that has to exist between three currency pairs that have a common currency. If one
knows two of those vols, it has to be the case that the third vol can be implied based on the covariance rule:
where σ EURUSD denotes the EURUSD volatility, σUSDJPY denotes the USDJPY volatility, σ EURJPY denotes the
EURJPY volatility and denotes the correlation between EURUSD and ρEURUSD,USDJPY USDJPY.
Often, however, theory does not converge to reality. That’s where the relative value traders can come in and try to
profit. See Figure 16.10, which shows software to run those calculations instantaneously.

Finding volatility opportunities

A colleague of mine had this great idea (he was not the first and not the last) and spent a month putting together a
humungous Excel sheet that would monitor all sorts of spreads. Their naïve belief was that banks could send to their
hedge fund clients all those calculations, and the hedge funds would find them so fantastic and compelling and such
a gold mine of trade ideas that they would reward the bank with more business. Apart from the general truth that
hedge funds have always been very good at taking free gifts, this process quickly turned into a grotesque exercise.
We would run the Excel sheet, check that it did not have errors and then save the output as a PDF file that we would
then email to our best clients. Pathetically, this file ended up being 20 pages of numbers, where outliers spreads, the
ones supposedly outside of their statistical norm and therefore subject to a future “return to the mean”, were
highlighted in red. Unluckily, a client receiving the file would have to peruse hundreds of numbers, without any
comment or indication that such a red number would have been better or worse than another red number. The bank
was basically churning numbers and dumping all the results on the client desks, saying: “Figure it all out by yourself
and, also, if you don’t mind, pay me for the inconvenience.” What we should have done is to select the very best few
spreads and made a persuasive argument to the clients as to how those are the ideas of the day. But, then again, this
was before the Volcker rule, so if we discovered a great arb – why would we not just exploit it ourselves?

As technology allowed a huge increase in number crunching capability on one’s desktop, buy-side traders
downloaded data from data vendor providers on decades of volatility surfaces and nuanced butterfly and risk
reversal data to divine where volatility was going next. We know many hedge fund traders that claimed to have
developed their own secret sauce by examining the data and backtesting to find “the trade”. Today, this numerical
sifting over market indicators is done quickly by machines. To correctly analyse all these patterns across hundreds
of currency pairs and across complex vol surfaces is a massive data project. However, you can also see the appeal.
With so many moving parts, it becomes very tempting to think: “I can find a unique dislocation that will make me
money.” Many of our friends and ex-colleagues who traded options spent a lifetime working with Excel to find
those patterns. This activity has been transformed by many more miners for gold, who have implemented the latest
AI systems that allow computers to figure out the secrets. We suspect, however, that no matter how smart the AI
program is at detecting patterns, domain knowledge will still be critical to capitalising on the pricing anomalies that
make money. For example, even after discovering a nuanced opportunity it is still the case in FX options that human
intervention is required to get the trade done on your side of the spread to make it financially worthwhile. Actually,
we need to reconsider that last statement. It may be that we’re just not ready to accept this final change where even
huge nuanced OTC option deals are done completely by computers and on the right side of the spread.
At the desk level, there are many tools now available to get a sense of the broad trends in volatility and to narrow
the search by first examining the visual clues. Figure 16.11 shows relative value of sifting in the FX volatility space.
This particular filter shows the current level of one-month implied volatility (the white circle on the rulers) as
compared to its three-year range (as represented by the ruler itself) and to its average (the white diamond on the
ruler). If the white circle is to the left of the white diamond, then implied volatility is historically cheap and “should”
be bought.

A couple of comments are in order. Old school trade idea theory postulates that, after you detect a variation from
historical behaviour, you should ask yourself why this is happening. Then you should ask yourself why this will
revert to the mean. Then you should ask yourself why this will revert now. The same school of trading argues that if
you do not understand why a trade is making money you should get out of it.8 With the advent of machine-learning,
the tables have now turned. Pattern recognition is a technique that by definition does not know why a trade or a
pattern is emerging; therefore, it will always be impossible to ask the computer model to give the reason why this
trade is making money. This reminds us of a typical situation in psychology and sociology studies, where the
researchers run experiments and discover a new pattern of human behaviour.9 Articles always end up with the
researchers trying to give a commonsense explanation of why the data exhibits this behaviour. Usually, the rational
explanation is half-baked and you can tell that it took them 10 minutes to put together. The casual reader of this
explanation, not a sociology expert, could easily come up with a dozen reasons why it could be the other way round
instead. Human nature likes a story; human nature likes explaining out the complexities of the world. This makes us
all very vulnerable as it leads us towards theories that can be explained with a soundbite, and away from
complicated and confusing theories that don’t make any intuitive sense but are at times the correct ones. Consider
the extraordinary contributions to this discussion from Brian Nosek at the University of Virginia. In a celebrated
2015 article,10 Nosek was able to show that less than 50% of the results of 100 published psychology studies could
be reproduced. Re-running the experiments gave different or opposing results from the original published study.
Similarly, finance in general suffers from many of the ailments of epistemology, so truth can be very hard to
discover.

Selective examination of relative vol trades

Without going into specific trades, consider a few basic elements from the categories outlined at the start of this
discussion.

One of the first tests to determine whether implied volatility should be bought or sold is to compare its current level
to its historical range and average. The comparison is then analysed with the help of percentiles. If volatility was
higher than it is today only 5% of the time, then the current level is at the 5% percentile. The smaller the
percentile, the more convincing the idea to sell the volatility. Of course, most of the time a trade idea is realised
using a specific trade. In this case, the trader would use the sale of an ATM straddle followed by dynamic delta
hedging to realise the concept of “selling volatility”. To be more complete, a volatility swap can be used as
potentially a more efficient tool but volatility swaps are beyond the scope of this book.
The second classic way to analyse implied volatility is to compare it to the realised volatility. There can be many
reasons why future expected volatility is higher than realised (past) volatility. To sell implied volatility, in the
absence of any obvious market condition, a relative value trader will take comfort in the fact that the implied
minus realised spread was higher than its current level only 5% of the time.
Figure 16.12 shows relative value sifting in the FX realised-implied spread space. This particular filter
shows the current level of one-month spread between implied volatility and realised volatility (the white dot on
the rulers) as compared to its three-year range (as represented by the ruler itself) and to its average (the white
diamond on the ruler). If the white dot is to the left of the white diamond, then implied volatility is historically
cheap and should be bought. Recall our delta hedging and gamma trading discussions throughout this book. A
stark reality pushes traders to put on the above positions. If, in fact, implied volatility is higher than the volatility
that will be realised over the designated upcoming period, the person that bought the option and delta hedged
will most definitely lose, while the person that sold and delta hedged will most certainly gain. Stated differently,
implied and realised have to converge eventually.
A third all-time classic metric for relative value trades is the risk reversal, as discussed in the previous section.
Traders can look at past levels of the risk reversal and compare them to the current level. All those systems only
make sense when the trader knows why the current level is different from historical levels. In one striking and
exhilarating example, I was attending a sales meeting at the end of March 2011. Somebody asked if there were
any interesting ideas coming from clients, and a senior salesperson mentioned that one of their clients was
studying the annual seasonality of USDJPY. The concept is that if you notice that USDJPY always falls in
March, then you should sell it at the beginning of March. Apparently this client had noticed some regularity in
past years. The whole discussion went south very quickly when another salesperson mentioned that Japan had
just experienced a gigantic earthquake on Friday, March 11, 2011. He asked on what basis the client was
expecting a repetition of previous market behaviour? Did the client realise that Japan will not have a humungous
earthquake every year in March? Silly ideas sometimes can have very long legs.
Figure 16.13 shows relative value sifting in the FX risk reversal space. This particular filter shows the
current level of one-month 25 delta risk reversal (the white circle on the rulers) as compared to its three-year
range (as represented by the ruler itself) and to its average (the white diamond on the ruler). If the white circle is
to the left of the white diamond, then risk reversal is historically cheap and should be bought. Unless there is a
clear reason why that is – for example, an earthquake has just struck Japan.
The last type of reversion to the mean analysis we will examine is the term structure analysis. Each currency has a
curve, called the term structure of volatility, which shows, for each tenor, the level of ATM implied volatility. In
a normal market, this curve should be slightly upward sloping, just as an interest rate curve would be, and for
similar reasons – which have to do with the concept of risk premium. If the term structure of volatility exhibits
some irregularity, such as a kink, a trader could enter into a trade that makes money in a scenario where this
irregularity disappears.
Figure 16.14 shows the volatility term structure on the eve of the Brazilian general election of October 5, 2014 (first
round). In the figure, we can see the volatility term structure in USDBRL as observed on three particular dates in
2014: August 8, September 8 and October 8. The August curve (in bright white) has a steep climb in the three-
month tenor, which encompassed the election results. The September curve (in pale white) then had a similar peak
in the one-month tenor, and the October curve (in dark grey) had no peaks as it was observed after the results of the
first round of the election. A naïve trader would look at the September curve and decide that the front-end of the
curve was too high with respect to historical averages, and the curve was inverted. The trade to execute in this case
would be to sell the one-month volatility and purchase a longer-tenor volatility. The idea would be to dynamically
delta hedge the position, which is not easy to do as you need to have a position that is spot-neutral. Therefore, you
would sell and buy volatility via delta-neutral or, approximately, ATM straddles. Then the position needs to be delta
hedged every day, or every five minutes, your choice.

What does the end-game look like in this case? The trader is betting that the term structure is too high for the one-
month point. However, if they keep this position for two weeks, then they are left short a two-week straddle (delta
hedged). Even if the one-month point is now lower and on par with past history, which was your original theory,
now they need the two-week point to be lower. This would likely not be the case because now the two-week point is
the closest to include the first round of the presidential election, and therefore the two-week point is now likely
much higher. This shift, from looking at one-month to looking at the two-week is called “the roll”, or sometimes
“rolling down the volatility curve”.
Assume that two-week implied volatility is lower than the level the one-month implied volatility was sold on
September 8. The good news is that even if the volatility level is the same, a one-month straddle is worth more than
a two-week straddle because of the longer tenor (remember the notion of time decay). Therefore, we can assume that
the trade to sell the one-month straddle and buy it back two weeks later made money.11 Also consider the long-end
of the trade, where they bought a longer-tenor straddle, say a six-month, and sold it after two weeks. In this case, the
time decay will hurt, but not as much as it helped for the two-week tenor, because as we know time decay is much
more impactful for short-term options.
These were just appetisers as to all the exciting things one can do with volatility. In this case, the Brazilian
election trade did not make money no matter at what point it was unwound. The concept of reversion to the mean
does not apply to high levels of implied volatility when there is a clear and relevant reason for volatility to be much
higher than historical levels, as in the case of a presidential election. Before betting on a reversion to the mean, do
your homework. However, as specialists of epistemology and the scientific method know, it is much easier if you
have to find a reason to invalidate a theory, rather than having to be confident a reason does not exist. Fans of the
philosophy of Karl Popper will rejoice.
CLEVER WAYS TO INCREASE SPECULATIVE LEVERAGE
Now that we’ve explored some of the human and market elements of speculating with FX options, we will finish
with a few fun trades that are common in the market, introduce a limited new set of bells and whistles – such as
knock-outs and reverse knock-outs – and conclude with a listing of many other ways that market views can be
expressed and tailored to one’s risk/reward profile. This section will analyse a couple of sample trades that allow a
speculator to significantly increase the leverage applied to their conviction without significantly increasing the risk.
It is by no means exhaustive.

