0% found this document useful (0 votes)
19 views30 pages

L14 Option MStep Binomial Tree Handout

This document outlines multi-step binomial tree models for pricing financial assets. It provides an example of a two-step binomial tree model, showing how to price a call option recursively by moving up the tree. Dynamic trading strategies are discussed, which adjust the portfolio composition at each node to replicate the derivative payoff. These self-financing strategies do not require external cash flows after initialization. Risk-neutral probabilities are also introduced, which must satisfy the no-arbitrage condition when used to price assets in the tree model.

Uploaded by

kokmunwai717
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
19 views30 pages

L14 Option MStep Binomial Tree Handout

This document outlines multi-step binomial tree models for pricing financial assets. It provides an example of a two-step binomial tree model, showing how to price a call option recursively by moving up the tree. Dynamic trading strategies are discussed, which adjust the portfolio composition at each node to replicate the derivative payoff. These self-financing strategies do not require external cash flows after initialization. Risk-neutral probabilities are also introduced, which must satisfy the no-arbitrage condition when used to price assets in the tree model.

Uploaded by

kokmunwai717
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 30

Lecture 14

Multiple-Step Binomial Tree Models

Yicheng Zhu

1 Apr 2022

1 / 30
Outline

1. Two-Step Binomial Trees and Dynamic Trading

2. Multi-Step Binomial Trees

3. Pricing on Multiple-Step Binomial Trees

2 / 30
Extending the One-step Binomial Tree Model

So far we have covered one-step binomial tree for option pricing.

I The model allows us to again facilitate the replication argument and price assets,
although the payoff function of the derivatives might no longer be liner (a straight
line).

I However, this model is very unappealing as the assumption (two states) is very strong
and far away from reality.

I One way to resolve the issue is by thinking about multiple steps for a binomial tree.

I The following example will illustrate the idea.

3 / 30
Two-step Binomial Tree Model: an Example

Suppose today at time t, stock ABC is traded at St $100 per share. 6 months later, it has
a p = 60% probability of going up by 2%, and a probability of 1 − p = 40% going down
by 1.96%. The stock price can further move with the same change in percentage and
probability 1 year later. The risk-free interest rate is 0%, constant.

St+1 = 104.04
p = 60%

St+0.5 = 102
p = 60%

1 − p = 40%
St = 100 St+1 = 100
p = 60%

1 − p = 40%
St+0.5 = 98.04

1 − p = 40%
St+1 = 96.12

4 / 30
Two-step Binomial Tree Model: an Example

St+1 = 104.04
p = 60%

St+0.5 = 102
p = 60%

1 − p = 40%
St = 100 St+1 = 100
p = 60%

1 − p = 40%
St+0.5 = 98.04

1 − p = 40%
St+1 = 96.12

Suppose at time t the investor wants to purchase a call option with strike price 98 and
maturity 1 year. What is the fair price for the option?

5 / 30
Recursive Pricing

One way of solving the problem is to solve the price of the call option recursively.

I You start with the terminal payoffs, and then can solve the prices of the call option
one period ahead.

I After solving the prices of the call option one period ahead of the maturity in all
states, move one step ahead.

I Each step requires fewer states to solve, and eventually we end up with the initial
point.

6 / 30
Two-step Binomial Tree Model: an Example (Cont’d)

We first consider ‘conventional’ method of replication. Consider the ‘upper’ sub-tree, a

St+1 = 104.04
p = 60%

St+0.5 = 102

1 − p = 40%
St+1 = 100

The payoff of the call option is given by $6.04 and $2, respectively, or

St+1 = 104.04
p = 60% Ct+1 = 6.04
St+0.5 = 102
Ct+0.5
1 − p = 40%
St+1 = 100
Ct+1 = 2

7 / 30
Two-step Binomial Tree Model: an Example (Cont’d)

St+1 = 104.04
p = 60% Ct+1 = 6.04
St+0.5 = 102
Ct+0.5 = 4
1 − p = 40%
St+1 = 100
Ct+1 = 2

We want to solve for the fair price of the option at t + 0.5, if the stock price goes up.
I The ∆ now is
6.04 − 2
∆= = 1.
104.04 − 100
I In addition, we need to carry 2 − 100 × 1 = −98 in zero-coupon bonds.

I That means that, at time t + 0.5, we need to pay 102 × 1 − 98 = 4 to replicate the
payoff of the option, if the stock price goes up.

I This must also be the price (value) of the call option.

8 / 30
Two-step Binomial Tree Model: an Example (Cont’d)

Now let’s consider the ‘lower’ half of the tree:


St+1 = 100
p = 60% Ct+1 = 2
St+0.5 = 98.04
Ct+0.5
1 − p = 40%
St+1 = 96.12
Ct+1 = 0

I The ∆ now is
2−0
∆= = 0.515.
100 − 96.12
I In addition, we need to carry 0 − 96.12 × 0.515 = −49.50 in zero-coupon bonds.

