EDHEC RISK AND ASSET

MANAGEMENT RESEARCH CENTRE

393-400 promenade des Anglais

06202 Nice Cedex 3
Tel.: +33 (0)4 93 18 78 24
Fax: +33 (0)4 93 18 78 40
E-mail: [email protected]
Web: www.edhec-risk.com

EDHEC Comments on the Amaranth

Case: Early Lessons from the Debacle

Hilary Till
Research Associate, EDHEC Risk and Asset Management Research Centre,
and Principal, Premia Capital Management, LLC
 Abstract

We examine how Amaranth, a respected, diversified multi-strategy hedge fund, could have lost 65%
of its $9.2 billion assets in a little over a week. To do so, we take the publicly reported information on
the fund’s Natural Gas positions as well as its recent gains and losses to infer the sizing of the fund’s
energy strategies. We find that as of the end of August, the fund’s likely daily volatility due to energy
trading was about 2%. The fund’s losses on 9/15/06 were likely a 9-standard-devation event. We
discuss how the fund’s strategies were economically defensible in providing liquidity to physical Natural
Gas producers and merchants, but find that like Long Term Capital Management, the magnitude of
Amaranth’s energy position-taking was inappropriate relative to its capital base.


About the author

Hilary Till is the co-founder of Premia Capital Ms. Till has recently authored and co-authored
Management, LLC, a principal of Premia Risk chapters in The New Generation of Risk
Consultancy, Inc., and a research associate with Management in Hedge Funds and Private Equity
the EDHEC Risk and Asset Management Research Investments (Euromoney, 2003), Intelligent Hedge
Centre. Fund Investing (Risk Books, 2004), Commodity
Trading Advisors: Risk, Performance Analysis, and
A Chicago-based proprietary trading firm with a Selection (Wiley, 2004), Core-Satellite Portfolio
focus in natural resources markets, Premia Capital Management (McGraw Hill, 2005), Hedge Funds:
Management, LLC specialises in detecting pockets Insights into Performance Measurement, Risk
of predictability in derivatives markets using Analysis, and Portfolio Allocation (Wiley, 2005),
statistical techniques. Premia Risk Consultancy, The Handbook of Inflation Hedging Investments
Inc. advises investment firms on derivatives (McGraw Hill, 2006), Hedge Fund & Investment
strategies and risk management policy. In Management (Elsevier, 2006), Portfolio Analysis:
addition, Ms. Till sits on the advisory board of the Advanced Topics in Performance Measurement, Risk
Tellus Natural Resources Fund, a fund of hedge and Attribution (Risk Books, 2006), Fund of Hedge
funds. Before co-founding Premia Capital, Ms. Funds: Performance, Assessment, Diversification
Till was chief of derivatives strategies at Putnam and Statistical Properties (Elsevier, 2006).
Investments where her group was responsible for
the management of all derivatives investments Ms. Till is a member of the curriculum and
in domestic and international fixed income, examination committee of the Chartered
tax-exempt fixed income, foreign exchange, and Alternative Investment Analyst Association® and
global asset allocation. Prior to joining Putnam a member of the hedge fund specialisation exam
Investments, Ms. Till was an equity derivatives committee for the Professional Risk Managers'
analyst and commodity futures trader with Harvard International Association.
Management Company, the investment manager
for Harvard University’s endowment. Ms. Till has a B.A. in Statistics from the University
of Chicago and an M.Sc. in Statistics from the
Ms. Till has written articles on commodities, risk London School of Economics where she studied
management, and hedge funds published in the under a private fellowship administered by the
Journal of Alternative Investments, AIMA Journal, Fulbright Commission.
Derivatives Quarterly, Quantitative Finance, Risk
Magazine, Journal of Wealth Management, and
the Singapore Economic Review. She has served
as a referee for the Financial Analysts Journal.


 Introduction

How can a respected, diversified multi-strategy that it will be cheap in the summer if there is a lot
hedge fund, whose size was reportedly $9.2 billion of supply, but expensive by a certain point in the
as of the end of August, lose 65% of its assets in winter. Mr. Hunter closely watches how weather
a little over a week? affects prices and whether conditions will lead to
more, or less, gas in a finite number of underground
This analysis will provide some preliminary storage caverns.”
answers and consider some early lessons from
this debacle. “Bruno Stanziale, a former Deutsche Bank colleague
and now at Société Générale, works with energy
Amaranth Advisors, LLC is a multi-strategy hedge companies that need to hedge their [forward]
fund, which was founded in 2000 by Nick Maounis production. In an interview in July, he contended
and is headquartered in Greenwich, Connecticut. Mr. Hunter was helping the market function better
The founder’s original expertise was in convertible and gas producers to finance new exploration,
bonds. The fund later specialized in leveraged loans, such as by agreeing to buy the rights to gas for
blank-check companies, and in energy trading. delivery in 2010. ‘He’s opened a market up and
According to Burton and Leising (2006), as of June provided a new level of liquidity to all players,’ Mr.
30th of this year, energy trades accounted for about Stanziale said.”
half of the fund’s capital and generated about 75
percent of their profits. “[Amaranth’s energy book] was up for the year
roughly $2 billion by April, scoring a return of
Davis (2006) has thus far provided the most 11% to 13% that month alone, say investors in the
complete picture of Amaranth’s energy strategies. Amaranth fund. Then … [the energy strategies] …
The following paragraphs quote from her article. had a loss of nearly $1 billion in May when prices of
gas for delivery far in the future suddenly collapsed,
Davis reported that Brian Hunter, Amaranth’s head investors add. [The energy traders] won back the
energy trader sometimes held “open positions to buy $1 billion over the summer …”
or sell tens of billions of dollars of commodities.”
Mr. Hunter was based in Calgary, Alberta. On Monday, 9/18/06, the market was made aware
of Amaranth’s distress. It turns out the fund
“Mr. Hunter saw that a surplus of [natural] gas lost about 35% of its assets during the week of
this summer [in the U.S.] could lead to low prices, September 11th. It lost -$560 million on Thursday,
but he also made bets that would pay off if, say, a 9/14/06 alone.
hurricane or cold winter sharply reduced supplies
by the end of winter. He was also willing to buy The fund scrambled to transfer its positions
gas in even further-away years, as part of complex to third-party financial institutions during the
strategies.” weekend of 9/16 to 9/17, but was unable to do
so. On Wednesday, 9/20/06, the fund succeeded
“Buying what is known as ‘winter’ gas years into the in transferring its energy positions to JP Morgan
future is a risky proposition because that market has Chase and Citadel Investment Group at an apparent
many fewer traders than do contracts for months -$1.4 billion discount to their 9/19/06 mark-to-
close at hand.” market value. This meant that the fund’s investors
had lost about -$6 billion or -65% since the end
“Unlike oil, [natural] gas can’t readily be moved of August.
about the globe to fill local shortages or relieve
local supplies.” Given these facts, we can draw six preliminary
lessons about what happened at Amaranth.
“Traders like Mr. Hunter make complex wagers on
gas at multiple points in the future, betting, say,


1. Lessons for Investors

Full and timely position transparency would not In summary, an examination of the fund’s monthly
have been necessary to raise a red flag about the sector-level p/l would have been sufficient to
fund’s Natural Gas trading. raise a red flag.

Since May, investors knew the energy portfolio That said, an analysis of the fund’s liquidity risk
had typical up or down months of about 11%. would have benefited from an understanding of
The monthly volatility of the energy strategies the precise positions of the fund.
therefore had been about 12%. Therefore, it would
not have been unusual for the fund’s energy trades
to lose -24% in a single month, corresponding to
a two-standard-deviation event.


 2. Lessons for Fund-of-Fund Risk Managers

One would expect that the fund did not entirely This procedure is described in Appendix A: Reverse-
accumulate its Natural Gas positions through Engineering Amaranth’s Natural Gas Positions.
the exchange-traded futures markets; one might
assume that a meaningful fraction of its Natural We do not represent that our analysis shows the
Gas strategies was also acquired through the Over- fund’s actual positions. Instead, our returns-based
the-Counter (OTC) derivatives markets. analysis shows positions that, at the very least,
were highly correlated to Amaranth’s energy
Even if a substantial amount of the Fund’s strategies.
positioning was through the opaque OTC markets,
an examination of the notional value of Amaranth’s According to our returns-based analysis, the fund’s
positions versus the open interest in the back-end positions were indeed massive relative to the open
of the New York Mercantile Exchange (NYMEX) interest in the further-out months of the NYMEX
futures curve would likely have shown that the futures curve.
fund’s positions were massive relative to the
prevailing open positions in the futures markets.
This would have provided a red flag of the liquidity
(or lack thereof) of the fund’s positions.

After the fact, we can do a returns-based analysis

of what the likely sizing of the fund’s positions
were, based on the publicly reported strategies of
the fund, and what the publicly stated path of the
strategy’s p/l had been since June.