Reverse knock-outs
The trade described below makes clever use of options to generate leverage and create better risk management for a
specific market view. This also will be a good example of a trade that typically does not require any real-time
management, and is therefore of the buy-and-hold type.

Turn the clock back to January 21, 2015. The PM is bullish on the EURUSD exchange rate, which means bullish
on the euro and bearish on the US dollar. They are preparing for the ECB rate decision and ECB president Mario
Draghi’s press conference the following day. The calendar of events can be seen in Figure 16.15. The PM expects
that the euro will rise as a consequence of the press conference, and more specifically they are targeting 1.23, which
last traded just about a month before.
 Figure 16.16 shows the history of spot EURUSD in the year preceding January 20, 2015.
The PM enters into two trades: buys euros versus US dollars six-months forward, spot is at 1.1550 and the six-
month forward is at 1.1574; and they also spend about US$56,000 to purchase a six-month 1.1574 euro call US
dollar put with a reverse knock-out at 1.23.
Let’s take a quick detour to define a reverse knock-out. In Chapter 13, we encountered the forward extra where
one of the option legs did not exist until spot reached a certain level. Then, out of thin air a significantly ITM option
appears and drags the effective rate back to close where it started. In this structure, we are introducing an option that
exists like a normal option, but if spot reaches a predetermined ITM level the option disappears. Typically, knock-in
and knock-out call options set the trigger below the strike, so that when the option is triggered it will be OTM. They
are very effective at reducing the upfront premium if the purchaser has a very well-defined view of future spot
behaviour.
Back to our story. If an option has the potential to disappear at some point, will it be as expensive as a regular
option, which will be there until expiration no matter what spot does? Of course not. As a comparison, a vanilla euro
call US dollar put with the same strike would cost about 3% of the euro notional. So the purchase of the barrier
option sounds very cheap. However, we need to be careful with our words at this point. The concept of cheap or
expensive is out of place in such a discussion. The question to ask is, “What is the probability that the barrier will be
triggered?” To figure that out, we can just price a EURUSD one-touch with the same expiry date and a trigger level
of 1.23.12 The undiscounted price of the one-touch divided by the payout of the one-touch is exactly equal to the
probability that EURUSD will ever trade at or above 1.23 within six months after the trade date. Therefore, the
actual price of the one-touch is slightly different due to discounting to present value, but it will give us an idea. In
this case, the price is about US$4.73 million/1.1550 = €4.095 million or 40.9% of euros payout.
Figure 16.17 shows some of the details of the position that the PM has taken. All are priced as of January 20,
2015. From left to right, you have the forward position, then the knock-out euro call US dollar put, then the
comparison to a much more expensive vanilla euro call US dollar put with the same strike equal to the six-month
forward rate, and then finally an estimate of the probability that 1.23 will trade in the following six months,
expressed as a one-touch option with a trigger at 1.23.
 How do we make sense of this numbers deluge? Let’s take it from the top. The PM is looking for the EURUSD
rate to go to a maximum of 1.23 in the next six months. They want to really pile on this trade because they have a
very strong conviction. They cannot just buy more forwards, as they will blow through their risk limits. Remember a
forward, like spot, hurts as much going against you as it is pleasurable when it goes in your direction. This is where
they get creative. An easy way to increase the exposure without increasing risk of loss would be to buy a euro call
vanilla option.13 However, that would cost close to 3% of the notional (the price in the figure is in USD out of a euro
notional), which would then put the breakeven of the trade at a minimum of 1.1574 × 1.03 = 1.1921 for the option in
isolation. If the target rate of 1.23 is reached, then the gain will only be ((1.23 – 1.1574)/1.23 × 2) – 3% = 8.80%,
which is good – but, given the risk being taken, they are looking for ways to increase it. Note that the return is
doubled, because in this scenario the exposure is now on €20 million: €10 million from the forward and another €10
million from the option. The really cool thing is the risk is only on half that amount because the losses can only
come from the forward trade, not from the option trade (other than the premium).
They consider this and are not happy about spending 3% upfront to buy unlimited profit as the euro rises to
infinity, when in fact they are convinced it will end up below 1.23. They get creative, and decide to purchase a much
cheaper option, which will disappear in case 1.23 trades. Now the premium is a mere 47bp, so ((1.23 – 1.1574)/1.23
× 2) – 0.47% = 11.33%, which is a substantial increase. Not exactly. Can you spot why? If 1.23 trades, the option
knocks out and they get nothing from the option. So the maximum return is close to but never 11.33%. Our
calculations are back of the envelope approximations, just like they would do in their head as they are thinking about
what to do next. In essence, we’ve found a way to turbocharge our view and find some leverage. Is this a free lunch?
Better returns, less risk of loss? No such thing in this business. An increase in all the good things comes at the
reduction of its probability of occurring. What if the knock-out option knocks out? This would be bad news. On
second thoughts, maybe not as bad as you might think – if the euro rises more than they expected and exceeds the
target level, they still have the forward position that will continue to make money. Assume the euro just touches
1.2362, then (1.2362 – 1.1574)/1.2362 = 6.37% = 5.90% + 0.47%, and the gain on the forward trade is enough to
pay for the purchase of the knock-out option in a scenario where the spot rise stops at precisely 1.2362.14
This raises the issue of where to place the barrier. This discussion involves the trade-off between the conviction of
a specific level being breached versus the cost of the premium. We will not delve into that here, but it can take on a
more concrete quantitative dimension if we first optimise the trigger versus cost, and then impose the trader’s
perspective. For the case in hand, suffice it to say that the PM needs to decide whether it may be better to place the
trigger: below 1.23, which would make the option cheaper, or place it higher than 1.23, and have a greater chance of
their leverage paying out.
The core of the argument is that, no matter what happens, the PM is satisfied that if they are right they will make
money. Additionally, this trade does not require any maintenance. The plan works on a buy-and-hold basis, unlike
our previous risk reversal example. What are we forgetting? Oh yeah, suppose they’re wrong? The worst-case
scenario is still a fall in the EURUSD exchange rate. They need to put on a stop-loss that reflects their risk appetite.
If the euro falls, how much pain are they willing to tolerate before getting stopped out? By the way, the other nice
thing about this trade is that if they get stopped out on the forward, the option still lives in case the euro reverses.
Lest you think we only discuss happy endings in this book, they were wrong. On January 22, 2015, Mario Draghi
delivered to the press an expansion of the asset purchase programme, including €60 billion of sovereign, agency and
European institutions’ bonds per month. The reaction of the market was very strong and the euro plummeted to
about 1.12.

Figure 16.18 shows the precipitous plunge in the euro after Draghi’s press conference, and the subsequent low
levels that it maintained during the following year. The forward would have expired in July 2015 at a market rate of
about 1.09, for a loss of about 6%. However, in all likelihood the PM would have stopped out before that level was
reached.

Leveraged spreads
As you may suspect, just like corporate treasurers, PMs hate paying premium, but unlike treasurers they love
leverage. One of the most beloved classic speculative trades is the leverage call spread (or leveraged put spread).
This is a trade that definitely requires dynamic maintenance, especially as spot nears the stop-loss level and because
it is a trade with a potentially unbounded loss.15 As usual, we will choose a historical setting, and in this case we
will have a PM with a strongly bearish market view on the Chinese renminbi on August 2015. For some context, the
renminbi was pegged at CNY8.25/US$ until 2003. Political chatter, and the market grapevine, then started to
forecast an imminent revaluation of the currency. This caused many hedge funds to take long Chinese yuan
positions, (short positions in USDCNY).16 These types of trades continued to lose money, however, as the renminbi
was not de-pegged. The NDF forward points became negative in the expectation of the revaluation.
Forward points for NDF contracts do not necessarily depend on any interest rate differential, and may solely
express the market expectation of future spot; this is true if there are no arbitrage opportunities between the onshore
and the offshore market. This is a very interesting case, which we discussed in Part I of this book, where the NDFs
were completely detached from the interest rate differential. The forward rates expressed through the NDFs were
based on supply and demand, whereas the onshore forward points were positive, since Chinese interest rates were
higher than US interest rates. More and more traders bought Chinese yuan forwards until the revaluation happened
in July 2005. But it was a meagre 2%. In the years that followed, the renminbi was officially floating and, generally
speaking, appreciated over time. This positive trend came to a stop for a couple of years during the GFC but then re-
started in 2010, and the renminbi continued to appreciate until its strongest level at the end of 2013.
The point to remember in this story is that the market was totally in love with the Chinese narrative, and the
storyline had the power to convince everybody. One of the most populous countries in the world opening its
financial markets and its consumer markets, ready to accept Western capital, with cheap labour and consumers with
increasing purchasing power. It was irresistible. As Montaigne wrote: “Nothing is so firmly believed as that which
we least know.”17 None of those hedge fund PMs in Greenwich knew much about China at that time. However, all
of them were bullish China and therefore, by a wrong logical inference, bullish the renminbi. It was called the “one-
way trade”, and it did not last forever. The economy started to suffer from too much credit creation, which caused
investors and savers to look for ways to take money out of the country. At the beginning of 2014, the renminbi
started to weaken, and this was followed in March by the widening of the +1%/–1% trading band to +2%/–2%, with
respect to the daily People’s Bank of China (PBOC) official fixing of the exchange rate. From then until August
2015, the fall was uneven but totalled close to 3% at the time. The last few months of 2015 exhibited crazy low
volatility. For example, in mid-July 2015 the three-month ATM implied vol was 1.5%. The official fixing was
decided by committee, so the low realised volatility was imposed to the extent possible by the authorities. Recall our
story about Bonny fighting with her risk manager about the low volatility of the renminbi. However, for implieds to
be that low means the market believed they would keep on doing this for the foreseeable future.

Figure 16.19 shows the history of the USDCNY exchange rate (this is an implied spot rate based on the NDF
quotes). A technically oriented PM may look at the chart and think that the time was ripe for another leg down in the
value of the renminbi, meaning that the USDCNY exchange rate would rise. A PM that’s comfortable with options,
however, would consider the following trade.
 Figure 16.20 shows a leveraged call spread in USDCNY traded on August 3, 2015, with a tenor of three months.
Note the bought strike is 6.1483, the ATM rate, on a notional of USD$10 million. The sold strike is the 27 delta,
equal to 6.2049, and its notional is US$20 million.
The PM buys a three-month ATM forward US dollar call Chinese yuan put on the offshore NDF exchange rate,
which fixes, for the purposes of calculating the final payout of options, on the official daily PBOC’s fixing. The
notional is US$10 million. This is a bearish trade on the renminbi. To finance this trade, the PM reasons that
although their view is bearish, they are not expecting a seismic move, but rather a moderate one. After all, this is a
currency exchange decided by a communist party committee. Since the PM is convinced that the upside is limited,
they are not too happy to pay this premium for potential moves they do not believe will happen. They are willing to
sell to the bank the far upside in USDCNY, represented here by a 26.93% delta US dollar call Chinese yuan put. The
resulting trade is called a 1×2 US dollar call spread. This is where they now make a stronger statement about their
conviction that the move will be limited. They sell twice the notional of the 26.93 delta US dollar call. The strike
was chosen in such a way that the total premium is close to zero. To recap, the ATMF call they bought cost
US$42,428, and selling twice the amount of the 26.93% delta earns US$42,428.
 What does this trade look like? Figure 16.21 shows the P&L of the leveraged spread: line ① is the P&L at
maturity. For low levels of spot, this line is calculated on the basis of mid-market rates; this is the reason why line
① is about US$10,000 above zero for low levels of spot, which tells us the magnitude of the bid/mid spread on the
transaction.
If they are completely wrong and the US dollar does not appreciate, or actually falls, they walk away with no gain
and no loss. If they are right, it can be something really interesting. In Figure 16.21, line ① is the payout at
expiration in three months, while the other lines are the market value at different points in time between inception
and the three-month date. As spot crosses 6.1483, which is the bought strike, line ① starts to ascend, since the
bought US dollar call will be ITM, but the sold US dollar calls will not be. The payout for that initial spot move
resembles a purchased US dollar call. At 6.2049 the payout is at the maximum possible level, which can be
calculated as US$10 million × (1 – 6.1483/6.2049) = US$91,218 or 0.9218%, a very nice payout for three months
work, annualised at 3.65%. Where’s the risk? The bad news comes north of 6.2049. The sold double-notional US
dollar call is ITM, and every extra basis point of rise in USDCNY returns one basis point of gain from the bought
call, but two basis points of loss from the sold double-notional US dollar call. This is why line ① starts to point
downwards after 6.2049. This continues without limit, and for that reason there is a point where the P&L will be
zero. This point, disregarding mid-market rates, is 2 × 6.2049 – 6.1483 = 6.2615. If spot rises above that level, then
the aggregate P&L of the structure is negative, and it becomes ever more negative as spot continues to rise.
The concept to appreciate with this trade is that, very often, PMs do not have simplistic views, such as: “The
renminbi will just fall.” Typically, market views have a timing component and a target rate. As we saw in the
previous examples, they may also dictate the path spot will take. In our case, the PM thinks that within three months
the target rate for the exchange rate will be 6.2049. The PM also thinks that it is relatively unlikely that the rate will
be higher than 6.2615. If they had not given 6.2615 much thought, then they definitely should have. The prospect of
an unlimited loss for very high levels of spot at expiry is completely unacceptable for a hedge fund, so there will be
a plan in place on how to deal with a scenario if spot rises above 6.2615. The plan may be to react in a very
instinctive way to stop the bleeding – just buy USDCNY forward when the market crosses 6.2615. This will
effectively move one-to-one as spot rises beyond that, but it gets ugly if spot reverses. If spot falls again and moves
towards 6.2049, the aggregate structure is not gaining anymore. The cure of being long USDCNY at 6.2615 is no
longer a cure but an ailment, and it would lose money. I’ve written many option tests for people I’ve taught options
to over the years – my favourite question is: “What is perfect hedge to a short option position?” The answer is
nothing. Just buy the option back.
Line ② in Figure 16.21 shows the price of the structure today. The price at the current spot level is zero, as it
should be, and for any higher spot level the value of the structure becomes negative. This means that the PM is ready
to accept a negative mark-to-market of the portfolio in the short term. However, if you look at that line ② you can
see that it is not very steep, and almost looks like it is close to being flat. This means that the current delta of the
structure is close to zero – or, stated differently, this structure is currently insensitive to the spot levels. That’s a big
deal in the sense that the trade is initially very low maintenance.
Line ③ is very interesting for another reason. It is higher than line ② for all instances. This means that, as time
goes by, for each given spot rate the structure gains in value, or in other words the structure is receiving time decay.
This is quite peculiar. A bought vanilla option’s value at any spot level will decline with the passage of time.18 In
this case, the opposite happens because the notional on the sold option is double the notional on the bought option,
so the sold option dominates and we have a constant gain as time goes by. It is as if the market were saying, “You
are taking a very large risk in selling that 6.2049 US dollar call, and if nothing happens I will pay you to take that
risk. If the rate rises a lot, however, you are going to suffer.”
So what happened? The structure was traded on August 4, 2015. On August 11, the PBOC, out of the blue,
devalued the renminbi. The exchange rate jumped from 6.2097 to 6.3246. A few days later, the central bank also
eased domestic interest rates, thus unleashing a wave of panic in the markets. At its September meeting, the Fed
declared that it had decided to not raise interest rates due to concerns about the Chinese economy, the first time in
memory that the Fed had cited a foreign economy as a factor in its decision.

Figure 16.22 shows in white the USDCNY exchange rate being devalued on August 11 2015. Line ① shows the
NASDAQ composite index falling and line ② shows the Shanghai Shenzhen CSI 300 equity index in the same
period. The horizontal line ③ is the strike on the bought 6.1483 US dollar call, the horizontal line ④ is the strike on
the double-notional sold US dollar call.
At 6.3246, the trade did not look good. Mark-to-market of the trade would have been a negative US$417,000,
which corresponds to a loss of 4.7%. When we talk about a 4.7% loss, this is very slippery ground. Recall our
discussion earlier in the chapter about how difficult it is to compare performance. In fact, the trade is zero-cost at
inception, so when we calculate the return what notional should we use? It might sound natural to use US$10
million, but one of the legs has a notional of US$20 million, so why 10 and not 20? The reality is that, in this
example, there is no mathematically sound definition of return on notional. Most of the hedge funds look at the
return on the risk deployed instead of return on notional. The risk can be calculated using VaR or volatility, or any
other measure, but it is important to realise that it is not easy to define risk. Many people will tell you that VaR is the
best measure, but just think of the fact that the definition of VaR depends on a choice of confidence interval
(usually, 5% or 1%), and this choice can be exploited by the PMs to enter into trades that would be much riskier if
examined under a different choice for measuring risk. At this point, we will not enter into a discussion that would be
more pertinent for a manual about the dark arts of risk management. This is an important issue for all the PMs trying
to enter into trades that are as large as possible, and equally important for all the risk officers trying to monitor and
limit the risk that their hedge fund assumes.
Let’s return to our initial emotional reaction to the central bank effectively devaluing the currency. Ouch! For a
trade that cost nothing to put on and earned theta every day, that stinks. If left unattended, at expiry the trade would
have been worth negative US$141,000, for a still sizeable loss of 1.41%, with an expiry spot reference of 6.3435 –
not that different from the devaluation rate of 6.3246. So how come the trade recouped some of the losses and went
from –4.7% to –1.41%? Most of the negative 4.7% value came from the sizeable time value in the double-notional
sold US dollar call. Volatility for that strike on August 12 had soared to 7.7% on the offer. Recall from Part II that,
as volatility increases for OTM or ITM options, the values do not just rise in a linear fashion. From August 11 until
expiration, the spot rate did not really move, effectively causing the sold US dollar call to bleed away the time value,
which was good news for the PM’s P&L. However, one wonders: would the PM have been composed enough and
had enough risk rope left so that after a mark-to-market hit like that, they would not have bailed and got out of the
short option at those ridiculous levels?
This trade was not a huge loss, but does represents a fairly typical example where the market view was generally
correct but circumstances and details can derail the best ideas. The need to leverage the trade, or in this case to
finance the premium, forced too tight a range in a situation where a committee meeting in Beijing can put the rate
anywhere they want. Like we saw with EURCHF in the previous chapter, this is a good example where delta
hedging and stop-loss levels are useless. The damage to the trade value was done on August 11 with the devaluation.
This caused the spot rate to gap, and as we know it’s all over before there is a chance to execute a delta hedge. The
rate jumped in a split second, when the official fixing was published.
The PM will undoubtedly ponder: why did I set the sold strike so low? The reason is that the delta on the sold
strike was about 27%, which is actually a reasonably low delta, but the real reason is that the implied volatility was
unrealistically low. The PM was forced to sell a US dollar call with a strike relatively close to the current market.
Put another way, the strike on the sold US dollar call was relatively far from the current market for a currency pair
with a volatility around 2%, which was the case on August 3 when the trade was executed. The volatility was so low
because USDCNY was a “managed peg”, or in other words a currency pair where the rate is not free to float and
determined by market forces – rather, it is decided every day by a committee. Despite what the PM was expecting,
on August 11, everything changed. On that day the exchange rate moved by about 2%, which gives an annualised
realised volatility of about 32% (2% multiplied by the square root of 255). So, in essence the PM sold for 3% an
option on a currency pair that for a day traded at an annualised volatility rate of 32%! Another way to think about
this trade as it pertains to situations like Chinese yuan, or even the EURCHF case, is that central authorities
sometimes have the capacity to put rates where they want instantaneously. Under those conditions, it does not make
sense to price options as if the price of the option is based on a normally distributed set of expectations. It is not.
You’re better off speculating with discrete payouts you can control.
This example on USDCNY is also very illustrative in terms of central bank power. It would seem, after hearing
the story, that the central bank, the PBOC, held absolute power and could determine the fate of all investors. The
reality is slightly different. On the one hand, yes, PBOC’s power is huge. Not only do they set the fixing every day,
but in addition each and every USDCNY trade done onshore has by law to be booked through the state trading
system, called CFETS. Each day, after the fixing is published, trades can only be executed if they are within –
2%/+2% from the fixing. Before March 2014, the window was –1%/+1%. Therefore, the authorities can completely
control the market as they can refuse to book any trade executed at a rate that they do not like. Local banks would
refuse to settle a transaction that is for all intent and purposes a rogue one. In addition, as we all know China has
accumulated over the years a huge reserve of dollars, which the central bank can use to intervene in the markets and
defend the renminbi.
 Figure 16.23 shows the amount of US dollar foreign reserves held by the PBOC up to the time of our USDCNY
trade under consideration. The latest data point reads US$3.65 trillion. Part of those reserves is invested in US
Treasuries. Foreign exchange reserves are a very effective defence against local currency weakness.
Figure 16.24 shows the amount of US Treasuries held by foreign central banks over time. This is a snapshot at the
time of our USDCNY sample trade, in August 2015. China and Japan are the largest holders. The aggregate amount
had been rising for about 10 years, although later, at the time of the money market funds reform in October 2016,
this amount fell for a while (see our treatment of the cross-currency basis for more details on the money market
funds reform).
It may appear that, for the central bank, the control of the exchange rate must have been an easy game to play. Not
so. This is one of the most fascinating aspects of the FX market, which puts it in a special category apart from all the
other financial markets. Foreign exchange is intimately linked with trade flows and the geography of capital. These
are not about Ivy League educated hedge fund PMs in Greenwich – they are about a much broader swathe of
society, such as owners of small firms, import/export businesses, people who have saved money for their entire life.
If they are unsure about the local Chinese financial system, or if they are worried for any other reason about their
money, it will be very difficult to stop them from getting money out of the country. The central bank will have a
very hard time blocking every little scheme that individuals can devise. This is not unique to China. Ask the central
banks of Russia in the 1980s, of Venezuela, Brazil, Argentina, or any country in the world about how to stop capital
flows, and the answer is that it’s not so easy.
 With all its power to fix the exchange rate, PBOC still could not ignore supply and demand, and in fact, by and
large, their fixing did not deviate from reality all that much. If it did, people would play games and find backdoor
tricks to get the money out. They would send truckloads of flat screen televisions as exports to Hong Kong, collect
export subsidies and turn the truck around and re-import the televisions to the mainland. They would import
shiploads of copper, nominally for industrial uses, backed by letters of credit to collect higher-yielding yuan-
denominated deposits at about 5% at a time when US dollar interest rates were close to zero. They would hire
foreign consultants and over-invoice their services to get money out of the country. They would buy bitcoin with
renminbi, transfer them to Canada into their cousin’s bitcoin wallet, and ask him to sell them for Canadian dollars.
They would buy condos in Vancouver and push its real estate market to astronomical levels where even senior video
games programmers could no longer afford to live in the city. This is the beauty of the foreign exchange market –
FX impacts the real economy and, since people are very smart, all the regulations in the world are not enough to stop
them from taking money out.

PORTFOLIO VIEW
We have discussed investment strategies for a single currency pair. Recall from the sections on corporates that there
is also a great deal of theory on the portfolio approach. If you recall when we examined the difference between
hedging and speculating, we pointed out that a major difference is that hedging is a problem with a solution. The
hedger expresses an objective, which could be to reduce the volatility of earnings year-on-year or maybe to defend a
budget rate. Whatever the objective, it is very narrowly defined, and can sometimes be expressed by a formula. We
will now bridge the two by setting clear objectives that can be ranked and measured with a formula so that, within
the context of the investable universe we’ve defined, we can arrive at an optimal solution.
For every FX position one can compute its expected Sharpe ratio as the ratio of its carry to its volatility. The carry
plays the role of the expected return, and its implied volatility plays the role of the realised volatility. A PM could
just select the currency that offers the best Sharpe ratio, but then they would be putting all their eggs in one basket.
By diversifying, the volatility can decrease substantially. In fact, and this is the magic of the numbers, the
correlation, or better the possibly negative correlation among different FX positions, can make the aggregate
volatility of an FX basket substantially lower.
Let’s examine the question of an investor who can invest US$1.00 in a portfolio of pre-selected currencies,
potentially shorting some of them, in a way that the aggregate US dollar notional invested is US$1.00. Suppose the
investor is targeting a specific carry level, and is wondering what specific choice of basket weights minimises the
volatility of the resulting basket. We can answer that question:
let C be the carry that is targeted;
let σi be the volatilities of the portfolio components;
let γi be the carry of currency i with respect to the US dollar; if the rate in currency i is higher than in US dollars,
then γi will be positive;
let wi be the notionals of each portfolio exposure; and
let ρij be the correlation between the i-th and the j-th component of the portfolio.

If we express wN as a function of the other wi’s due to the fact that the sum of the weights must equal 100%, we
have (Let α = γΝ – 1 – γΝ):

The condition on the carry C can be expressed as:

And, therefore, we can also solve for wN-1; when we solve for both wN and wN-1, we get:

In other words, we are parametrising all the admissible portfolios using only N–2 independent weights, because we
have N unknowns and two conditions: the carry and the aggregate notional. Then, we can write the variance of the
portfolio as:
Now we want to impose a minimum point for the variance, and therefore require that for each weighting the
derivative of the variance with respect to that weighting shall be zero:
The solution of this linear system of equations gives the weights on the portfolio. Table 16.1 is an example with
such an optimisation in the case of a portfolio of the G10 currencies (excluding the US dollar, which will act as
functional currency).
We obtain the series of optimised portfolios shown in Table 16.3 for each choice of carry from –2% to 8.50%.
In Figure 16.25, for every possible basket position that the PM might choose, the carry is on the horizontal axis
and the implied volatility of the basket on the vertical axis. This is a very similar set-up as the ones we used for
portfolio theory in the context of corporate hedging. For the highest Sharpe ratio, the PM would like the portfolio to
lie as much to the right and as much towards the bottom as possible. In white dots we show the portfolio constructed
by investing 100% of the notional on a single currency. In black is the optimal portfolios calculated and shown in
Table 16.3. Note that the black portfolios represent a very substantial improvement vis-à-vis the original set of the
white portfolios. As the carry becomes negative, the weight on the euro and Swiss franc (which both have negative
carry) increases.
A STRATEGY FOR EVERY VIEW
We’ve looked at the thought processes and the numbers that drive speculators. For the remainder of this chapter, we
want to be sure you have a brief overview of a few of the basic – and not so basic – option tools that speculators use
to express their views. We will list them and briefly comment on their characteristics. However, this presentation
will not be comprehensive and nor will we go very deeply into the uses and abuses.
Basic speculative structures

Spreads
There can be call spreads and put spreads. This is a simpler less risky case of the Chinese yuan leveraged spread
discussed earlier in the chapter.

Long a call spread: Buy call with strike “A” and sell call with strike “B”.
This strategy can be used to hedge, but it provides only a limited hedge for a short position. Protection is in
place from strike “A”, with the limit being set by the short strike “B”.
It is mostly used by speculators, to reduce the cost of an outright purchase of an option by selling a strike
which the speculator thinks will be the high of the spot move.
This is a great strategy for currencies that have a high skew. This means that OTM options have a
significantly higher volatility than the ATM strike, so you buy the “A” strike at a relatively low volatility
level and sell the “B” strike at a much higher volatility.
This strategy is suitable for a mildly bullish view on the market. It gives 100% participation in favourable
underlying moves between the two strikes.
Compare the two call spreads in Figures 16.26 and 16.27. The first strategy depicted above is a 1.19/1.23 euro call
spread that costs 100 euro points, and if the euro rises to 1.23 it will give a payout of 400bp. The second strategy
depicted above is a 1.21/1.25 euro call spread that costs 40 euro points, and if the euro rises to 1.25 it will give a
payout of 400bp. This is a strategy that, when held to expiration, will have a predictable outcome based on where
spot ends up.
Is the second strategy clearly superior? When initiating the trade, many speculators look at this like a gambling
bet. The first spread returns 5:1, while the second bet returns 10:1. Stated differently, for the same 400bp in payout,
in the former case I have to pay 100bp and in the latter I only have to pay 40bp. In both cases, I think the euro is
going up. This is faulty logic. Great odds can be created by moving the strikes away from the forward. However, the
further away one moves from the forward the lower the probability of earning anything and the higher the
probability that the premium will be lost. The 1.19 strike is already ITM and the strategy is a safer bet, which is why
the premium is higher. Stated more precisely, in the case of the 1.19/1.23 with the forward at 1.20, the probability
that this strategy will be ITM is something greater than 50%. In the second case, it is definitely something less than
50%.
The second comment about spreads is how they decay. The punchline is that if spot moves quickly ITM with lots
of time left, owning this spread will be very disappointing. It takes a long time to realise the promised loot. Stated
differently, one would want to keep the term of spreads closer to the short term than the long term. In addition to
spreads where options expire at the same time, there are spreads where the legs expire at different times, allowing
the trader to express a timing and maybe a volatility view. Those are called calendar spreads.

The straddle
Long position: Buy a call and buy a put with strike typically ATMF.
However, the strike is often shifted slightly to make the straddle zero delta.19 The practical reason is obvious,
it saves one additional tiny trade. Recall the purchase of the call, which requires that you sell the underlying
and a purchase of the put, which requires that you buy the underlying. Therefore, the two deltas cancel out.
For a short position, the trader would sell both.
Buying a straddle is done when one expects implied volatility to increase. This is the most impactful way to
take that position because, as you may remember from Part 2, ATMF options have the highest vega or
sensitivity to volatility.
This is also a very expensive strategy! The owner needs to be right and quickly. Going long a straddle is often a
silly strategy in some ways. The owner expects volatility to rise around these levels. However, by definition if
 we are around these levels it’s unlikely volatility will rise. Then, some traders will say: “I know it will move, I
just don’t know which way. With a straddle I will make money when it moves either way.” In theory, that’s
correct but the cost is so high that it has to move a very long way just to break even. If you really have no idea
which way it will move, maybe you should take the day off from trading and wait until you are better equipped
to make a decision.

Let’s expand on the speculative effectiveness of a straddle. Buying a straddle because the view is that spot will
move is all good. Buying a straddle because implied volatility will rise is a different story altogether. Suppose that
implied volatility indeed rises, so that the straddle value increases and you are celebrating. This is very unlikely to
happen without spot moving. Assume spot moved so that now the straddle strike is no longer ATM, making the
structure long or short the exchange rate. In other words, now there is an implied volatility long position, but also a
long or short currency position. If the trader still maintains that their interest is in implied volatility and not the
exchange rate, then they should get rid of that exchange rate position by delta hedging. This is easy to do. As the
trader delta hedges, they are isolating the P&L, which will change only as a consequence of changes in the implied
volatility. This is great, and exactly the original objective.
However, now there’s another problem – suppose spot moved higher and is sitting far away from the straddle
strike. Now the problem is that the vega of the straddle has decreased. If implied volatility rises the straddle value
rises again, but in a much slower fashion than when the strike was ATM. The structure is losing its exposure to
volatility. This is an issue and there are various ways to fix it, but none are perfect. One typical fix is to sell the
straddle and buy a new one, which would be zero delta and therefore very high vega. Great solution, but it will
require you to incur transaction costs. There are instruments that would give a trader exposure to pure volatility
equal to buying a straddle and the instrument would maintain the vega until expiration. Those instruments are
volatility and variance swaps and they are beyond the scope of this book.
 Finally, selling straddles when the market has gotten very excited and looks like it will calm down makes sense.
It’s consistent logic – if nothing happens soon, vols will fall and the impact will be powerful. Even if the trader is
slightly wrong, it will have to move a long way for the position to really hurt.

