Lecture 7: Binomial Option Pricing – Part I

EF 4321: Derivatives and Risk Management, City University of Hong Kong

© Team EF4321, 2024

All rights reserved
In this lecture …

§ We answer the following questions:

§ What is a binomial tree, and how to set it up?

§ What are risk-neutral pricing, dynamic replication, and backward induction?

§ How to use binomial trees to price options?

§ Readings:
§ McDonald Ch 10, 11

§ Hull Ch 13

Fall 2024 EF 4321 Lecture 7 2

1. An example

Fall 2024 EF 4321 Lecture 7 3

Replicating a call

§ Stock with price S = $60 and one-period risk-free rate of 20%

§ Over next period stock price either falls to $30 or rises to $90

§ Call option with strike price K = $60 pays either $0 or $30

Fall 2024 EF 4321 Lecture 7 4

Replicating a call

§ How do you price the call? Replication (with stock and bond)

§ Buy Δ = 0.5 share of stock and lend 𝐵 = −$12.50 (i.e., borrow $12.50)

Fall 2024 EF 4321 Lecture 7 5

Replicating a call

§ Portfolio replicates option payoff => C = $17.50

§ Solving for replicating portfolio:

§ Buy Δ shares of stock and lend B

§ If stock price rises to $90, we want the portfolio to be worth

$90×Δ + 1.2×𝐵 = $30

§ If stock price drops to $30, we want the portfolio to be worth

$30×Δ + 1.2×𝐵 = $0

§ Δ = 0.5 and 𝐵 = −$12.50 solve the two equations with the two unknowns

Fall 2024 EF 4321 Lecture 7 6

Replicating a call

§ “Delta” (Δ) is chosen so that the value of the replicating portfolio (Δ×𝑆 + 𝐵) has the same
sensitivity to the value of the underlying asset price S as the option price C

§ Δ is called the hedge ratio or “delta” of the option

Fall 2024 EF 4321 Lecture 7 7

Relative pricing

§ Very important result

§ The option price does not directly depend on the probabilities of a stock price up- or
down-move

§ Intuition:

§ If 𝐶 ≠ Δ×𝑆 + 𝐵, there exists an arbitrage opportunity

§ Probabilities do affect prices, but their effects are already fully reflected in stock prices and
bond prices (risk-free interest rate)

§ Taking stock and bond prices as given, options are redundant and probabilities do not directly
matter in pricing

Fall 2024 EF 4321 Lecture 7 8

Binomial tree

§ Unfortunately, this simple replication argument does not depend work with three or more
payoff states

§ Rather than increase the number of payoff states per period, increase the number of
binomial periods => binomial tree

Fall 2024 EF 4321 Lecture 7 9

2. Binomial model

Fall 2024 EF 4321 Lecture 7 10

Notation

§ Define
§ u = 1 + return if stock price goes up

§ d = 1 + return if stock price goes down

§ r = continuously compounded risk-free rate

§ p = probability of stock price up-move

§ h = length of one binomial period

§ No arbitrage requires 𝑑 ≤ 𝑒 !" ≤ 𝑢

§ Stock and option payoffs:

Fall 2024 EF 4321 Lecture 7 11

One-period replication

§ Payoff of portfolio of Δ shares and B dollars of lending (a bond)

§ Replication requires

§ Two equations in two unknowns with solution

§ Option price by no-arbitrage: 𝐶 = Δ×𝑆 + 𝐵

Fall 2024 EF 4321 Lecture 7 12

Graphical interpretation

§ The replicating portfolio describes a line with the formula

𝐶! = Δ×𝑆! + 𝑒 "! ×𝐵
§ … where 𝐶! and 𝑆! are the option value and stock price after one binomial period, respectively

Fall 2024 EF 4321 Lecture 7 13

Risk-neutral pricing

§ Now we have looked at one-period binomial trees with only two payoff states. What if
there are more payoff states? à multi-period binomial trees

§ Pricing with binomial trees (multiperiod): you can go with either of the two methods

§ 1. Risk-neutral pricing

§ 2. Dynamic replication + backward induction

Fall 2024 EF 4321 Lecture 7 14

Risk-neutral pricing

§ Define

§ No-arbitrage condition 𝑑 ≤ 𝑒 "! ≤ 𝑢 implies 0 ≤ 𝑝∗ ≤ 1

§ Rearrange option price

Fall 2024 EF 4321 Lecture 7 15

Risk-neutral pricing

§ Interpretation of 𝑝∗
§ Expected return on the stock

§ Suppose we were risk-neutral

§ Solving for p

§ Very, very important result

§ 𝑝∗ is the probability which sets the expected return on stock equal to the riskfree rate => risk-
neutral probability

Fall 2024 EF 4321 Lecture 7 16

Risk-neutral pricing

§ Very, very, very important result

§ The option price equals its expected payoff discounted by the riskfree rate, where the
expectation is formed using risk-neutral probabilities instead of real probabilities

§ => this is called risk-neutral pricing

§ As you may have noticed, this is simply a rebranding of the numbers in the replication pricing
method

§ … but it turns out to be a really convenient rebranding oftentimes and thus is quite popular

§ Risk-neutral pricing extends to multiperiod binomial trees and applies to all derivatives
which can be replicated

$∗
Derivatives price = 𝑃𝑉" E 𝑝𝑎𝑦𝑜𝑓𝑓

Fall 2024 EF 4321 Lecture 7 17

Intuition

§ Step 1
§ Derivatives are priced by no-arbitrage

§ No-arbitrage does not depend on risk preferences and probabilities

§ Step 2
§ Imagine a world in which all basic security prices are the same as in the real world but
everyone is risk-neutral (a risk-neutral world)

§ The expected (log) return on any security equals the risk-free rate r

Fall 2024 EF 4321 Lecture 7 18

Intuition

§ Step 3
§ In the risk-neutral world, every security is priced as its expected payoff discounted by the risk-
free rate, including derivatives

§ Expectations are taken with the risk-neutral probabilities 𝑝∗

§ Step 4
§ Derivative prices must be the same in the risk-neutral and real worlds because there is only
one no-arbitrage price

Fall 2024 EF 4321 Lecture 7 19

Two-period binomial tree

§ Stock and option payoffs

§ By risk-neutral pricing, we can directly get

Fall 2024 EF 4321 Lecture 7 20

Three-period binomial tree

§ Stock and option payoffs

§ By risk-neutral pricing

Fall 2024 EF 4321 Lecture 7 21

Backward induction

§ Alternatively, we can explicitly dynamically replicate the option with the underlying asset
and bonds

§ What we know: How to replicate an option in a one-period binomial tree

§ Question: How to replicate an option in a multiperiod binomial tree?

§ Answer: Transform the multiperiod problem to one-period problems that we already know
how to tackle, through backward induction

Fall 2024 EF 4321 Lecture 7 22

Backward induction

§ Consider a two-period binomial tree as an example

§ Backward induction: We start by assuming we are at a node just one period prior to
expiration

§ Two possible nodes: u and d

Fall 2024 EF 4321 Lecture 7 23

Backward induction

§ Suppose we are at node u. Looking forward, we face a one-period binomial tree

§ This allows us to find 𝐶% through replication

𝐶%% − 𝐶%' 𝑢𝐶%' − 𝑑𝐶%%
Δ& = , 𝐵% = 𝑒 ("! , 𝐶% = Δ% 𝑆𝑢 + 𝐵%
𝑆𝑢(𝑢 − 𝑑) 𝑢−𝑑

§ Suppose we are at node d. Looking forward, we face another one-period binomial tree

§ This allows us to find 𝐶' through replication

𝐶'% − 𝐶'' 𝑢𝐶'' − 𝑑𝐶'%
Δ) = , 𝐵' = 𝑒 ("! , 𝐶' = Δ' 𝑆𝑑 + 𝐵'
𝑆𝑑(𝑢 − 𝑑) 𝑢−𝑑

§ (Exercise: verify that 𝐶% and 𝐶' are the same as what you would get from the risk-neutral
pricing method)

Fall 2024 EF 4321 Lecture 7 24

Backward induction

§ After getting 𝐶% and 𝐶' , we can continue the backward induction: Assume we are two-
periods before the expiration. (In a two-period tree example, that means we are at node 0)

§ Looking forward, because we already know 𝐶% and 𝐶' (payoffs we will get from selling the
option at nodes u or d), we face a one-period binomial tree

§ This allows us to find 𝐶* through replication

𝐶% − 𝐶' 𝑢𝐶' − 𝑑𝐶%

Δ* = , 𝐵* = 𝑒 ("! , 𝐶* = Δ* 𝑆 + 𝐵*
𝑆(𝑢 − 𝑑) 𝑢−𝑑

§ (Exercise: verify that 𝐶* is the same as what you get from the risk-neutral pricing method)

Fall 2024 EF 4321 Lecture 7 25

Dynamic replication

§ Δ and B are different at different notes => dynamic replication

§ In year 0, buy Δ shares and lend B

§ In year 1, sell the portfolio for either C# or 𝐶$ , which yields exactly what is required to purchase
a new portfolio (“self-financing”) consisting of either (Δ# , B# ) or (Δ% , B% ), depending on the node

§ In year 2, liquidate the portfolio to replicate the payoff from the option

§ Δ increases from 0 to 1 as the option becomes further in-the-money (as will be clear in a
later example)

Fall 2024 EF 4321 Lecture 7 26

Dynamic replication

§ Dynamic replication with a self-financing strategy:

§ Time 0: cost to buy option ß cost to enter the strategy

§ Time t (0 < 𝑡 < 𝑇): no option cash flows ß no strategy cash flows (due to self-financing)

§ Time T: option payoff realized ß portfolio payoff realized

§ No arbitrage: Strategies with the same cash flows in the future (all times and all
scenarios) have the same cost today
§ => the cost of the option = the cost of the initial replicating portfolio

§ Backward induction: starting from the payoff of the last period and work backwards, until
getting the option price at time 0

Fall 2024 EF 4321 Lecture 7 27

Calibrating binomial trees

§ Say we want to price an options contract with binomial trees

§ How do we set up the parameters in the binomial tree?

Fall 2024 EF 4321 Lecture 7 28

Log returns

§ Properties of log or continuously compounded returns

§ The logarithmic function computes returns from prices
𝑆&(!
𝑟&,&(! = ln
𝑆&

§ The exponential function computes prices from returns

𝑆&(! = 𝑆& ×𝑒 )!,!#$

§ Log or continuously compounded returns are additive

01 141

𝑟*++,*- = ; 𝑟.2+&!-3,. = ; 𝑟$*5-3,$

./0 $/0

Fall 2024 EF 4321 Lecture 7 29

Volatility

§ The variance of the annual log return is

01 161
𝑉𝑎𝑟 𝑟+,,%+- = 𝑉𝑎𝑟 A 𝑟.2,3!-4,. = 𝑉𝑎𝑟 A 𝑟'+7-4,'
./0 '/0

§ Suppose returns are uncorrelated over time and that each month or day returns have the
same variance, then
1 1
𝜎 1 = 12 × 𝜎.2,3!-4 = 252 × 𝜎'+7-4

where 𝜎 1 denotes the variance of annual returns

§ Taking the square root yields annualized volatility

𝜎 = 12 × 𝜎.2,3!-4 = 252 × 𝜎'+7-4

Fall 2024 EF 4321 Lecture 7 30

Calibrating u and d

§ In the absence of uncertainty, a stock must appreciate at the risk-free rate so that next
period it is equal to the forward price
𝑆38! = 𝐹38! = 𝑒 "! 𝑆3

§ With uncertainty, we model the stock price evolution as

§ … where 𝜎 is the annualized volatility of log returns, and 𝜎 ℎ is standard deviation of log
returns over a period of length h

§ Therefore, 𝑢 = 𝑒 "!89 ! and 𝑑 = 𝑒 "!(9 !

§ We refer to a tree constructed this way as a “forward tree”

Fall 2024 EF 4321 Lecture 7 31

Estimating historical volatility

§ We can estimate volatility from historical data using the sample standard deviation of log
returns scaled to an annual horizon

§ For example,

§ Monthly returns

§ Weekly returns

§ Daily returns

§ The shorter the return horizon, the more likely that the mean is ignored

Fall 2024 EF 4321 Lecture 7 32

3. Application

Fall 2024 EF 4321 Lecture 7 33

One-period European call

§ Assume
§ Non-dividend paying stock with current price S = $41

§ Annualized volatility of the stock is 𝜎 = 30%

§ Continuously compounded risk-free interest rate is r = 8%

§ European call with a K = $40 strike and T = 1 year to expiration

§ Consider first a one-period tree with h = T = 1 year

§ Calculate the final stock prices

§ Calculate the final option values

§ Calculate Δ and B

§ Calculate the option price

Fall 2024 EF 4321 Lecture 7 34

One-period European call

§ Calculate the final stock prices

§ Calculate the final option values

§ Calculate Δ and B

§ Calculate the option price

Fall 2024 EF 4321 Lecture 7 35

One-period European call

§ Graphically

Fall 2024 EF 4321 Lecture 7 36

One-period European call

§ Graphically
§ Stock prices

Fall 2024 EF 4321 Lecture 7 37

One-period European call

§ Graphically
§ Option payoffs

Fall 2024 EF 4321 Lecture 7 38

One-period European call

§ Graphically
§ Replicating portfolio

Fall 2024 EF 4321 Lecture 7 39

One-period European call

§ Graphically
§ Option value

Fall 2024 EF 4321 Lecture 7 40

One-period European call

§ Alternatively, use risk-neutral pricing with

Fall 2024 EF 4321 Lecture 7 41

Two-period European call

§ Extend the previous example to price a 2-year option: final stock prices

Fall 2024 EF 4321 Lecture 7 42

Two-period European call

§ Extend the previous example to price a 2-year option: option payoffs

Fall 2024 EF 4321 Lecture 7 43

Two-period European call

§ Extend the previous example to price a 2-year option: replicating portfolio at year 1

Fall 2024 EF 4321 Lecture 7 44

Two-period European call

§ Extend the previous example to price a 2-year option: option values at year 1

Fall 2024 EF 4321 Lecture 7 45

Two-period European call

§ Extend the previous example to price a 2-year option: replicating portfolio at year 0

Fall 2024 EF 4321 Lecture 7 46

Two-period European call

§ Extend the previous example to price a 2-year option: replicating portfolio at year 0

Fall 2024 EF 4321 Lecture 7 47

Increasing the number of steps

§ Dividing the time-to-expiration into more binomial steps yields a more realistic tree with a
larger number of outcomes at expiration
§ Consider again the 1-year European call option example

§ Let all inputs be the same except there are three binomial periods

§ With h = 1/3, u and d are smaller, because holding fixed annual volatility, the stock moves by
less over 4 months than 12 months

§ Given a new u and d, 𝑝∗ also changes

Fall 2024 EF 4321 Lecture 7 48

Increasing the number of steps

Fall 2024 EF 4321 Lecture 7 49

European put

§ Consider a 1-year European put with strike price K = 40

§ [Exercise 7.1] Price that European put with a 3 binomial periods

§ The only difference relative to the call tree are the terminal payoffs: use max{0,K-S} instead of
max{0, S-K}

Fall 2024 EF 4321 Lecture 7 50

European put

Fall 2024 EF 4321 Lecture 7 51

Takeaways…

§ 1. Understand what binomial trees are and what they do in option pricing.

§ 2. Setting up a binomial tree

§ Risk-neutral pricing, dynamic replication, backward induction

§ Calibration of a forward tree

§ 3. Use binomial trees to price options

Fall 2024 EF 4321 Lecture 7 52