BLACK-SCHOLES OPTION PRICING MODEL

Black and Scholes Merton’s approach:

Used the capital asset pricing model to Setting up a riskless portfolio consisting of the
determine a relationship between the option and the underlying stock and arguing
market’s required return on the option and that the return on the portfolio over a short
the required return on the stock. period of time must be the risk-free return.
 Not easy because it depends both on the
stock price and time.

Merton’s approach to deriving the Black-Sholes-Merton model

 How volatility can be either estimated from historical data or implied from options
prices using the model?

 How the risk neutral valuation argument can be used?

 How the Black-Scholes-Merton model can be extended to deal with European call
and put options on dividend-paying stocks?

Risk neutral valuation

GEOMETRIC BROWNIAN MOTION

GOAL: a continuous price process

The goal is to upgrade from the binomial model;

We want a stochastic price process that satisfies two properties:

Is a variable whose value randomly changes over time

1. the price can change at any time – continuous – time;

2. and the price, at any given time, can assume any value in the interval [0, ∞] –
continuous-variable

To obtain such a price process  Brownian Motion

BROWNIAN MOTION (a.k.a. Wiener process)

DEFINITION: BROWNIAN MOTION

The process Z is a Brownian Motion if it satisfies the following properties:

 1. Z0 = 0
2. Zt2 – Zt1 ~ N(0, t2 – t1), ∀t2 > t1

The Brownian Motion increments from date t 1 to t2 are normally distributed with
mean 0 and variance t2-t1

3. Zt2 – Zt1 and Zt4 – Zt3 are independently distributed ∀t4 > t3 ≥ t2 > t1

The Brownian Motion increments over non-overlapping time intervals are

independent

4. Z has continuous sample paths;

There are no jumps: if we observe the Brownian Motion at very short time
intervals we can’t see any gaps in its process;

THE BROWNIAN MOTION DIFFERENTIAL REPRESENTATION

We can rewrite property 2 Zt2 – Zt1 ~ N(0, t2 – t1), ∀t2 > t1
as follows:

Z t 2−Z t 1=ε √ t 2−t 1 , where ε ~ N(0,1)

In the special case where t1=0 and t2 = t we have:

Z t−Z 0=ε √ t−0 ⟺ Z t =ε √ t

≡0
From this we obtain the most common representation (differential form) of Brownian
Motion:

where d= instantaneous change (i.e. over an infinitesimal time interval)

GEOMETRIC BROWNIAM MOTION: THE PRICE PROCESS

The Brownian Motion will be used as a building block in the construction of the price
process for the underlying asset:
 it delivers us the uncertainty we observe in prices;
 but just by itself (dSt = dZt) it would be a poor price process:
 o the Brownian Motion can assume negative values, but prices don’t;

o even though prices go up and down all the time, over the long run they
follow an upward trend, whereas the Brownian Motion has no trend (the
mean value of dZt is zero);

o and the prices of different assets have different levels of uncertainty

(volatility)

The Geometric Brownian Motion (defined below) satisfies all the above requirements

We will use to model the price process for the underlying asset:

μ=expected return on stock per year

σ =volatility of the stock price per year

GEOMETRIC BROWNIAM MOTION PATH:

WHAT IS THE DISTRIBUTION FOR ST IF IT FOLLOWS A GBM?

In order to price options, we need to know what the distribution of the underlying asset
prices at expiration (ST) is.

If the stock price process is described by a Geometric Brownian Motion what is the
distribution of the future date t prices, ST?
 Because our price process incorporates uncertainty, we will never know, before
date t, what St will be;
 We will only be able to obtain the distribution of possible ST values;

We need to delve into “scary” mathematical concepts

FINAL ANSWER:
The price in the future date t is given by:

which is random because of the Brownian Motion Z t=√ t ε

Since ε ~N(0,1) we can work out the distribution of St as being:

That is St is lognormally distributed.

Where is r coming from? What happened to μ ?
If we look at the GBM and at the expression for S t we will notice that r (the risk-free
interest rate) has replaced μ (the asset’s expected rate of return):

Expression for St :

This happened because we jumped straight to the end of the story (missing the derivations
has this drawback)

We missed the part of the story where we switched to the risk neutral world:
 the reason why we need to be in the risk neutral world is the same as in the
binomial model;

BLACK – SCHOLES FORMULA

RISK-NEUTRAL VALUATION
In the risk-neutral world we can price a call option as:
where:
EQ [...∨F0 ] = expectation in the risk-neutral world (hence the Q) using the information
available at date 0 ( F 0 ¿

If we were not in the risk-neutral world we would need to discount at the expected rate of
return of the call, not r:
 but we don’t know the rate of return;
 and that’s why we price the option in the risk-neutral world where:

