Sei sulla pagina 1di 4

Monte Carlo Methods in Finance

Homework: Chapter 7
Please enter your answers in the homework unit at the end of the Chapter.
The solutions to this homework will be posted in a separate unit after the due
date.
1. Consider an asset whose initial price is S(t
0
) = S
0
. Let C(S
0
, K, r, T, )
be the pricing formula of a European call option on this asset in the
Black-Scholes model. The Octave/MATLAB function that implements
this pricing formula is priceEuropeanCall. It has been distributed with
Chapter 7 (see Unit 7.0). The closed form expression for the delta of this
option in the Black-Scholes model is
=
C(S
0
, K, r, T, )
S
0
= (d
+
),
where
d

=
1

T
log
_
S
0
Ke
rT
_

1
2

T
and (z) is the cdf of the standard normal (normcdf).
Write an Octave/MATLAB function to compute the value of delta using
this expression. The header of this function, deltaEuropeanCall, should
be
function delta = deltaEuropeanCall(S0,K,r,T,sigma)
%% deltaEuropeanCall: Delta of a European call option
%
%% SYNTAX:
% delta = deltaEuropeanCall(S0,K,r,T,sigma)
%
%% INPUT:
% S0 : Initial value of the underlying asset
% K : Strike
% r : Risk-free interest rate
% T : Time to expiry
% sigma : Volatility
%
%% OUTPUT:
% delta : Delta of the option in the Black-Scholes model
1
%
%% EXAMPLE:
% S0 = 100; K = 90; r = 0.05; T = 2; sigma = 0.4;
% delta = deltaEuropeanCall(S0,K,r,T,sigma)
What is the delta of a European call option of parameters S
0
= 100,
K = 90, r = 0.05, T = 2, = 0.4?
1.2653
0.7408
0.5327
0.2113
-0.1749
2. An alternative method to compute the of the option is to numerically
approximate the derivative using nite dierences

num
=
C(S
0
(1 +h), K, r, T, ) C(S
0
(1 h), K, r, T, )
2S
0
h
.
Consider the European call option of the previous exercise. Which value
of h yields the lowest approximation error?
h = 10
1
h = 10
3
h = 10
5
h = 10
7
h = 10
9
Hint: To compute the option price use the function priceEuropeanCall.
To compute the error of
num
, use the output of the function deltaEuropeanCall
as the exact delta.
3. Consider the call option from the previous exercises. We will now approx-
imate the delta of the option by the expression

MC
=
C
MC
(S
0
(1 +h), K, r, T, ) C
MC
(S
0
(1 h), K, r, T, )
2S
0
h
,
where C
MC
(S
0
, K, r, T, ) is a Monte Carlo (MC) estimate of the price of
the option. To compute this price estimate use the function priceEuropeanCallMC.
To compute the numerical approximation to the derivative use the h that
you determined in the previous exercise. What is the relative error of this
MC estimate of the delta of the option using M = 10
5
simulations?
Relative error 0.16%
2
Relative error 1.6%
Relative error 16%
Relative error 160%
Relative error 16, 000%
Hint: The Monte Carlo estimate of the price of the option is itself a ran-
dom number and therefore exhibits uctuations. Since we are not using
common random numbers, the uctuations in the two terms of the numer-
ator of the numerical dierentiation formula are independent. Therefore,
the variance of this Monte Carlo estimate of the delta is

2
[
MC
] =

2
[C
MC
(S
0
(1 +h), ...)] +
2
[C
MC
(S
0
(1 +h), ...)]
(2S
0
h)
2
,
the sum of the variances of each price estimator divided by the factor
in the denominator squared. Use this expression to obtain [
MC
], the
standard deviation of
MC
, which we will take as the absolute error of
the MC estimate of the delta. Compute the relative error as the ratio
between the absolute error and the exact delta, which is the output of the
function deltaEuropeanCall.
4. In this exercise we will consider again the problem of computing the delta
of the option by numerical dierentiation of the Monte Carlo price es-
timates. However, this time, we will use the same random numbers to
compute C
MC
(S
0
(1 +h), K, r, T, ) and C
MC
(S
0
(1 h), K, r, T, ). Since
we are using common random numbers to compute the terms that are in-
volved in the subtraction that appears in the numerator, the uctuation
errors are highly correlated and approximately cancel.
The procedure is as follows:
Compute the delta along the mth trajectory

(m)
=
payo(S
(m)
+
(t
0
+T)) payo(S
(m)

(t
0
+T))
2S
0
h
,
where
S
(m)

(t
0
+T) = S
0
(1 h) exp
__

1
2

2
_
T +

TX
(m)
_
X
(m)
N(0, 1), m = 1, 2, . . . , M
and, for a European call option,
payo(S
T
) = max(S
T
K, 0).
Use the h that you determined in a previous exercise.
3
Estimate the delta by averaging over the simulated trajectories and
then multiplying by the discount factor

MC
= e
rT
1
M
M

m=1

(m)
.
The standard deviation of the Monte Carlo estimate of the delta is
stdev
MC
= e
rT
1

M
stdev
_
_

(m)
_
M
m=1
_
.
Using common random numbers, what is the relative error of the MC
estimate of the delta of a European option with parameters S
0
= 100,
K = 90, r = 0.05, T = 2, = 0.4, with M = 10
5
simulations?
Relative error 0.35%
Relative error 3.5%
Relative error 35%
Relative error 350%
Relative error 35, 000%
5. In this exercise we analyze the eectiveness of stratied sampling to re-
duce the variance of the Monte Carlo estimate of the price of an option.
Consider a European call option with parameters S
0
= 100, K = 90,
r = 0.05, T = 2, = 0.2. With respect to standard Monte Carlo, the
standard deviation of the option price estimate using M = 10
4
simulations
when these simulations are stratied in 5 strata is reduced by a factor of
approximately
0.5
1.5
2.5
3.5
Hint: Look for help on priceEuropeanCallMC_stratifiedSampling.m
4

Potrebbero piacerti anche