I That means that, at time t + 0.5, we need to pay 98.04 × 0.515 − 49.50 = 0.99 to
replicate the payoff of the option, if the stock price goes down.

9 / 30
Two-step Binomial Tree Model: an Example (Cont’d)

Now we know perfectly how much the option should cost at time t + 0.5.

St+0.5 = 102
p = 60% Ct+0.5 = 4
St = 100
Ct
1 − p = 40%
St+0.5 = 98.04
Ct+0.5 = 0.99

I The ∆ now is
4 − 0.99
∆= = 0.76.
102 − 98.04
I In addition, we need to carry 4 − 102 × 0.76 = −73.52 in zero-coupon bonds.

I That means that, at time t, we need to pay 100 × 0.76 − 73.52 = 2.48 to replicate
the payoff of the option.

I The call option should then cost $2.48.

10 / 30
Dynamic Trading

All the calculation before implies a trading strategy that can replicate the payoff of the call
option:

I Start with ∆ = 0.76, and the corresponding zero coupon bonds.

I After 1 period, at t + 0.5, adjust the stock holding to the corresponding ∆ of the
nodes.

I The trading strategy here contains adjustment of the portfolio, while in previous
models the portfolio composition is fixed.

I The strategy here is called dynamic replicating trading strategies, while holding a
fixed portfolio is called static replicating trading strategies.

I Unsurprisingly, dynamic replicating trading strategies are useful only if we work with
binomial trees with more than one step.

11 / 30
Dynamic replication

I The portfolio also contains no external cash flow after initialization.

I If you adjust your portfolio, you fund the adjustment using bonds and then put the
bonds in your portfolio, and as a result the value of your portfolio does not change.

I The fact that the strategy does not need further external cash flows suggests that
this strategy is a self-financing dynamic trading strategy.

I Definition: A self-financing dynamic trading strategy is a strategy that does not have
any infusion or withdrawal of money after the initial purchase/sale of assets at time 0.

I The present value of the payoff at time T from such a strategy is the time 0 cost of
setting up the strategy (otherwise arbitrage: buy low, sell high)

12 / 30
Risk Neutral Pricing

I The risk-neutral probability q of going up at each node must suffice

1 = q + (1 − q)
1
S = qS × 1.02 + (1 − q)S × .
1.02
I As a result, q is given by
1
1 − 1.02
q= 1
= 49.5%.
1.02 − 1.02

I Question: Why q are the same for all nodes?

13 / 30
Risk Neutral Pricing

I The call option price if the stock price goes up in the first period then must be

6.04 × q + 2 × (1 − q) = 4.
I The call option price if the stock price goes down in the first period then must be

2 × q + 0 × (1 − q) = 0.99.
I The call option price at initiation then must be

4 × q + 0.99 × (1 − q) = 2.48.

14 / 30
Risk Neutral Pricing (cont’d)

I Again the result must be the same for the two approaches as risk neutral pricing is
equivalent to absence of arbitrage.

I However, the calculation with risk-neutral pricing is must easier.

I Absence of arbitrage argument, however, tells what actual trading strategies are
needed to replicate the payoff of the derivative.

15 / 30
Outline

1. Two-Step Binomial Trees and Dynamic Trading

2. Multi-Step Binomial Trees

3. Pricing on Multiple-Step Binomial Trees

16 / 30
Multiple-step Trees

I The most important advantage of two-step binomial tree over one-step binomial tree
is that the former allow a much richer characterization of the terminal payoff.

I However, for each step, there could be only two outcomes, and this allows to replicate
the dynamics of the derivative price movement using the underlying and bond only.

I A natural extension is to cut the price movement into even finer steps, but keep the
binomial tree structure. This allows us to get a much accurate approximation of the
reality.

17 / 30
Setting Up Multiple-step Trees

I We want to use a n-step tree to approximate the return of an asset from time t to
t + T.

I Question: How to set up the tree?

I A standard algorithm goes like this:

I We chop off the time interval [t, t + T ] evenly into n small intervals

I Each interval is of size h = T /n. We denote the steps i = 0, 1, ..., n

=⇒ i × h = calendar time on the tree

18 / 30
Calibrate Multiple-step Tree Parameters

I Assume the asset has log expected constant return µ and constant volatility σ

I Then, over a small interval of time h:

I Expected return is  
St+h
Et = e µ×h
St
I Conditional variance is
" 2 #
St+h
Et − e µ×h = σ2 × h
St

I We want our tree match these moments for each step.

19 / 30
Calibrate Multiple-step Tree Parameters (Cont’d)

I Assume that the stock prices evolves according to:



Si × (1 + u) with probability p
Si+h =
Si × (1 + d) with probability 1−p

I We now choose u, d and p to approximate the expected return µ and variance σ 2 on


the tree

I The following conditions must be satisfied:


 
Si+1
1. Expected Return: Et Si
= e µ×h =⇒

p × (1 + u) + (1 − p) × (1 + d) = e µ×h

 2 
Si+1
2. Variance: Et Si
− e µ×h
= σ 2 × h =⇒
 2  2
p 1 + u − e µ×h + (1 − p) 1 + d − e µ×h = σ 2 × h

20 / 30
Calibrate Multiple-step Tree Parameters (Cont’d)

I We have two equations, but three parameters, and can not uniquely solve for the
parameters.