3. Lessons for the Hedge Fund Industry, Including Analogies
to the Long Term Capital Management (LTCM) Debacle
There are at least three analogies to the LTCM This natural-gas-oriented hedge fund had been
debacle. founded by a former NYMEX President, Robert
Collins; and its shutdown sent shock waves
[A] Risk metrics using recent historical data would throughout the industry.
not have been helpful in understanding the
magnitude of moves during an extreme liquidation- A near-month calendar spread in Natural Gas
pressure event. experienced a 4.5 standard-deviation move
intraday before the spread market normalized by
There was a preview of the intense liquidation the close of trading on 8/2/06. We assume that
pressure on the Natural Gas curve on 8/2/06, the this move occurred because of MotherRock’s
day before the energy hedge fund, MotherRock, distress.
announced that they were shutting down.

Figure 1 - panel A illustrates the intraday and three-month behavior of the September-vs.-October
Natural Gas (NG U-V) spread.

The intraday peak-to-trough move in the NG U-V spread was 12c on 8/2/06

Figure 1 - panel B

As of 8/1/06, the daily standard deviation of the NG U-V spread had been 2.67c based on the previous three months of data.
Therefore, the spread’s intraday move, which is illustrated in Panel A, was 4.5 (= 12/2.67) standard devations. 
 3. Lessons for the Hedge Fund Industry, Including Analogies
to the Long Term Capital Management (LTCM) Debacle
We might assume that MotherRock had been short compared to Amaranth’s September experience.
the NG U-V spread; that is, they were long October
and short September. Why make this assumption? In Appendix B: Recent Volatility Analysis of
The brief intense rally in this spread on 8/2/06 is Fund’s Reverse-Engineered Positions, we discuss
consistent with the temporary effects of a forced what the volatility of Amaranth’s energy trades
liquidation of a short position in this spread. is likely to have been, based on a returns-based
analysis. Figure 2 reproduces one of the graphs
As it turned out, the scale of MotherRock’s losses, in Appendix B.
which were likely up to $300 million, was small

Figure 2

As of the end of August, the daily volatility of historically allowed international banks to lay
Amaranth’s inferred Natural Gas positions was off illiquid fixed-income risk. These hedge funds
2% based on the previous three months of trading then have traditionally hedged this exposure with
experience. highly correlated, but more liquid, instruments
(although post-1998, they have no longer done
The fund’s loss on Friday, September 15th (inferred so with Treasuries).
from Appendix A’s returns-based analysis) may
then have been a 9-standard-deviation event. In the case of Natural Gas, there is a natural
commercial need for an institution to provide a
Jorion (1999) informs us that based on LTCM’s return for storing Natural Gas for later use during
target volatility, the scale of LTCM’s losses in peak winter demand. In the United States, there
August 1998 would have been an 8 standard- is also inadequate storage capacity in Natural Gas
deviation event. for peak winter demand. Therefore, the winter
Natural Gas contracts have been trading at ever
[B] The strategies that LTCM and Amaranth increasing premiums to summer and fall months
employed are and were economically useful. to (1) incentivize storage; and (2) provide a return
in the future for creating more production and
LTCM’s core strategy had been “relative-value storage capacity. The natural commercial position
fixed-income investing”. is to lock in the value of storage by buying
summer and fall Natural Gas and selling winter
Relative-value fixed income hedge funds have Natural Gas forward. There has been no natural


3. Lessons for the Hedge Fund Industry, Including Analogies
to the Long Term Capital Management (LTCM) Debacle
other side to this trade in sufficient commercial For Amaranth (and other energy hedge funds),
magnitude, which is where the usefulness of such the benefit of providing liquidity to Natural
financial participants as Amaranth comes into Gas producers and merchants is as follows. If
play. The hedge fund could have also provided a trader were long winter Natural Gas versus
liquidity to commercial participants by buying other sectors of the Natural Gas curve, that
winter Natural Gas and then hedging itself with trader’s portfolio would perform very well during
spring contracts. Hurricanes (like 2005’s Hurricane Katrina) and
also during exceptionally cold winters (such as
In Appendix C: Background on the U.S. Natural during February 2003.) In summary, the hedge
Gas Market, we discuss a number of the technical fund would be positioned for extraordinary gains
features of the Natural Gas market. We do so if such weather shocks occurred. But the issue
in order to provide a better understanding of again becomes one of appropriate sizing relative
Amaranth’s Natural Gas strategies. to a fund’s capital base.

[C] Even If a Strategy is Economically Viable, All

Strategies Have Capacity Constraints.

LTCM’s well-known strategy of (for example)

buying (relatively illiquid) 29.5-year Treasury bonds
and shorting 30-year on-the-run Treasury bonds is
obviously defensible. But even the Treasury market
can come under liquidation-pressure stress when
position sizes reach sufficient magnitude, as seen
during the LTCM crisis.


 4. Lessons for Energy Hedge Fund Risk Managers

Veteran commodity traders do use measures have been to examine what the range of the
like Value-at-Risk calculated from recent data Natural Gas spread relationships had been in the
to evaluate risk. But they also employ scenario past. In that case, one would have found how
analyses to evaluate worst-case outcomes. A risky the fund’s structural position-taking was in
natural scenario analysis for Amaranth would its magnitude.

5. Lessons for Multi-Strategy Hedge Fund Managers

The commodity markets do not have natural trader out of a position? In the case of Amaranth,
two-sided flow. For experienced traders in the there was no natural (financial) counterparty who
fixed-income, equity, and currency markets, this could take on their positions in under a week
point may not be obvious. (or specifically during a weekend when the fund
initially tried to transfer positions to a third party).
The commodity markets have “nodal liquidity”. If a The natural counterparties to Amaranth’s trades
commercial market participant needs to initiate or are the physical-market participants who had
lift hedges, there will be flow, but such transactions locked in either the value of forward production
do not occur on demand. or storage. The physical-market participants
would likely have had physical assets against
For experienced commodity traders, a key part of their derivatives positions so would have had
one’s strategy development is a plan for how to little economic need to unwind these trades at
exit a strategy. What flow or catalyst will allow the Amaranth’s convenience.

6. Lessons for Policy-Makers

The derivatives markets are wonderful risk-transfer But obviously the magnitude of Amaranth’s
mechanisms for many economically essential positions was inappropriate for the size of their
activities. Amaranth was indeed providing an capital base, as with LTCM. In Appendix D:
economic service. It is economically desirable for The Post-Liquidation Experience, we note that
the capital markets to incentivize the creation of the assumed Amaranth spreads have (thus far)
sufficient storage capacity of Natural Gas for peak stabilized since the positions were transferred to
winter demand in the U.S. two financial institutions.

10
7. Conclusion

In summary, how could Amaranth have lost (3) Risk metrics using recent historical data would
65% of its $9.2 billion assets under management have vastly underestimated the magnitude of
in a little over a week? moves during an extreme liquidation-pressure
event;
According to published reports, Amaranth Advisors,
LLC employed a Natural Gas spread strategy that (4) If the fund’s risk managers had employed
would have benefited under a number of different scenario analyses that evaluated the range
weather-shock scenarios. These strategies were of Natural Gas spread relationships that have
and are economically defensible, but the scale occurred in the not-too-distant past, they would
of their position-sizing relative to their capital have realized how massively risky the fund’s
base clearly was not. Using a returns-based structural position-taking was in its magnitude;
analysis to infer the sizing of their positions, we
found that their energy portfolio likely suffered (5) It is essential for commodity traders to
an adverse 9-standard-deviation event on the understand how their positions fit into the wider
Friday (September 15th) before the fund’s distress scheme of behaviors in the physical commodity
became widely known. markets: before initiating any large-scale trades
in the commodity markets, a trader needs to
We can draw six early lessons from this debacle: understand what flow or catalyst will allow a
trader out of a position; and
(1) Investors would not have needed position-level
transparency to realize that Amaranth’s energy (6) Amaranth was likely indeed providing an
trading was quite risky. A monthly sector-level economic service for physical Natural Gas
analysis of their profits and losses (p/l) would participants; this hedge fund provided liquidity
have revealed that a –24% monthly loss would for physical-market participants who could then
not have been unusual; lock in the value of forward production or the
future value of storage. But even so, like Long
(2) If investors did have position-level transparency, Term Capital Management, the scale of Amaranth’s
they would have likely noted that the fund’s over- spreading activities was much too large for its
the-counter Natural Gas positions were massive capital base.
compared to the prevailing open interest in the
exchange-traded futures market, which would
have given an indication of how illiquid their
energy strategies were;

11
 Endnotes

The author of this article would like to note that The author is also an advisory board member of
the ideas in this article were jointly developed with the Tellus Natural Resources Fund.
Joseph Eagleeye, co-founder of Premia Capital The source of the price and inventory data used
Management, LLC, http://www.premiacap.com. in this article is from Bloomberg.

References

● Barr, Alistair, “Amaranth Sells Energy Portfolio to J.P. Morgan, Citadel,” MarketWatch, 9/20/06.

● Burton, Katherine and Matthew Leising, “Amaranth Says Funds Lost 50% on Gas Trades This Month,”
Bloomberg News, 9/18/06.

●Burton, Katherine and Jenny Strasburg, “Amaranth Plans to Stay in Business, Maounis Says,” Bloomberg
News, 9/22/06.

● Chambers, Matt and Cassandra Sweet, “Amaranth Debacle Raises Regulation Issues,” Dow Jones
Newswires, 9/22/06.

● Davis, Ann, “How Giant Bets on Natural Gas Sank Brash Hedge-Fund Trader,” Wall Street Journal,
9/19/06.

● Goldstein, Matthew, “Heat is On at ICE,” TheStreet.com, 9/22/06.

● Jorion, Philippe, “Risk Management Lessons from Long-Term Capital Management,” University of
California at Irvine, Graduate School of Management Working Paper, June 1999.

●Lammey, Alan, “Choppy Winter Forecasts Suggest Healthy 1 TcF End to Drawdown,” Natural Gas
Week, 12/5/05, pp. 2-3.

● Trincal, Emma, “Amaranth is Bleeding Assets, But Keeps Talking to Investors,” HedgeWorld, 9/21/06.

● Trincal, Emma, “Amaranth Advisors: Who is to Blame?,” HedgeWorld, 9/26/06.

● Weisman, Andrew and Jerome Abernathy, “The Dangers of Historical Hedge Fund Data,” Risk Budgeting
(Edited by Leslie Rahl), Risk Books, London, 2000.

● White, Ben, “Amaranth Chief Defends Policies,” Financial Times, 9/23-24/06.

12
Appendix A: Reverse-Engineering Amaranth’s
Natural Gas Positions
One would expect that the exact Natural Gas position-level transparency. Instead, investors
positions that were held by the Amaranth Multi- have to infer the exposures of a hedge fund from
Strategy Funds will (and should) be confidential the fund’s return data.
for an extended period of time.
In our analysis, we draw from the spirit of
According to Barr (2006), Amaranth sold its entire Weisman and Abernathy (2000), who discuss
energy-trading portfolio to J.P. Morgan Chase a clever way to infer a hedge fund’s key risk
and Citadel Investment Group on Wednesday, factors and leverage level from performance data.
September 20th. Amaranth apparently did so at a The following analysis draws from the publicly
significant discount to the portfolio’s then mark- reported information that was available as of
to-market value. Until JP Morgan and Citadel have Monday, 9/25/06. If any of the publicly available
comfortably unwound or materially restructured information becomes revised or updated, then this
the risk of this portfolio, one would expect that analysis will need to be correspondingly revised.
these positions will remain confidential. Therefore,
detailed post-mortems on Amaranth’s energy From Davis (2006) and Burton and Strasburg
strategies will correspondingly have to wait until (2006), we note that Amaranth apparently held
the exact positions in this portfolio are made Short Summer /Long Winter Natural Gas spreads
public. as well as Long March / Short April Natural Gas
spreads, including in deferred-delivery years,
Nonetheless, a substantial amount of information possibly through 2011.
has thus far been made public regarding this
debacle. Two of Amaranth’s Natural Gas spread We can create two spreads: (1) a Natural Gas
strategies have been frequently mentioned in spread combination in the March-April contracts
press reports. The size and timing of the fund’s for delivery in 2007, 2008, 2009, 2010 and 2011;
gains and losses in energy trading have also been and (2) a Natural Gas spread combination of
exhaustively detailed in the public domain. We Long Winter (December, January, February, and
therefore have enough information to perform March) and Short Summer (June, July, August,
a simple returns-based analysis on the fund’s and September) for delivery in 2007/8 through
energy strategy. 2010/11.

Returns-based analysis is a well-known technique Let’s examine how each spread has performed
in the hedge-fund industry since investors and since the end of July.
risk managers frequently are not provided with

Figure A-1

13
 Appendix A: Reverse-Engineering Amaranth’s
Natural Gas Positions

Figure A-2

Figure A-3

Figure A-4

Note: The spread combination’s standard deviation was calculated from three months of daily data from 5/31/06 to
8/31/06.

14
Appendix A: Reverse-Engineering Amaranth’s
Natural Gas Positions
We cannot represent that Spread (1) and Spread Note particularly, Figures A-2 and A-4. If one used
(2) were Amaranth’s actual positions, but given daily data from 5/31/06 to 8/31/06 to calculate the
the magnitude of the moves illustrated in Figures historical standard deviations of Spread (1) and
A-1 through A-4, we can say that at the very Spread (2), the moves on 9/15/06 and 9/18/06 were
least, their positions were likely highly correlated massive in their magnitude. This is summarized
to those in this analysis. in Figure A-5.

Figure A-5
Magnitude of Moves Based on Recent
Three-Months of Data in Standard Deviations

9/15/2006 9/18/2006

Spread 1: Natural Gas March - April Spreads: -7.3 -7.9

Spread 2: Natural Gas Summer - Winter Spreads: -14.5 -30.7

We can now infer the size of Amaranth’s positions We can solve for the sizing of the fund’s spread
in Spreads (1) and (2) based on the gains and positions because we now have two equations with
losses for the fund’s energy book. Drawing from two unknowns. Let n = the number of NYMEX-
Burton and Strasburg (2006) and White (2006), equivalent contracts for Spread (1); and let q =
(a) the fund lost -$560 million on September the number of NYMEX-equivalent contracts for
14th, and (b) lost about -35% during the week of Spread (2). Solve for n and q based on:
September 11th. According to Trincal (2006), the
fund had approximately $9.2 billion in assets, as
of the end of August. Therefore, the losses during 560 million = (n * -6,000) + (q * -3,720);
the week of September 11th may have been about and
-$3.2 billion. 3.2 billion = (n * -30,650) + (q * -48,950).

Spread (1) declined -$0.60 on 9/14/06, or -$6,000

per spread combination. (The contract size for
Natural Gas futures contracts on the New York n = 86,308 spreads;
Mercantile Exchange (NYMEX) is 10,000 MmBtu; and
we also say that the contract multiplier for q = 11,331 spreads.
this commodity is 10,000.) During the week of
September 11th, this spread declined -$3.065, or
-$30,650 per spread combination.

Spread (2) declined -$0.372 on 9/14/06, or -$3,720

per spread combination. During the week of
September 11th, this spread declined -$4.895, or
-$48,950 per spread combination.

15
 Appendix A: Reverse-Engineering Amaranth’s
Natural Gas Positions
Again, no representation is being made that these The answer is approximately. This spread
were the fund’s actual positions. But these inferred combination would have made $1.2 billion from
positions are very consistent with the publicly the end of May through the end of August. This
known facts of this debacle. is shown in Figure A-6. We would expect that the
fund’s positions were not static this summer, but
We can double-check whether this position-sizing given how close our derived portfolio’s profits and
is consistent with other known facts about the losses (p/l) are to the fund’s actual energy-book p/l,
fund’s energy portfolio. The energy portfolio we may conclude that our derived positions are
apparently made about $1.3 billion from June similar (or highly correlated) to what the fund’s
through August. Would the position-sizing solved core energy positions were this summer.
for above provide such trading gains?

Figure A-6

Figure A-7 shows how the inferred Natural Gas positions performed during the first two weeks of
September.

Experienced Natural Gas participants will note extensive over-the-counter bets …” Goldstein
that the magnitude of the inferred positions is (2006) adds that it is expected that Amaranth’s
greater than what would be expected from an over-the-counter Natural Gas trades were executed
examination of the futures open interest on via the InterContinental Exchange (ICE). In other
the NYMEX. Write Chambers and Sweet (2006), words, Amaranth’s positions were not (entirely)
“Amaranth’s trades are thought to have involved accumulated via NYMEX futures positions.
16
Appendix B: Recent Volatility Analysis of Fund’s
Reverse-Engineered Positions
Appendix A attempts to reverse-engineer managers might have become so wrong-footed
Amaranth’s Natural Gas positions. We cannot in evaluating the risk of their energy trading.
say that we have identified their actual positions.
Instead, we can say that at the very least, their A simple Value-at-Risk / recent volatility analysis as
positions were highly correlated to those in this of the end of August probably would have vastly
analysis. underestimated the risk of their strategies during
a liquidation-pressure event. Figure B-1 shows
We can now attempt to come up with some the daily P/L from 6/1/06 through 9/15/06 of the
explanations for how Amaranth and its risk fund’s inferred Natural Gas positions.

Figure B-1

Figure B-2 shows the progression in the fund’s p/l in standard-deviation terms, which reached an extreme
on Friday, September 15th.

Note: The standard deviation of the inferred energy portfolio’s p/l was calculated from three months of daily data from
5/31/06 to 8/31/06

As of the end of August, the fund’s risk managers in p/l of 10%. Such a monthly volatility is
may have expected that a one standard-deviation consistent with the p/l swings reported by
move in the energy strategy’s p/l was 2%. This Davis (2006).
translates into a monthly standard deviation
17
 Appendix C: Background on the U.S. Natural Gas Market

Natural Gas derivatives trading has offered the northern states of the United States. Natural
hedge funds a potentially alluring combination gas is also a key energy source for air-conditioning
of scalability and volatility, and also at times, demand during the summer.
pockets of predictability. Traders can access
these markets through the transparent New York There is a long “injection season” from the spring
Mercantile Exchange (NYMEX) for exchanged- through the fall in which Natural Gas is injected
traded exposure, or they can do so through the and stored in caverns for later use during the
opaque InterContinental Exchange (ICE) for over- long winter season.
the-counter exposure.

The key economic function for Natural Gas is to

provide for heating demand during the winter in

Figure C-1 illustrates the normal seasonal pattern of builds and draws in Natural Gas throughout the year.

This graph specifically shows the U.S. Department of Energy’s total estimated storage data for working natural gas inventories
averaged over the period, 1994 to 2005.

Several technical points make Natural Gas an This has a direct impact on the pricing relationships
especially volatile commodity market: between different delivery months for Natural
Gas.
● Natural Gas production has not kept pace with
increasing demand for this commodity; In all commodity futures markets, there is a
● The U.S. Natural Gas markets are largely insulated, different price for a commodity, depending on
at least in the short-term, from global energy when the commodity is to be delivered. For
factors, since only a small amount of U.S. Natural example, a futures contract whose delivery is in
Gas needs are met through imports of Liquid October will have a different price than a contract
Natural Gas (LNG); whose delivery is in December. Commodity traders
● There is insufficient storage capacity of Natural will frequently specialize in understanding the
Gas to meet peak winter demand; and factors that impact the spread between two
● At the end of winter, inventories have to be delivery months; this is known as calendar-spread
cycled out of storage, regardless of price, in order trading. In our example, a futures trader may trade
to maintain the integrity of storage facilities. the spread between the October vs. December
futures contracts.
In essence, the technical issues with Natural Gas
18 mean that it is only a quasi-storable commodity.
Appendix C: Background on the U.S. Natural Gas Market

Figure C-2 shows the futures curve for Natural Gas as of 9/26/06. We refer to the term structure of a
commodity futures market as a curve since each delivery-month contract is plotted on the x-axis with their
respective prices on the y-axis; thus, a curve is traced out.

When the near-month futures contracts trade at The example provided above is actually a simplified
a discount to further-delivery contracts, one says version of how storage operators can choose
that the futures curve is in contango. When the to monetize the value of their physical assets.
near-month futures contracts instead trade at a Sophisticated storage operators actually value
premium to further-delivery contracts, one says their storage facilities as an option on calendar-
the futures curve is in backwardation. spreads. Storage is worth more if the calendar
spreads in Natural Gas are volatile. As a calendar
One can note that the yearly futures curves for spread trades in steep contango, storage operators
Natural Gas in Figure C-2 mirror the average can buy the near-month contracts and sell the
inventory build-and-draw pattern of Figure C-1. further-out month contracts, knowing that they
The prices of summer and fall futures contracts can ultimately realize the value of this spread
typically trade at a discount to the winter through storage. But a preferable scenario would
contracts. The markets thus provide a return for be for the spread to then tighten, which means
storing Natural Gas. An owner of a storage facility that they can trade out of the spread as a profit.
can buy summer Natural Gas and simultaneously Later if the spread trades in wide contango again,
sell winter Natural Gas via the futures markets. they can reinitiate a purchase of the near-month
This difference will be the storage operator’s return versus far-month Natural Gas spread. As long
for storage. When the summer futures contract as the spread is volatile, the operator/trader can
matures, the storage operator can take delivery of continually lock in profits, and if they cannot trade
the physical Natural Gas, and inject this Natural out of the spread at a profit, they can then take
Gas into storage. Later when the operator’s physical delivery and realize the value of their
winter futures contract matures, the operator storage facility that way.
can make delivery of the physical Natural Gas by It is our expectation that both storage operators
drawing physical Natural Gas out of storage for and Natural Gas producers were the ultimate
this purpose. As long as the operator’s financing counterparties to Amaranth’s sizeable spread
and physical outlay costs are under the spread trading.
locked in through the futures market, then this
operation will be profitable.

19
 Appendix C: Background on the U.S. Natural Gas Market

Why are Natural Gas spreads so volatile? It is current demand, and the futures curve has
only when a commodity is fully storable that traded in steeper contango to provide a further
commodity spreads can be predictably stable. enhanced return for storage. This occurred in
In that case, the determining factor between the aftermath of Hurricane Katrina in 2005.
the value of one contract versus a later-month
contract is the cost of storing and financing At the start of the winter, if there are predictions
the commodity from one period to the next. of an exceptionally cold winter, the winter
contracts trade at a large premium to spring
With U.S. Natural Gas, storage capacity has actually contracts in order to encourage supplies to be
declined since 1989. Further, domestic production brought out of storage immediately, and to
has not kept pace with demand. These factors have discourage any non-essential use of Natural Gas.
caused massive volatility in the outright price of This occurred in December of 2005, even though
Natural Gas and in the price relationships between storage at the start of the season was quite high.
different sectors of the Natural Gas curve. To give
one an idea of Natural Gas’ volatility, as of 9/26/06, At the end of the winter, if there is a cold shock
the implied volatility of one-month, at-the-money and inventories are at their seasonal low, the end-
Natural Gas options is 92.5%. This is the case even of-winter contracts can also explode relative to
though there are no hurricanes, heat-waves, or later-month contracts in order to limit current use
cold-shocks presently confronting this market. of Natural Gas to absolutely essential activities.
This scenario occurred in the winter of 2002/3
The outright price of Natural Gas as well as the spread and is illustrated in Figure C-3. Lammey (2005)
relationships in this market are highly sensitive to quotes a futures trader regarding the extremely
the prevailing storage situation for the commodity. cold winter of 2002-3: “I remember that season
well, because we started off the winter with
During the summer if there are hurricanes in intense cold, and ended the season late with
the U.S., concerns emerge that not enough intense cold – and many participants in the
Natural Gas will be produced and stored for industry were seriously worried that there might
winter needs. In that scenario, the front-month not be enough gas to get us across the finish line.”
contract’s price has exploded to discourage

Figure C-3 Panel A: February 2003’s Near-Stock-Out Scenario

20
Appendix C: Background on the U.S. Natural Gas Market

Figure C-3 Panel B

Instead, if the winter is unexpectedly mild, and opportunities around seasonal inflection points
there are still massive amounts of Natural Gas in for Natural Gas use. The summer/fall injection
storage, then the near-month price of Natural Gas season creates opportunities in the summer/fall
plummets to encourage its current use and the versus winter Natural Gas spread relationship.
curve trades in contango in order to provide a return The end-of-winter period creates opportunities
to any storage operator who can still store gas. This in the March-versus-April Natural Gas spread.
occurred during the end of the winter in early 2006. As discussed in Appendix A: Reverse-Engineering
Amaranth’s Natural Gas Positions, it appears
As one may surmise for the above scenarios, the that Amaranth was precisely involved in these
U.S. Natural Gas markets provide many spreading sorts of opportunities on a massive scale.

21
 Appendix D: The Post-Liquidation Experience

The assumed Natural Gas positions in the Amaranth day after the announced transfer of Amaranth’s
energy book are discussed in Appendix A: Reverse- energy book to two financial institutions. This
Engineering Amaranth’s Natural Gas Positions. is illustrated in Figure D-1. Since that time, the
We note that these positions appeared to have assumed Amaranth spreads have stabilized and
reached their trough on Thursday, 9/21/06, one slightly recovered (thus far.)

Figure D-1

22
EDHEC is one of the top five business schools in EDHEC pursues an active research policy in the
France owing to the high quality of its academic field of finance. Its Risk and Asset Management
staff (100 permanent lecturers from France Research Centre carries out numerous research
and abroad) and its privileged relationship with programs in the areas of asset allocation and risk
professionals that the school has been developing management in both the traditional and alternative
since its establishment in 1906. EDHEC Business investment universes.
School has decided to draw on its extensive
knowledge of the professional environment and
has therefore concentrated its research on themes
that satisfy the needs of professionals. Copyright © 2006 Edhec

EDHEC RISK AND ASSET

MANAGEMENT RESEARCH CENTRE

393-400 promenade des Anglais

06202 Nice Cedex 3
Tel.: +33 (0)4 93 18 78 24
Fax: +33 (0)4 93 18 78 40
E-mail: [email protected]
Web: www.edhec-risk.com