The strangle
Long position: Buy a call and buy a put.

The simultaneous purchase of a call and a put with different strikes set equal distance apart around the
forward rate.

This strategy is like the straddle, with similar greeks. Therefore, the strategy benefits from a rise in implied
volatility as with the straddle, but to a lesser extent.

There is a lower premium outlay/cheaper than purchasing a straddle, but the break-even levels are even
further away.

It can be used when a rise in volatility/range break is expected, but with less confidence than in the case of
using a straddle.

The butterfly
Long: Buy a straddle and sell a strangle.
The best use of a straddle and a strangle is when you combine them into a butterfly (see Figure 16.29).
Butterflies can also be constructed with four legs in a straight combination as mentioned above, or they can be
constructed in three legs of the same option with different strikes.
The greeks are nuanced depending on where spot is.

We’ve discussed this strategy enough, but just for completeness let’s look at the risk reversal.

The risk reversal

The simultaneous purchase of a call and sale of a put with different strikes set equal distance apart around the
forward rate.
First-generation exotic options
We now want to introduce some of the more exotic structures that are common in the market, starting with the
characteristics of trigger options. For an in-depth discussion and fun stories of how they are used, you’ll have to wait
for our next book! These are the basic tools of exotic structures.
To begin, we will define the two types of triggers (barriers) that we refer to in the rest of the section. Some of
these should already be familiar to our readers.

Discrete barriers
While normal barrier options can be triggered at any time during the option’s life, a discrete barrier can only be
triggered against a fixing – for example, a central bank fixing. The advantage is that trigger events are less subject to
intraday volatile market movements.
It is an all-time classic that clients and salespeople will realise placing triggers reduces cost, so they naturally add
a trigger without thinking it through. This is clearly in the category of “a little knowledge is dangerous”. So we’ve
had over the years speculators asking us for a call option with a strike placed above the current market level and
with a discrete knock-out placed below the current market level. The knock-out is only monitored at the expiry of
the option. If at expiry the spot is below the knock-out level, then the option disappears, otherwise the option is the
same as a vanilla call option with the same strike. Think about this for a minute before you read on. When asking to
price this “exotic” option, they expect the price to be lower than the price of the corresponding vanilla call option
with the same strike. You’ve probably already realised that this knock-out option is identical to the vanilla call
option with the same strike. The reason is that there is no scenario where the knock-out option disappears but the
corresponding vanilla call option with the same strike is ITM. Now you know what passes as “inside the beltway”
humour at structuring desks.

Fade-in barriers
A fade-in barrier option is like a normal barrier, but is triggered on a day-by-day and pro rata basis. It is a strip of
consecutive discrete window barriers, each having a one-day window with a fixing source as a reference. Each
passing day, a fraction of the total notional is triggered into or out of existence. Fade-in barriers are similar to fade-
out options in that they end up with the same notional when given the mirrored trigger conditions.

The common structures are as follows

Knock-out options:
A regular knock-out is an option that disappears if spot trades at a predetermined knock-out trigger, which is
set OTM with respect to spot at inception.
The knock-out is cheaper than the plain vanilla equivalent since there is a chance the option will disappear.
They are highly dependent on the trigger level and implied volatility.
They are usually bought by speculators to reduce the upfront premium.
Where to place the triggers is an art. At one level of sophistication, they are placed beyond technical
support/resistance levels. As mentioned earlier, one can mathematically work out the most efficient level
given cost, volatility and objectives.
Of course, one can always sell a knock-out option if spot is expected to trade at the trigger level.
Knock-in options:
A regular knock-in is an option that only exists/appears if spot trades at a predetermined trigger, which is set
OTM with respect to spot at inception.

The knock-in is cheaper than the plain vanilla equivalent since there is a chance it will not exist. As with the
knock-out, it is dependent on trigger level, direction and volatility.
Speculators would buy a knock-in if they believe the market will move in the opposite direction from the final
and more significant move.

An example will clarify. Suppose spot is 100 and the trader wants to buy a call struck at 105 that knocks-in at 95. If
spot never trades at or below 95, then the option does not exist. On the other hand, the moment spot trades at 95 then
the option exists and becomes identical to a 105 vanilla call. To make money from this option, spot needs to go
down from 100 to 95 and then back up to at least 105. Depending on the tenor of the option, this might be more or
less likely. Generally speaking, those options will be a fraction of the price of a vanilla option.
Double one-touch:
A fixed, known payout is received if spot trades at/or beyond either of two predetermined triggers at any time
before expiry.
Triggers are set above and below the spot level at inception.
This option is bought on the expectation that implied volatility will rise, or there will be a breakout of pre-
specified range.
This option is sold if volatility is expected to fall and spot is expected to stabilise.
It is less risky than selling vanilla options such as strangles because the maximum loss is capped if the triggers
are hit.

Finding the right speculative strategy

By just using the above structures in creative ways, tens of different hedging and speculative products can be created
to reflect very intricate risk profiles and market views. For practical purposes, to help make sense of this clumsy
cookbook of diverse structures, and mostly for the fun of it, we show here a way to find a speculative strategy that
matches some of the main views across spot, volatility and risk profile. In case you take the following listing too
seriously, the global head of structuring at a large European bank used to admit to poking fun at interviewees by
asking them what a structurer would recommend if the client is bullish. Most of the candidates would shower him
with an esoteric combination of options, whereas the simplest answer is just to buy the underlying outright.
 Also, keep in mind that most trainees or juniors on any trading desk sooner or later are given the project of
compiling the infamous “cookbook”. They have to ask the other members of the team for structures and variations
of structures. The book is typically formatted with one page per structure. Although in the end most structures across
all bank structuring desks will be identical or very similar, every bank gives its version a different “cool”, or
intuitive, name in an effort to create a marketing edge. In the end, the name is made up on the fly, devoid of
descriptive power and forgotten in 90 seconds. Ignoring this backstory, buy-side clients will then call another bank
or pricing data provider and name the structure they have bought as if this made-up name were known to everybody.
They are usually surprised to realise they are victims of marketing and the human attraction to name-dropping.
In the list displayed on figures 16.30–32, we throw in some of the names we used at our respective banks over the
years. This provides a partial summary of these structures.

1 Most of the time this type of trade is actually not done in equal amounts, but by using various more sophisticated measures. One example could be to
enter the two trades so that they have the same VaR instead of the same notional.
2 Mostly due to their high savings rate and low domestic interest rates. This makes it attractive to invest savings pretty much anywhere else
internationally.
3 We have measured the standard deviation of the distribution of risk reversal values over some historical time period. Note that, as the risk reversal can
assume both positive and negative values, it is not a lognormal distribution, so we have not taken the distribution of the returns of the risk reversal but
just the distribution of the values.
4 95% VaR corresponds to a move of about 1.64 standard deviations.
5 All notionals are calculated with a time horizon of three months, and all volatilities are annualised.
6 Assuming that implied volatility equals the instantaneous volatility at each instant, and that there are no trading costs and that the premium of the
structure was originally zero. A bit of a simplification, but part of our point here is that things become immensely complicated as soon as the position
is dynamically delta hedged.
7 Our example includes a number of simplifications. On April 6, the spot rate was 85.54, which means the calculation of target P&L for the trade is no
longer correct. Christine would make US$30,800, if the risk reversal moved to her target level, but our calculation kept the spot unchanged. At 85.54,
 this is not true. Nevertheless, assuming Christine delta hedged and vega hedged the trade continuously, her final P&L due to change in spot would be
zero, due to changes in the level of ATM implied volatility would be zero, and therefore her P&L could only be due to various second-order effects,
of which the risk reversal is the largest. Other effects will be due to the butterfly, time decay, change in interest rates, and so on.
8 Every trader knows the old adage: if you don’t know who the sucker is at the poker table, it means you are the sucker.
9 Consult the book by Shankar Vedantam in the References.
10 B. Nosek, 2015, “Estimating the Reproducibility of Psychological Science”, Science, 349(6251), aac4716, August 28, 2015.
11 We are assuming that spot did not move, which is obviously a very big assumption, otherwise it would be very easy to make money by selling all sorts
of options all the time.
12 A one-touch is exactly what it sounds like. If at any point during the life of the contract spot trades at that level, a payout is due and the option expires.
There are many variations, but suffice to say this is the closest thing to a bet you’d make at a casino that you’ll find in the options market. Pay a
premium X and, if the level is touched, the bet returns some multiple of X.
13 A vanilla option is an asset: it cannot generate a future loss. On the other hand, a forward trade carries a risk as it can become a liability if the market
moves against you. The credit risk in a vanilla option trade is limited to the risk that the option buyer does not pay the premium on the option. As the
premium is usually due in two days from the trade date, this is a two-day credit risk, which is minimal.
14 In this scenario, it is also true that one would be better off not buying an option that eventually knocked-out. Our point is that that the damage from the
option being knocked-out is not as bad as one might think.
15 They are not unbounded in all cases, depending on the notional currency.
16 The positions were mostly traded via NDF forwards. At the time, the currency was completely non-convertible and non-deliverable, so US-based hedge
funds could only trade NDFs.
17 M. de Montagne, 1595, Essays, Book 1, Chapter 31.
18 In most instances, apart from pathological cases with very high interest rate differential and deeply ITM options.
19 The details here are not straightforward. You might argue that a straddle with a strike equal to the forward rate is always (roughly) zero delta.
Conventions, however, make this statement tricky. If you are a German trader receiving the premium of a straddle in US dollars, then you are actually
long US dollars, not flat. Therefore, the market convention will be to move the strike on the straddle to make the German trader short US dollars in
the same amount as the premium they will receive when they sell the straddle. Therefore, just by magic at the time of the trade, they will be at zero
delta, even when the US dollar-denominated premium is taken into account.
 Epilogue

