Financial Mathematics

Lecture 2 – Binomial Trees, Chapter 12

Jasper Anderluh

Some notes BR

Binomial Trees - Introduction

We start financial derivative pricing with the Binomial

Tree model by CRR 1979.

Why this model ?

It is a basic method, easy to understand that:

Explains the concept of hedging.

Shows the risk neutral measure.

Reveals the elementary sensitivities of

option prices to interest rate and volatility.

Also: for time step to 0, approaches Black Scholes formula

 One Step Model (1)

Consider the very simple model for the stock price:

22 (ST-X)+ = 1

20

18 (ST-X)+ = 0
3 months

Suppose we have a call 21 option that expires in three

months.

What is the payoff at exercise?

One Step Model (2)

The actual probabilities of the stock moving up or

down are given by:

22 (ST-X)+ = 1
0
0.7
p=

20 r = 12% annually

1-p
=0
.30
18 (ST-X)+ = 0
3 months

What is the value of the option?

 One Step Model (3)

Let’s think differently about this. Suppose I sell you

the option, then initially I:

Receive the option value f on my bank account;

Will buy the number ! of shares, paying them

from my bankaccount.
At expiry I:

I sell my ! stocks, receiving the proceeds on my

bankaccount;
Pay you 1 in case the stock is at 22, otherwise
zero.

One Step Model (4)

So, for my bank account, I have :

22 (ST-X)+ = 1
Bank : (f – 20!)e0.25x12%-1+22!"

20 r = 12% annually

Bank : (f – 20!)e0.25x12%+18!"
Bank : f – 20!"

18 (ST-X)+ = 0

3 months = 0.25

How to proceed?
 One Step Model (5)

The fair option price should follow from:

As I start with zero in the bank,

The option price should be considered fair if I

also end up with zero in the bank account,

In both stock up or down scenarios.

Is this (mathematically) possible?

One Step Model (6)

Computation yields:

0.63 for the option value f.

Where did the probabilities p and

(1-p) enter the equation ?

If we compute the discounted expected value we

find 0.68.

How can these values be different?

 Generalization

We can generalize this example:

uS0 (ST-X)+ = fu
Bank : (f – S0!)eTxr%-fu+uS0!"

S0 r% annually

Bank : (f – S0!)eTxr%-fd+dS0!"
Bank : f – S0!"

dS0 (ST-X)+ = fd

expiry at T

Then we find the following equation:

Risk Neutral Measure (1)

Studying the general equation, we observe:

The p in the general equation is, for now, an

artificial probability that we found by manupilating
the general expression.

The actual probability that is driving the market

price of the stock never enters the equation !!

What if we compute the expecation of the

stock price, using this artificial probability?
 Risk Neutral Measure (2)

Computing :

Results in : Valuation Formula !

So, apparantly, in the world having these probabilities,

the value of an asset S is determined by discounting
its expected value with the risk-free rate!

No compensation in the discount rate for risk,

so we value like we are risk-neutral.

Risk Neutral Measure (3)

The same holds for a derivate:

Valuation Formula !

Like the stock, we find its value by discounting its

expected value with the risk-free rate!
Text

How is this related to the CAPM you have been

tought in the Investment Theory course?
F4E module
 Risk Neutral Measure (4)

CAPM states the following:

You should receive average return related to the

beta of your investment.

This beta depends on the correlation with the

“market”, i.e. the rest of the stocks.

So, for risk that you can diversify away, you

receive no extra return above the risk free rate.

After selling your option and buying delta

stocks, your portfolio has become riskless.
According to CAPM, you should receive the risk-
free rate, exactly that is what we do!

2 or More Step Trees (1)

Now, two possible outcomes is to simple.

Suppose, we extend to:

uS0 (ST-X)+ = fu

S0 S0 (ST-X)+ = fm
Bank : f – S0!"

dS0 (ST-X)+ = fd

expiry at T

What is the (mathematical) problem here?

 2 or More Step Trees (2)

We can also extend in another way, covering 2

periods:

u1S0 u1u2S0 (ST-X)+ = fuu

u1d2S0 (ST-X)+ = fud

S0 d1u2S0 (ST-X)+ = fdu
d1S0
d1d2S0 (ST-X)+ = fdd
!T !T

How many outcomes after N extensions?

2 or More Step Trees (3)

By setting ui=u and di =d we get the recombining

tree:

u2S0 (ST-X)+ = fuu

uS0

udS0 (ST-X)+ = fud

S0

dS0
d2S0 (ST-X)+ = fdd
!T !T

The number of outcomes changes from 2N to

(N+1).
 2 or More Step Trees (4)

We proceed by calculating backwards every cell

according to the one-step tree:

u2S0 (ST-X)+ = fuu

uS0

fu
udS0 (ST-X)+ = fud
S0
fd
dS0
d2S0 (ST-X)+ = fdd
!T !T

First we compute fu and fd from the right-hand cells.

Then we compute f from fu and fd.

2 or More Step Trees (5)

Using the formula we already have for one cell:

We find with in the first step:

And we use this to obtain the option value f:

 Put / American

As the formula we have is entirely defined in terms of

the payoff (fuu, fud, fdd), we can value a put easily by:

fuu = (K - u2S0)+, fud = (K - udS0)+, fdd = (K - d2S0)+

For the American option: typo: uS_0-K should be K-uS_0

We can excerise intermediately.

The option value at an intermediate node:

Value of intermediate exercise Value of a European option with

payoff fuu and fud.

Delta

As in the one-period tree, remember for every cell:

We construct a portfolio of ! shares and the received

amount f.
In such a way that if we sell the ! stocks in the next
step, we have precisely the amount of money in our
bank account equal to the option value in this next
step, both in the up and down scenario.

So, every node in the tree has its own delta,

computed by:
 Matching Volatility with u and d (1)

Now, we have to set the parameters u and d in such a

way that we can use the tree to calculate actual
option prices. Remember,

In the risk-neutral world we calculated with p such

that E[S!t] = er!t
In the real world we calculated with p such that
E[S!t] = eµ!t

We define the volatility # in the real world by stating

that the variance of the return in a period of !t should
equal:

Matching Volatility with u and d (2)

We can deduce that for some 0<p<1 resulting in a

drift " we find by equating variances:

Variance of the Variance of the return

return in the tree in the real world

In the limit for !t to zero, we have the solution :

What is special about this result?

 Matching Volatility with u and d (3)

The result is of interest because:

It is obtained for a general $.

Therefore, it shows that for small !t, the volatility

can be set by choosing u and d,

Regardless of the expected return of the stock, i.e.

regardless of the value of p.

u and d are the same for both the

real and the risk-neutral world!

That is necessary for measuring volatility from

historical stock price paths to price options!

Increasing the number of steps

The question is, what happens if we increase the

number of steps?

We see convergence to some limit. This limit appears

to be the Black-Scholes price.
 Other Options

Some examples for valueing other options:

Option on Replace r by

Stock paying continuous dividend q. r-q

Stock Index r-qavg

Currencies r-rforeign

Futures 0

It always boils down on what do I receive on or pay

from my bank account for financing the position in the
underlying.