I One easy way to fix this is to assume

(1 + u)(1 + d) = 1,

and this implies that the price remains the same after one ‘up’ and one ’down’.

I The solution is then


√ √
e µ×h − (1 + d)
1 + u = e σ× h
; 1 + d = e −σ×h
; and p = .
u−d
I Next figure shows an example from BinomialTree.xls

21 / 30
Multiple-step trees: a pricing tool

I Time index: i = 0, 1, ..., n. Node index: j = 0, 1, ...., n


I Because we start from up-left corner
I an up movement in S between i and i + 1 correspond to same index j;
I a down movement in S between i and i + 1 corresponds to an increase in the index j.

22 / 30
Outline

1. Two-Step Binomial Trees and Dynamic Trading

2. Multi-Step Binomial Trees

3. Pricing on Multiple-Step Binomial Trees

23 / 30
Multiple-step Trees: Derivative Pricing

I The simplest way of pricing derivatives using multi-step binomial tree is to use
risk-neutral pricing.

I The risk neutral probability of going up is given by


e r ×h − (1 + d)
q=
u−d
Since u, d, h are the same for each step, q remains constant. r is the risk-free rate.

I Move backward on the tree one step at a time:


Vi,j = e −r ×h × Eq [Vi+1 |(i, j)]
where i, j is the (time, node) state on the tree.

I For instance, for call options, start from the end of the tree with the final condition

Cn,j = max(Sn,j − K , 0) for j = 0, 1, ..., n


I Then move backward applying

Ci,j = e −r ×h × (qCi+1,j + (1 − q)Ci+1,j+1 )

24 / 30
Multiple-step trees: example

Option Price

25 / 30
Multiple-step trees: example with 250 steps

time ==> 0.000 0.000 0.001 0.001 0.002 0.002 0.002 0.003 0.003 0.004 0.004
i==> 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0
0 3.876 4.193 4.527 4.879 5.247 5.633 6.036 6.457 6.894 7.348 7.818
1 3.561 3.861 4.177 4.511 4.863 5.231 5.618 6.021 6.442 6.879
2 3.262 3.545 3.845 4.161 4.495 4.847 5.216 5.602 6.005
3 2.980 3.246 3.529 3.828 4.145 4.479 4.831 5.200
4 2.715 2.964 3.230 3.513 3.812 4.129 4.463
5 2.466 2.699 2.949 3.214 3.497 3.796
6 2.233 2.451 2.684 2.933 3.198
7 2.016 2.218 2.436 2.668
8 1.814 2.002 2.204
9 1.628 1.801
10 1.455

26 / 30
Multiple-step Trees: Probability Distribution of ST

I What is the probability distribution of the stock price at maturity ST ?

I Consider an n-step tree

I There are n + 1 nodes at maturity, j = 0, 1, ..., n


I The top node j = 0 can only be reached with n ups u and 0 downs d
I Node j = 1 can only be reached with n − 1 ups u and 1 down d
.. ..
. .
I Node j = n can only be reached with 0 ups u and n downs d

I If each up u has risk-neutral probability q, what is the probability of reaching


exactly node j?
 
n!
P (node = j at T = n × h) = × (q)n−j × (1 − q)j
j!(n − j)!
where n! = 1 × 2 × 3 × ... × n (note: 0!=1)

27 / 30
Multiple-step trees: probability distribution of ST

As n increases to infinity, the distribution at T = n × h of the stock price ST converges to


a log-normal distribution.

Probabilities of ST for n = 250 and T = 1


0 06
0.06

0.05

0.04

0.03

0.02

0.01

0
0 100 200 300 400 500
Ͳ0.01
STOCK PRICE

28 / 30
Multiple-step trees: probability distribution of ST
Or, plotting the probability density with respect to log(ST ), we obtain

0.06

0.05

0.04

0.03 BinomialTree
NormalDist

0.02

0.01

0
3.5 4 4.5 5 5.5 6

Log Stock Price

This will be used as an assumption directly for a model later in the course.
29 / 30
Conclusion

I Binomial trees are very useful tools for derivative pricing


I They can be easily programmed on computers to price derivative security recursively
(sometimes called backward induction..)
I We can use them to compute the positions in the dynamic replication portfolio.
I As we will see, we can also use binomial trees to compute the value of American options
(much harder than Europeans), and price more complicated exotic. securities

I Risk neutral pricing is another useful tool for pricing securities.


I It is just a tool, and it does not require the assumption that agents are risk neutral!
I The risk neutral methodology stems from the fact that we can use a replicating
portfolio to hedge the exposure from a derivative security.

30 / 30

You might also like