We have come to the end of our journey, moving from the very simple trades that exchange one currency for another
to the TARN, which is sophisticated enough to take up many pages to explain and many more to understand. This is
where a trade ceases to be a trade and becomes a riddle.
The book you have just finished could have been two different books – a technical book with all the quantitative
FX details, or a book of anecdotes, greed, suffering, luck, tears and laughter. We’ve tried to give you the most
relevant aspects of both. The story of foreign exchange is not a story of formulas, instruments and rules, but instead
one of surprising events that impact almost everyone – from the average global citizen looking to go on vacation in a
different country, to the huge corporation, to the banker, to governments and to international institutions. Very few
financial instruments have such a universal and direct impact. We’ve approached the mathematics and technical
details with the respect they deserve, but always with an eye to democratising the concepts and, where possible, the
numbers. To this end, we have discussed every important concept from a few different perspectives, looking to put it
in a realistic context so that we can show you the human impact.
The main reason we felt compelled to write this book is that, after almost half a century of “free” floating
currencies and a static way of doing things, we believe the entire system is changing in ways that will make it
unrecognisable in just a few years. Again, we could have told the story of the FX market transformation in terms of
how regulation, electronic trading, algos and blockchain will completely alter the technical and quantitative
thinking, but what fun is that? Besides, we would have missed the best part: the human element. The human element
is unique in FX. What made the FX and FX options markets really special is the Wild West feel. This attracted a
different breed of participant than, say, your corporate bond or equity dealer. The kid from the local shop that was
good with numbers got a chance to make more money than a Harvard grad with connections. It also opened the door
to a new breed of macho trader that did not view institutions as sacrosanct, and so would regularly take on and
defeat governments and central banks. The guy with the really long name that would never be hired into investment
banking had an advantage because he knew how to use Lotus 123 and was not intimidated by advanced statistical
methods.
As a handful of banks consolidated all the volume and most of the profits, a lucrative business evolved into the
most unregulated, largest, freest trading environment of any asset class. Participants had complete freedom to dream
up any risk they liked and create the most fun structures ever created. This in turn attracted a bunch of
mathematically inclined engineers that realised all the mathematics they had learned could be used not to create
physical structures, but financial structures that gave them the advantage and allowed them to earn unimaginable
bonuses. As mentioned in Part I, the epicentre of dominance (and compensation) on the sell-side had moved from
spot trading to the sales desk, and now to the electronic execution desks. Hedge funds and the prop trading desks at
banks, while they were allowed to exist, always held sway as the top dogs. They executed the coolest, bravest trades.
The interbank brokers1 did unimaginably well – they were in the middle of every trade, taking a tiny cut from every
deal. The people that made all this possible, the corporate treasurers and the asset managers, never managed to enter
the inner profitability sanctum. They never earned as much nor got the glory, and were generally content to feast off
the crumbs. Banks eventually had to address this situation as well.2

The human element

The old, classic FX market allowed for an intimacy of relations and interactions that in the end led to its demise. All
the FX traders that are accused of wrongdoing, in the current regulatory and cultural settings, are baffled. How could
the rough and tumble rules, the way of sharing information, and the way business was conducted for decades,
become illegal and immoral overnight and, worse yet, retroactively! It was commonsense that salespeople wanted to
execute the biggest hedge fund deals, and then use that information to figure out which way the exchange rate would
move. No one thought of that as insider information. It was an edge that they earned by establishing a relationship
with that hedge fund manager. To add insult to injury, the institutions that promoted and condoned this activity are
now throwing these people to the regulatory wolves.3
We have tried to give you a palpable sense of all these characters, the relationships and pressures that drove
behaviour. We also played amatuer behavioural psychologists, and took on some sacred cows, including: success in
finance is based on meritocracy; a person fails or succeeds on their own; and the idea that anyone with raw talent
can become a successful trader, given the opportunity. Finally, we claim that there is no such thing as trading talent,
only the ability to find an edge. This single person’s, or firm’s, edge may not be obvious even to the person
exercising it. It only appears to the casual observer as talent and smarts until the broader community realises that it
may simply be an information edge or a latency edge, or the short-lived discovery of a successful pattern. The story
of finance is a story of looking at the rules, and looking again, and carrying on looking until you find a trick, loop,
edge or paradox. Once one is uncovered and acknowledged, the broader society tries to eliminate it to create a “fair
playing field”. Below is our grand finale, with all the major themes coming back to take a final bow.

Finance is a meritocracy and talented people will be successful. This idea is appealing to our human nature. It is
easier to accept in finance because we assume it is very easy to measure the contribution of an employee in
terms of dollars. Whether it’s number of sales or the trading book P&L, it is only natural to conclude that a
quantitative business is more likely to be meritocratic. In actuality, the quantitative nature of the job will make it
much easier to associate a number to each employee’s yearly performance, but at the same time will make it
easy for smart people to game the system. For some very entertaining analysis, compare the evaluation
methodologies at a bank with those used by college faculty.
To be successful in finance you need teamwork. Over the years, management has tried to design systems that truly
reward teamwork. However, teamwork has turned out to be a very hard nut to crack. The myth is that one gifted
individual salesperson, or trader or visionary manager, can do it all by themselves. We have not found that to be
the case. Without the backing of significant capital, a really excellent back-office confirmation clerk, or a
diligent assistant, any trader or salesperson will have a hard time succeeding. One needs to be a team player.
The truth is, however, compensation and year-end discussions are conducted in exactly the opposite spirit. All
major banks (some famously so) use a normal distribution to dole out bonuses, with a disproportionate share
going to the right tail. The bottom 10% usually gets fired. This does not promote collaboration, but unhealthy
competition. Frankly, the two concepts discussed, of meritocracy and teamwork, are at odds. One can either
create a true meritocracy or reward teamwork – you cannot do both. The moment you measure people’s
individual contribution, and proclaim one performance superior to another by the reward you provide, you are
giving a powerful incentive not to help others. Think of the case where two people are equally valued but, by
definition, one has to be ranked higher. Then, by decree one person has to get paid more. What’s that second
employee going to do next year? They will not work closely with the person that got more money for doing the
same work. Compensation is not disclosed publicly but the pecking order is visible to all. Very few
organisations have been able to solve this paradox effectively. In real life being a team player is absolutely
essential, and will result in a happier career. However, it will not speed up your promotion or your ascent, it will
actually slow it down because no mathematical formulas can calculate teamwork.
This takes us to one of the central themes of this book, the expert problem. As a general rule, results of human
endeavour improve with study and practice. There is even a back-of-the-envelope rule that to really get good at
anything you need 10,000 hours of practice. Obviously, for a brain operation you will surely end up in better
shape if a brain surgeon operates on you instead of a plumber. However, to fix your bathroom you certainly end
up in better shape if the plumber fixes it instead of the brain surgeon. There is only one area of human activity
where this principle does not hold – if you need financial advice on your investments, there is no proof that a
financial professional, corporate treasurer, hedge fund PM or market-maker trader will be any better than your
Uber driver. We hope we have given you plenty of reasons throughout the book to understand why this is.
The expert problem brings up another theme, the role of luck. It is easy to believe that luck does not really play a
large role in a career in finance, because of the law of large numbers. The reasoning is that traders make many
bets during their careers, and bad luck will be averaged out so that the good traders will always rise to the top
and be recognised for their true talent. Well, an objective scorecard for portfolio managers does not really exist.
The key tool for a portfolio manager career is the track record, which is the average yearly return on the
investment managed. This is prominently displayed on every resume and is the key factor for hiring portfolio
managers. Yet, no practical way exists to check those numbers.
If a trader says they made an average return of 15% over the last 10 years by working at three different
hedge funds, there is no way to check those numbers as the information is not public, and is not calculated by
the hedge funds, and there is no generally accepted calculation method. Besides, it is not true that over a large
number of trades bad luck washes out. Think about those portfolio managers who had a position right before a
bubble burst versus those who closed the position the day before and are now heroes and have a successful
career. Those who were planning to hold the position for one more week blew up and lost their job, and have
since moved on to other careers. The difference between hero and idiot is very thin.
Compensation. Since the global financial crisis and the witch-hunt of financiers that followed, the industry has done
a lot of soul-searching about how much an employee should be paid. The original school of thought was that if
 we can measure the revenues that different employees produce and one employee produces 10 times more than a
second employee, then the former should be paid 10 times more. This argument does not consider that anybody,
even a monkey, might be able to produce a certain level of revenues when trading a large, liquid, client-driven
trading book, or as a salesperson covering large clients that would produce a minimum revenue amount under
any conditions. The other issue that arose is that, in some cases, an algorithm could produce twice the revenue
that a human could. Should that person then be paid less, or even be fired? In finance, these issues are front and
centre because ability cannot be taught and is indistinguishable from luck (see the expert problem above). This
has created an environment where there is no formal training, and people are left to their own devices to learn
the job and decide how to go about doing it, while management worries about hiring “talent”. In very few
industries, and this even excludes acting, is training as conspicuously absent as in finance.
Finding an edge. As mentioned, the Rothschilds complained about the use of the telegraph to transmit prices from
Philadelphia to New York since it made it “impossible” to make money in the markets. They were unhappy
about the elimination of an unfair advantage in information. For them, making money in finance meant by
definition having an “edge”, having a system, exploiting a loophole. They never were ashamed of that. Today,
an unprecedented amount of regulatory effort has been produced to eliminate unfair advantage and create a
playing field for all investors.
We do not believe there is such a thing as pure financial talent, and with a completely even field nobody
would be able to consistently beat the market except for the large role played by chance. To say things in a
different way: it is our personal opinion that if financial markets were a completely even playing field, nobody
would be able to consistently outperform, but you would still have rock star portfolio managers with stellar track
records because of the large role that luck plays in building trading records.
Algorithms have profoundly changed the way we look at trading talent. There are two main skills that
computers have been very good at: speed and cognitive bias. Speed is the easiest to explain and exploit. If you
think about the time when, for example, US non-farm payroll numbers are released, it is very clear to everybody
how a computer program can easily read the data release, decide whether it is good or bad news for the US
economy and the US dollar, and initiate trades that try to take a position ahead of everybody else. Clearly,
humans cannot act as fast as computers, and therefore any trader whose talent resided in trading data releases is
now squarely beaten by thousands of computer programs around the world that can trade faster than they can.
Another important edge concerns traders whose trading strategy consists in exploiting cognitive biases in
other human traders. As Warren Buffett famously said, “buy when everybody is selling”. Investors have been
making money for ages by exploiting crowd mentality, and being contrarian when the market gets carried out
and positioning becomes crowded and prices become over- (or under-) inflated. This trading approach, in its
simplest form, has basically just required an uncommon degree of rationality in the presence of specific market
situations, but nothing more. Of course, there are many more forms of cognitive bias that human traders can
exploit, our present example being just the simplest.
In the literature and research papers, studies have revealed time and time again that people are systematically
very bad at getting rid of the trading errors induced by cognitive bias, even after those errors are explained to
them. If an investor is able to force upon themself the psychological discipline needed to avoid those mistakes,
then the money is theirs for the taking. But, again, it is very easy to imagine computer programs able to exploit
cognitive bias much more efficiently than humans – for the simple reason that computers don’t suffer from it in
the first place, whereas human traders have to unlearn their gut reactions.
Regulation. This brings us to another of the key points that you have read about throughout this book, which is the
presence of regulation. As we have just said, some of the regulation that is flooding the system is trying to create
a level playing field where it is impossible for any market player to exploit an advantage and make money off
the other players. Simple examples of this effort are famous convictions of portfolio managers who were able to
procure non-public material information and trade off it, or traders that conflated their desire for personal gain
with their fiduciary responsibility as client advisors. The financial system might seem at times similar to those
computer operating systems that have existed for a very long time, and which are backwards-compatible. This is
a great advantage as it allows companies to keep the same software and databases, and still update to the latest
and greatest features. Yet, as the system is updated over the years, the burden of legacy code becomes a tangle
of different ways to get things done using layers that have been built using different tools and philosophies. At
some point, it becomes almost impossible to introduce new features without making the system ridiculously
complicated and clumsy. The pace of innovation slows down, and there is a tipping point where it becomes
more practical and economical to build a new system from scratch rather than continuing to patch up the old
system. However, human paradoxes of decision-making prevent it. The short-term pain of migrating to a new
system, and potentially losing large numbers of users in the process, is too big and prevails on the long-term
 gains of having a better system. The community plods along and progress stops. The same thing is happening to
finance, where every new piece of groundbreaking regulation imposes enormous compliance efforts for no
increased revenues, no increased speed, no better markets.

The maths
The people stories are the fun part. However, one needs to understand the maths to understand the motivations. In
finance, the technical details constitute a logical and rational structure, upon which funny anecdotes and stories get
tangled like flies on a spider’s web. They represent the singularities, the mistakes, the unexpected. We wanted to
show that the rules and the mathematics provide the scaffolding for the events that constitute the drama of finance,
and people are its actors. Human emotions drive the game of chess that distributes money around the globe. We
explored the fundamentals of the scaffolding – the basic numbers behind all FX transactions – and highlighted
where supply demand or computational shortcomings have allowed human drivers to insert themselves. Where the
maths got complicated, we drove the thinking to its intuitive roots. Where inconsistencies, conflicting objectives or
curious supply/demand anomalies existed, we showed how mathematical constructs helped explain and resolve
them. In many cases, these mathematical solutions were impossible a few years ago because they were
computationally too demanding, the speed and breadth of communication was not sufficient, or simply the maths
was held captive by a few that were benefiting.
The ability to know when you know is key. The ability to add rigorous proof to intuition is very important. This is
a business where shouting matches happen regularly. If you know you are right and your opponent is wrong, and
you are sure about it, you can escalate the shouting match until you win. However, the more you escalate, the higher
the stakes, so the ability to know that you will win is very crucial. A lot of people will tell you that they know
because it is “intuitive”, or it “makes sense”, but all that is useless. You have to have a logical argument, a sequence
of deductions that complete the proof. Once you have that, then you are able to explain it to other people.
The ability “to be sure” often goes together with that other human quality, attention to detail. In finance, as in
American football, little things mean a lot. When financial firms try to hire mathematical talent and attention to
detail, they often default to using education as a proxy. Education credentials have grown to be a necessary
component of the hiring process. As a sift they may be very effective, but we would point out that very little of what
one learns in college will be used in day-to-day financial jobs. Trainees used to start in the mailroom and progress to
back office, and then become traders – and all the way up they learned by doing. Prestigious degrees are overvalued,
and only used as necessary features of a successful résumé. We’ve attended many financial job fairs as recruiters,
and were always be swamped by a wave of applicants. All of them only talked about their grades, and told us about
their academic excellence as the main criteria we should consider. Many were unable to communicate effectively,
unable to listen, unable to develop a conversation. But they all had a 4.0 average.
Nonetheless, there are many education-taught notions and theories that are very helpful in the mathematics of the
financial markets. For example, it is crucial to understand that expectations are the sum of each outcome times its
probability of happening. This is one of the pillars. Many people really do not understand this or misapply it. This is
a concept we used extensively when talking about the value of an option – however, many will come to me with a
sales estimate that is the sum of the expected revenue from that client times the probability of that sale happening.
This looks sophisticated and feels right, but the truth is the sale has a binary outcome: it will happen and we get all
that revenue, or it will not and we get zero. There is no outcome where we will get 30% of the expected sale.
Typically, there is no relationship between one client buying this product and another client buying another product.
To put them together in a formula that assumes interdependence is useless. Another key notion that has been
discussed extensively in this book is that of a derivative (or rate of change). Many people do not understand that
concept, as it is a bit more difficult to figure out by oneself. You really need somebody to explain calculus to you
before you start figuring it out (if you are not Isaac Newton or Gottfried Leibniz).
This process of mathematical understanding, when successful, comes with a pleasure, an enjoyment and an
appreciation for mathematical work. Those who don’t like it are usually also not very good at it. The joy of maths is
a tingling that you can feel in your brain the moment you understand a complicated piece of structure. Once you
experience that tingling, you want more and more of it. Yes, it is like a drug. This desire to understand will help on
the job, but it will also represent a distraction. A great financial career can be damaged by excessive intellectual
curiosity. As the head of equity quantitatives trading at a large US bank once told me, if I have to press this key a
100 times to make a dollar, I will do it. People who are intellectually curious will have a hard time doing it, as it
would be too boring.4 I have often been lured by exciting intellectual quests that had no revenue prospects, much to
my supervisors’ dismay. It was very hard for me to extricate myself from such quests that make me get out of bed in
the morning, but you can see how damaging this could be for one’s career.5 Once, I was almost fired because my
boss told me to code a backtester for carry trade FX strategies. I loved the project so much that for two weeks,
sitting in the middle of the trading desks, I isolated myself and spent every single minute learning visual basic and
coding. My boss did not say anything for the whole two weeks, but at the end almost fired me and I had to work
very hard to earn his respect back.
Just as mathematics can be a distraction, it can also be a very effective weapon. It can be used to climb the
intellectual pecking order. Traders and strategists will fight to establish their own turf, and will make up treacherous
riddles to reveal each other’s ignorance, to impose their superiority by means of knowledge. I remember a very
obnoxious trader on the rates swap desk at a major US banks asking us recruits to calculate silly complex numbers
such as ei to catch people off guard, unless the recruits knew the answer of course.6 Once you can dominate your
colleagues intellectually, it becomes easier to dominate them in terms of sharing the revenues from the business.
The attraction of using mathematics as a political tool of war is that a mathematical advantage can hardly be lost.
The last bit of paradox we want to share is that mathematics really cannot be taught. The mathematics done on
trading floors around the worlds is just smart people tinkering with numbers until they get answers. It is nurtured by
patience, resilience and speed. Most of the learning happens when people hear about problems; then they disappear
for a couple of hours, and when they come back they have figured out how the solution works. Most mathematical
knowledge has no value if you cannot find the answer by yourself or if you cannot understand the proof.7
So what is the intrinsic value of mathematical knowledge? For one, employing people who understand the
mathematics of financial prices allows banks to market products that look fabulously attractive to those who are not
familiar with all the details. As we have said before, the ability of a structurer is to dig a wedge as wide as possible
between the price a computer model calculates and the educated guess of a client. Showing a structure that the client
thinks should cost a million dollars and sell it for six hundred thousand dollars constitutes a structuring success. It is
the ability to trick the human mind into psychological pitfalls in estimates. However, the bank computer model
thinks the structure is worth two hundred thousand dollars. The difference, four hundred thousand dollars, was
created by the structurer out of thin air. It is the gift of the structurer to be able to summon the music of the numbers
and make money magically appear. This is where mathematics makes mark-up possible.
Against the terrifying power of mathematics in the financial battlefield, the story started to change with the global
financial crisis when public opinion, politicians and regulators decided that mathematical complexity, which was in
the service of leverage, was not a good thing. It became their desire to make financial markets more commoditised,
with few highly liquid instruments, easy to evaluate and easy to trade. This put an end to the power of mathematics
in finance, and shifted the balance of power to solid balance sheets, conservative credit deployment and logistical
efficiencies.
The above is the intrinsic value of mathematical knowledge from a marketing point of view. On the other side of
the aisle, we have the value from the trading point of view. This is the ability to more accurately estimate the
probabilities of real-life outcomes, which in turn influence financial asset prices. People think that only the expert
trader will be able to precisely calculate the impact of both political tensions in the Korean peninsula and shale oil
production increases on the value of the dollar, and translate those factors into price changes. The head of risk at one
of the major quant hedge funds has famously quipped that the difference between a trader and the guy in the street is
that the trader knows the difference between 40/60 odds and 60/40 odds. If you have read this book, we hope you
are now convinced that no, it is not just the expert trader, a computer can also do that. Therefore, just as a clever
marketer will be given a shorter and shorter leash to play with complex option payouts, so the trader will slowly be
replaced by faster and cheaper computer models.

CONCLUSION
Finance, and FX markets in particular, are different from any other industry. Many of the anecdotes you have read in
this book would not have happened in a different industry. As Umberto Eco wrote in 1980: when theory falls short,
the time has come to tell stories. FX is the best place to find these stories because money is easy to store and to
transfer, friction is low and leverage is easy. People can have creative ideas in any field of human activity, but in the
real economy barriers to entry are often insurmountable. In the quintessential finance movie Wall Street, Martin
Sheen plays a worker who despises his son for working in finance. Finance people, Martin Sheen says, just move
money around like parasites, while honest people make things and work in manufacturing. Maybe so, but in finance
people can put ideas and capital to work at lightning speed, and can change the course of the world. If they find a
trick, a loop, an edge, a paradox, they will be able to exploit it because it takes only two seconds to enter into a trade.
By contrast, imagine that in a certain chemical sector, a particular compound is produced in two or three plants in
the world. In that case, the price might not be set based on cost, but rather on demand, especially if the cost of
building another of those plants is prohibitive due to new environmental regulations. Then, the owners of those few
plants can enjoy the profits due to such an enormous barrier to entry.
This does not happen in finance. Trading operations are very quick at entering new markets – for example,
mortgages or inflation-linked bonds or cryptocurrencies. Any newly created trading desk finding ways to make
money could very easily scale up their operations due to the lack of friction in the system. It is just as easy to
execute a ten million dollar trade as it is to execute a hundred million dollar one. Scaling from ten million to a
hundred million requires nothing, no further investment in computers, no extra permissions, no approval from
regulators. This is slowly changing, as banks’ balance sheets, which generate the ability to create leverage, are
coming under much greater regulatory scrutiny and the ability to leverage has become a scarcer resource (see above
about the cross-currency basis not being arbitraged precisely because of these issues).
Finance and FX markets are changing in ways that will make them unrecognisable in a matter of years.
Regulation is playing the same role as the paucity of chemical plants plays in our example above. One of the big
questions this book raises is whether or not the hindering of leverage is a good or bad thing, whether or not the
creation of barriers to entry in this industry is a good or bad thing. Why is the market changing now? Human nature
is what drives the history of all financial transactions. However, we suspect human nature, talents and flaws do not
change – they are the same as they were during the Roman Empire or in Homer’s time. Is it then technical progress
that is the engine of change? As transactions become more disintermediated, bustling exchanges vaporise into
software venues, and algorithms replace barking traders. However, the engine of change is not technical progress –
the true engine of change is the discovery of arbitrage, of easier ways to make money, of shortcuts. Some are
statistical, some mathematical, some behavioural, some legal and some illegal. The marketplace changes when a few
curious people, facilitated by technology, unimpeded by regulation and tradition, figure out quirks and details that
the crowd took for granted.

1 We did not talk a lot about the brokers. These were a handful of shops that facilitated trades in spot, forwards and options between banks, and who
earned a small fee on huge volume. They competed on fees and product offering, but mainly by establishing better relationships with the bank traders
so that they got to see the volume. As the markets became electronic, they came under immense pressure and either merged or disappeared.
2 See https://www.sec.gov/news/pressrelease/2016-152.html
3 T. Schoenberg and P. Hurtado, 2018, “HSBC to Pay $100 Million to End US Currency-rigging Probe”, Bloomberg Markets, January 18 (available at
https://www.bloomberg.com/news/articles/2018-01-18/hsbc-to-pay-100-million-to-end-u-s-currency-rigging-probe).
4 Remember the celebrated story of Michael Burry, a PM who suffered from a mild form of autism. He was portrayed in the book The Big Short, but
Michael Lewis. Burry had this ability to read hundreds of pages of CDOs offering memoranda without collapsing with boredom. This ability allowed
him to correctly forecast the collapse of the real estate bubble that gave rise to the GFC.
5 On a semi-personal note, we can point out a few silly stereotypes: that people coming from Mediterranean cultures aren’t in general more or less
successful that any others, but they tend to be very good at pleasurable endeavours, as opposed to other cultures that are more able to bend their will
and desire to any enterprise to which they set their aim. Mediterranean people give their best when they are spurred on by pure passion, and find it
difficult to pursue to the end necessary but boring tasks. Both authors agree that this is not a biased generalisation!
6 A very similar episode happened about nine years before, on the streets of Paris. My PhD supervisor wanted to impress and maintain his intellectual
superiority by showing off his renaissance history knowledge, so he asked me, point-blank and completely out of context, to name the city where
emperor Charles V, the last emperor to receive papal coronation, was crowned. To be honest, it was a question for history buffs, and one with a very
unexpected answer. And, no, the PhD was not in history, and, yes, to his bad luck, I happened to know the answer. Every dog has its day – I usually
didn’t know the answers to his questions, and have never since been so happy about rote memorisation of history.
7 There is an intriguing theory about why mathematics seems to be more ingrained in Asian cultures than in Western cultures in the beautiful book
Outliers, by Malcom Gladwell.
 References

Accenture. “Blockchain Technology Could Reduce Investment Banks’ Infrastructure Costs by 30 Percent, According to Accenture Report”. (January 17,
2017), https://newsroom.accenture.com/news/blockchain-technology-could-reduce-investment-banks-infrastructure-costs-by-30percent-according-to-
accenture-report.htm
Acerbi, Carlo, Claudio Nordio, and Carlo Sitori. “Expected Shortfall as a Tool for Financial Risk Management”. Abaxbank (February 1, 2008).
Andreou, A. “The Rise of Money Trading has Made our Economy all Mud and no Brick”. The Guardian, November 20, 2013.
https://www.theguardian.com/commentisfree/2013/nov/20/money-trading-economy-foreign-exchange-markets-economy
Atalay, Bulent. Math and the Mona Lisa, Smithsonian Books, 2006, New York.
Bank of England. “Fair and Effective Markets Review: Final Report”. (June 2015), http://www.bankofengland.co.uk/markets/Documents/femrjun15.pdf
BIS. “Foreign Exchange Turnover in April 2016”. Triennial Central Bank Survey, September 2016a, http://www.bis.org/publ/rpfx16fx.pdf
BIS. “FX Global Code: May 2016 Update”. May 2016b, http://www.bis.org/mktc/fxwg/gc_may16.pdf
Black, F., and M. Scholes. “The Pricing of Options and Corporate Liabilities”. Journal of Political Economy 81, no. 3 (May–June, 1973): 637–54.
Brealey, Richard, and Stewart Myers. Principles of Corporate Finance. New York: McGraw-Hill, 1996.
Dawkins, Richard. The Selfish Gene. Oxford: Oxford University, 1976.
Dodd, Randall. “Exotic Derivatives Losses in Emerging Markets: Questions of Suitability, Concerns for Stability”. IMF (2009),
http://www.financialpolicy.org/kiko.pdf
Dolmetsch, C. “‘That’s What I Did’ is Traders’ Worry as Bond Trials Play Out”. Bloomberg, June 15, 2017.
https://www.bloomberg.com/news/articles/2017-06-15/-that-s-what-i-did-is-traders-worry-as-bond-trials-play-out
Eco, Umberto. Il nome della rosa, trans by the authors (Milano, Bompiani, 1980).
Gladwell, Malcom. Outliers. New York: Back Bay Books, 2011.
Hull, John C. Options, Futures, and Other Derivatives. Upper Saddle River: Prentice Hall, 2003.
Hurtado, P. “The London Whale”. Bloomberg, February 23, 2016. https://www.bloomberg.com/quicktake/the-london-whale
Jacque, Laurent L. Global Derivative Debacles, from Theory to Malpractice. Singapore: World Scientific, 2010.
Johnson, R. “Long-term Investors Embrace FX Algos”. Greenwich Associates (June 14, 2017), https://www.greenwich.com/fixed-income-fx-cmds/long-
term-investors-embrace-fx-algos
Kahneman, Daniel. Thinking Fast and Slow, New York: Farrar, Straus and Giroux, 2013.
Lack, Simon. The Hedge Fund Mirage: The Illusion of Big Money and Why It’s Too Good to Be True. Hoboken: Wiley, 2012.
Landler, M. “As Exchange Rates Swing, Carmakers Try to Duck”. New York Times, January 17, 2004. http://www.nytimes.com/2004/01/17/business/as-
exchange-rates-swing-carmakerstry-to-duck.html?_r=0
Larson, E., T. Schoenberg, and C. Dolmetsch. “As Big Banks’ FX Case Ends, Judge Urges Probe of Traders”. Bloomberg, January 5 2017.
https://www.bloomberg.com/news/articles/2017-01-05/rbs-jpm-citi-barclays-fined-2-5-billion-in-fx-rigging-case
Levine, M. “Bank FX Fine Scorecard (Follow Along at Home!)”. Bloomberg View, May 20, 2015. https://www.bloomberg.com/view/articles/2015-05-
20/bank-fine-scorecard-follow-along-at-home-
Lewis, M. The Big Short: Inside the Doomsday Machine. New York: W. W. Norton, 2011.
Lewis, M. Flash Boys: A Wall Street Revolt. New York: W. W. Norton, 2015.
Lewis, M. The Undoing Project. New York: W.W. Norton, 2016.
McLaughlin, D. “Six Banks Pay $5.8 Billion, Five Guilty of Market Rigging”. Bloomberg, May 20, 2015. https://www.bloomberg.com/news/articles/2015-
05-20/six-banks-pay-5-8-billion-five-plead-guilty-to-market-rigging
Mahmud, S. J., and A. Williams. “MiFID II Set to Disrupt Investment Research Worldwide”. Bloomberg Intelligence, September 19, 2016.
https://www.bloomberg.com/professional/blog/mifid-ii-set-disrupt-investment-research-worldwide/
Markowitz, Harry. “Portfolio Selection.” Journal of Finance 7, no. 1 (March 1952): 77–91.
Martin, M., and G. Finch. “JPMorgan to Move Hundreds of Staff to Three EU Offices on Brexit”. Bloomberg, May 3, 2017.
https://www.bloomberg.com/news/articles/2017-05-03/jpmorgan-to-move-hundreds-of-staff-to-three-eu-offices-on-brexit
Mihm, S. “High-speed Trading Used to Mean Carrier Pigeons”. Bloomberg View, April 2, 2014. https://www.bloomberg.com/view/articles/2014-04-
02/high-speed-trading-used-to-be-for-the-birds
Neftci, Salih N. An Introduction to the Mathematics of Financial Derivatives. San Diego: Academic Press, 1996.
Nguyen, L. “Rookie Currency Traders are Causing Big Problems”. Bloomberg, March 20, 2017. https://www.bloomberg.com/news/articles/2017-03-
20/rookie-currency-traders-are-causing-trouble-at-crucial-moments
Nosek, Brian. “Estimating the Reproducibility of Psychological Science”. Science, 349, no. 6251, aac4716 (August 28, 2015).
Oltermann, P. “Greenpeace Loses £3m in Currency Speculation”. The Guardian, June 16, 2014.
https://www.theguardian.com/environment/2014/jun/16/greenpeace-loses-3m-pounds-currency-speculation
Onaran, Y. “The Volcker Rule”. Bloomberg, August 8, 2017. https://www.bloomberg.com/quicktake/the-volcker-rule.
Politi, D. “Argentina’s Big Mac Attack”. New York Times, November 24, 2011. https://latitude.blogs.nytimes.com/2011/11/24/argentinas-big-mac-attack/?
_r=0
Poundstone, W. Fortune’s Formula: The Untold Story of the Scientific Betting System That Beat the Casinos and Wall Street. New York: Hill and Wang,
2005.
Rosenberg, M. R. Exchange Rate Determination: Models and Strategies for Exchange Rate Forecasting. New York: McGraw-Hill, 2003.
Schwager, Jack D. The New Market Wizards. Glenelg: Marketplace Books, 2008. Shankar Vedantam, The Hidden Brain, Spiegel & Grau, New York, 2010
Smith, G. “Why I am Leaving Goldman Sachs”. New York Times, March 14, 2012. http://www.nytimes.com/2012/03/14/opinion/why-i-am-leaving-
goldman-sachs.html
Smith, N. “How Finance Took Over the Economy.” Bloomberg View, April 20, 2016. https://www.bloomberg.com/view/articles/2016-04-20/how-finance-
came-to-dominate-the-u-s-economy
Smith, R. “The Private Equity Firm That Quietly Profits on Top-selling Drugs.” The New York Times, July 8, 2017.
https://www.nytimes.com/2017/07/08/business/dealbook/drug-pricesprivate-equity.html
Sterngold, J. “Currency Option Gets Respect”. New York Times, August 2, 1984. http://www.nytimes.com/1984/08/02/business/currency-option-gets-
respect.html?pagewanted=all
Taleb, Nassim. Dynamic Hedging: Managing Vanilla and Exotic Options. New York: Wiley, 1997.
Taleb, Nassim. Fooled by Randomness: The Hidden Role of Chance in Life and in the Markets. New York: Random House, 2005.
Taleb, Nassim. The Black Swan: The Impact of the Highly Improbable. New York: Random House, 2007.
Wilmott, Paul. Paul Wilmott Introduces Quantitative Finance. New York: Wiley, 2001.
Xu, B., and Y. Liu. “The Case Study: How BMW Dealt With Exchange Rate Risk”. Financial Times, October 29, 2012.
https://www.ft.com/content/f21b3a92-f907-11e1-8d92-00144feabdc0