9/25/2019

Models We Examine

 We will examine two widely-used classes of models here:

 The binomial model.
 The Black–Scholes model. Options - Model 1
 The Black-Scholes model
 offers closed-form solutions for option prices, but
 is relatively limited (e.g., no closed forms with early
exercise).
 The binomial model is much more flexible and intuitive.

1 2

Binomial model Binomial Model's Flexibility

 With frequent price changes, we can create a large number

 The "binomial" assumption: Price moves from its current level of possible future prices and a variety of future price
S to one of two possible levels, uS or dS. distributions.
 u: "up" move. Occurs with probability p.  In particular, the model's parameters can be chosen to
 d: "down" move. Occurs with probability 1 — p .
approximate the Black-Scholes price process arbitrarily
 The risk-free interest rate is denoted r. closely.
 the rate of interest applicable to the time-period represented
by each step of the binomial tree.  Key input parameter in any option pricing model:  , the
annualized volatility of the stock.
 r is the gross rate of interest per time step.
 To approximate the Black-Scholes model, set
 That is, ₹1 invested at the beginning of the period will grow to
₹ r at the end of the period.
where h the length in years of one step of the tree.

3 4

1
 9/25/2019

Call option pricing problem

110 1.02 10
Imp: We must have d < r < u
100 1 C=?

90 1.02 0
 Binomial models used in practice use 100 periods or more,
typically many more.
Stock Cash Call
 We begin by examining option pricing in a one-period
model.

p = 0.75, 1-p =0.25

5 6

Moving to multiple periods

Pricing by replication  Suppose u and d are constant.

 After one period, there are two possible prices for the asset,
Can you price the Puts? namely uS
A quicker and useful method and dS.
Why this terminology?  In the second period, each of these two prices can itself go up by a
Why does it work? factor of u or down by a factor of d.
 Therefore, there are four possible paths that prices can take:
u followed by u.
u followed by d.
d followed by u.
d followed by d.

7 8

2
 9/25/2019

Two period tree

 However:
 u followed by d  terminal price = d (uS ) = udS.
 d followed by u  terminal price = u (dS ) = udS.
 So: only three distinct terminal prices:
 u 2S
 udS
 d 2 S.
 This feature is known as recombination of the Binomial tree

9 10

Pricing a call option with same parameters as

before

Price of Put option ?

Extending from 2 to n steps

The real power of Binomial model - Pricing

American options

11 12

3
 9/25/2019

 The Black-Scholes-Merton model is unambiguously the best

known model of option pricing.
 Technically, the BSM model is more complex than the
 It is also one of the most widely used: it is the benchmark binomial or other lattice models.
model for pricing options on The BSM model is a continuous-time model: prices in the
Equities. model are allowed to change continuously rather than at
Stock indices. discrete points in time as in the binomial model.
Currencies (the "Garman-Kohlhagen model"). Requires the use of sophisticated math!
Futures (the "Black model").

 The Black model is also commonly used to some interest-rate

derivatives such as caps and floors.

13 14

 Unlike the binomial or other lattice models, option prices in As in all option pricing models, the principal
the BSM model can be expressed in closed-form, i.e., as assumption of Black-Scholes-Merton concerns the
particular explicit functions of the parameters. evolution of the stock price.
This makes computing option prices very easy Under their assumptions, Returns on the stock
Developing and verifying the intuition about option pricing follow a log-normal distribution
and hedging behavior is made easier
It also makes computing option sensitivities very easy

15 16

4
 9/25/2019

The Log-Normal Assumption Lognormal density function

What does mu=0.10 mean?

The log-normal assumption says that the (natural)

log of returns is normally distributed: if S0 denotes
the current price, and St the price t years from the
present, then

17 18

The number σ is called the volatility of the stock.

The BSM model takes this volatility to be a
constant. In principle, this volatility can be
estimated in two ways:
From historical data. (This is called historical
volatility.)
From options prices. (This is called implied
volatility.)

How do you compute this volatility from past data?

19 20

5
 9/25/2019

LN distn and simple returns

 Suppose returns are lognormal over a one-year horizon with

µ = 0.10 and σ = 0.40
Normal distn is one of the easiest distn to work
 Over the one-year horizon, this means:
with and has powerful properties.
Expected log-returns: 10% Almost all properties of the normal can be carried
Variance of log-returns: 16% forward to the lognormal.
 What is the mean and variance of simple one year returns?

Upshot: Don’t use log returns and simple returns

interchangeably.

21 22

An example Coming back to BSM option pricing model

Consider a horizon of three months, and suppose a The model assumes that this price evolves
stock has lognormal returns with µ = 0.10 and according to a geometric Brownian motion.
σ = 0.30 What is a "geometric Brownian motion?“
Suppose the current stock price is S=110. First, Returns on the stock follow a log-normal
What is the 95% confidence interval for the stock distribution
price in 3 months?

23 24

6
 9/25/2019

Other assumptions Variables

The risk-free interest rate r is constant. (Since we  0 : current date

are in a continuous-time world, the interest rate is  T : Maturity date (So time-left-to-maturity is also T )
also expressed in continuously-compounded  K : strike price of option.
terms.)  S0 : current price of stock.
No dividends during the life of the option. (We will  ST : stock price at T.
drop this later.)
 µ, σ : Expected return and volatility of stock (annualized).
The options are European in style. (No closed-
 r : risk-free rate of interest.
forms are possible for American options.)
 C , P : Prices of call and put (European only).

25 26

BSM formula

The call-pricing formula in the Black-Scholes model is

C = S0 · N (d1) − e −rT K · N (d2)  The price of a put in the Black-Scholes model is given by

where N (·) is the cumulative standard normal distribution [N (x) is the P = [PV (K ) × N (−d2)] − [S0 × N (−d1)],
probability under a standard normal distribution of an observation less
than or equal to x ], and where d1 and d2 are as defined earlier.

27 28

7
 9/25/2019

Try problems
Ch11: 5, 9
Ch12: 15, 16, 22
Ch13: 1

29