It is inconvenient to work with the max ( . ) operator, and so we will replace it by an

indicator function:

Therefore, we can write the call price as:

(I replace the maximum operate for ST-K:

The expectation of the indicator function (the second expectation) is the probability of the
event (ST > K) happening:
 if we have trouble seeing this, consider that S T has equal probability of assuming n
different values {ST,1, ST,2, … , ST, m, ST, m+1, …, ST, n} and that the first m of those values
are such that ST > K
m
 In that case the probability of ST >K is clearly and:
n

The second expectation is then:

 Probability of the event:

N the
cumulative distribution

where N( . ) is the standard normal cumulative distribution function (CDF) and:

The second equality follows from substitution of ST;

The third equality follows from taking logs on both sides;
The last equality follows from the fact that the standard normal distribution is symmetric
around 0:
 therefore, the probability of being above -d2 is the same probability of being below
d2;

Now, let’s deal with the first expectation:

The third equality follows from writing the expectation in its integral form, where
2
−ε
2
e is the standard normal pdf;
∅ (ε )=
√ 2 π
The last equality follows from the fact that S T > K when ε >−d 2, as we determined
previously.

Substituting for ∅ (ε ):

Making the change of variable θ=ε −σ √ T , which implies that dθ=dε :

where:

Plugging everything together, we obtain the Black-Scholes formula for a call option:
where:

If we repeat all of this for a put option, we obtain:

(For a Put, we need to change the payoff, because this isn’t symmetric)

Of course, that we can arrive at the formula for the put price using formula for the call
price and the put-call parity (remember, we are pricing European options).

COMPUTING N( . )

N(.) is the standard normal cumulative distribution function.

Table for N(x) when x≤0:

REPLICATING PORTFOLIO:
If we remember, call and put options can be replicated by taking a position ∆ on the
underlying asset and a position B0 on the risk-free asset;
The value of the option is then the value of the replicating portfolio:

Just by inspecting the Black-Scholes formula, we can determine the replicating portfolio of
a call:

For a put option:

So:

Q(ITM ¿¿ call )=Q(ST > K)=N (d 2 )¿

N(d2) = probability of exercising the call;

 PROBABILITY OF EXPIRING ITM
If we notice:

is the risk-neutral probability of a call option expiring ITM

Obviously, the probability of a put option expiring ITM (i.e. a call expiring OTM) is:

Black-Scholes partial differential equation: riskless and

replication portfolios

EXAMPLE 1 – Pricing a put option with Black-Scholes Model

Consider the following European put option:
 Maturity in 3 months;
 Strike of 5€;
 Underlying asset is Cimpor shares
 No dividend paid in the next 3 months;
 Contract size is 100;
 Cimpor’s spot price is 4.6€
 Volatility is 25%
 3-month risk-free rate is 1%

1. Price this put using the Black-Scholes formula;

2. Describe the replicating strategy for this option today;

When we used the binomial model, the answer to these questions were:

p0=0.502 € x 100=50.02 €

∆=−0.709 x 100=−70.9∧B 0=3.763 € x 100=376.3 €

The model parameters are:

We start by computing d2 and d1

Next, we look up on the tables the values closest to N(-d1 ) and N (-d2) and interpolate them:

Table for N(x) when x≥0:

Finally, we plug everything into the Black-Scholes formula:

The price of the put is then:

0.4807€ x 100 = 48.07€
The replicating portfolio is:
By the way, the risk-neutral probability of exercising this option at expiration is:
N ( - d2) = 0.7610 = 76.1%

Now, suppose the stock price 0.125 years from today is 5.025€:
 this corresponds to the price after one up move when we used the binomial model on Day 07

If we compute the price of the put in this scenario using the Black-Scholes formula using these
parameters:

we obtain:

But the value of the replicating portfolio is :

Shouldn’t these values coincide?

 yes and no;
 supposedly yes, because this is a replicating portfolio;
 but in fact no, because a lot of time has elapsed and the stock price has changed substantially;

The Black-Scholes model assumes that the replicating portfolio is continuously adjusted, that is,
after a very short time interval and after a very change in the price of the underlying asset:
 in practice, even daily adjustments on the replicating portfolio might not be frequent enough;

To work around this problem, we will have to hedge not only the delta (as we are doing here), but
other greeks.
VOLATILITY ESTIMATION: HISTORICAL VS. IMPLIED
All parameters needed to use the Black-Scholes formula are readily observable, except for
the annualized volatility of the underlying asset’s rate of return σ :

 K and the expiration date, from which we compute T, are written on the option
contract;
 S0 can be observed in the market;
 r can also be observed in the market, being the risk-free rate of return with a
maturity T (or close to it);

To use the Black-Scholes formula we will need to estimate σ .

Recall that assuming a GBM as the stock price process (and working with real-world
probabilities) we have: