Stochastic Calculus for Finance II

-
some Solutions to Chapter IV
Matthias Thul

Last Update: August 1, 2011

Exercise 4.1
This proof is fully analogous to the one of Theorem 4.2.1. We want to show that for
0 s t T
E[I(t)|F(s)] = I(s).
Assume again, that the s [t
l
, t
l+1
) and t [t
k
, t
k+1
) for l k. We start by splitting
up the sum into an F(s) measurable part and a part independent of F(s) and obtain
I(t) =
l1

j=0
(t
j
) [M (t
j+1
) M (t
j
)] + (t
l
) [M (t
l+1
) M (t
l
)]
+
k1

j=l+1
(t
j
) [M (t
j+1
) M (t
j
)] + (t
k
) [M(t) M (t
k
)]
The rst sum is F(s) measurable. The conditional expectation of the second term is
E[ (t
l
) [M (t
l+1
) M (t
l
)]| F(s)] = (t
l
) [E[ M (t
l+1
)| F(s)] M (t
l
)]
= (t
l
) [M(s) M (t
l
)] .
Here we have used that M (t
l
) is F(s) measurable. Furthermore, the martingale
property of M(t) implied that E[ M (t
l+1
)| F(s)] = M(s). The conditional expectation of
each element of the second sum is

The author can be contacted via <<firstname>>.<<lastname>>@googlemail.com and

http://www.matthiasthul.com.
1
E[ (t
j
) [M (t
j+1
) M (t
j
)]| F(s)] = E[ E[ (t
j
) [M (t
j+1
) M (t
j
)]| F (t
j
)]| F(s)]
= E[ (t
j
) [E[ M (t
j+1
)| F (t
j
)] M (t
j
)]| F(s)]
= E[ (t
j
) [M (t
j
) M (t
j
)]| F(s)]
= 0.
Here, we rst used the tower law of iterated conditioning, then took out all F (t
j
)
measurable terms from the inner conditional expectation and nally used the martingale
property of M(t) again. The same steps can be used to show that
E[ (t
k
) [M(t) M (t
k
)]| F(s)] = 0.
It follows that
E[ I(t)| F(s)] =
l1

j=0
(t
j
) [M (t
j+1
) M (t
j
)] + (t
l
) [M(s) M (t
l
)]
= I(s) (q.e.d.)
Exercise 4.4 (Stratonovich integral)
(i) Since the elements in the sum are independent, we start by computing the mean
and the variance of
_
W
_
t

j
_
W (t
j
)
_
2
. This is completely analogous to the proof
of Theorem 3.4.3.
E
_
_
W
_
t

j
_
W (t
j
)
_
2
_
= Var
_
W
_
t

j
_
W (t
j
)

= t

j
t
j
E
_
_
W
_
t

j
_
W (t
j
)
_
4
_
= 3E
_
_
W
_
t

j
_
W (t
j
)
_
2
_
2
= 3
_
t

j
t
j
_
2
Var
_
_
W
_
t

j
_
W (t
j
)
_
2
_
= 3
_
t

j
t
j
_
2

_
t

j
t
j
_
2
= 2
_
t

j
t
j
_
2
Here, we have used that the forth moment of a normal random variable is three
times its variance squared. The mean and variance of Q
/2
are then given by
E
_
Q
/2

=
n1

j=0
E
_
_
W
_
t

j
_
W (t
j
)
_
2
_
=
n1

j=0
t

j
t
j
=
n1

j=0
t
j+1
t
j
2
=
1
2
T
2
Var
_
Q
/2

=
n1

j=0
Var
_
_
W
_
t

j
_
W (t
j
)
_
2
_
=
n1

j=0
2
_
t

j
t
j
_
2
=
n1

j=0
1
2
(t
j+1
t
j
)
2

||||
2
n1

j=0
1
2
(t
j+1
t
j
)
lim
||||0
Var
_
Q
/2

= 0
1
2
T = 0 (q.e.d.)
Since the expected value of Q
/2
is the same for all partitions, but the variance as
limit zero when the maximum partition size gets smaller, we can conclude that Q
/2
has limit
1
2
T.
(ii) We expand the sum in the denition of the Stratonovich integral and add the terms
necessary to obtain an Ito integral over the partition Q
/2
.
_
T
0
W(t) dW(t)
= lim
||||0
n1

j+0
W
_
t

j
_
(W (t
j+1
) W (t
j
))
= lim
||||0
n1

j+0
_
W
_
t

j
_
W (t
j+1
) W
_
t

j
_
W (t
j
) W
_
t

j
_
2
2W (t
j
) W
_
t

j
_
W (t
j
)
2
_
= lim
||||0
n1

j+0
_
W
_
t

j
_
W (t
j+1
) W
_
t

j
_
2
_
+ lim
||||0
n1

j+0
_
W (t
j
) W
_
t

j
_
W (t
j
)
2
_
+ lim
||||0
n1

j+0
_
W (t
j
)
2
2W (t
j
) W
_
t

j
_
+ W
_
t

j
_
2
_
=
_
T
0
W(t)dW(t) + lim
||||0
n1

j+0
_
W (t
j
)
2
W
_
t

j
_
2
_
2
=
1
2
W
2
(T) (q.e.d.)
In the last step, we have used the result that
_
T
0
W(t)dW(t) =
1
2
W
2
(T)
1
2
T, as
e.g. shown in Example 4.3.2 and the quadratic variation of the Stratonovich integral
that we calculated in part (i).
3
Exercise 4.5 (Solving the generalized geometric Brownian motion
equation)
Let f(t, x) = ln x. We have
f
t
= 0,
f
x
=
1
x
,

2
f
x
2
=
1
x
2
and
(dS(t))
2
=
2
(t)S
2
(t)dt.
Applying Itos lemma, the dierential of the log stock price d ln S(t) becomes
d ln S(t) = df(t, S(t))
=
1
S(t)
dS(t)
1
2
1
S
2
(t)
(dS(t))
2
= (t)dt + (t)dW(t)
1
2

2
(t)dt
=
_
(t)
1
2

2
(t)
_
dt + (t)dW(t).
In integral form, we get
ln S(t) = ln S(0) +
_
t
0
_
(s)
1
2

2
(s)
_
ds +
_
t
0
(s)dW(s). (1)
Taking the exponential yields
S(t) = S(0) exp
__
t
0
_
(s)
1
2

2
(s)
_
ds +
_
t
0
(s)dW(s)
_
.
The distribution of S(t) will in general not be log-normal. In order to have log-
normality for S(t), we require ln S(t) to be normally distributed. By Theorem 4.4.9, the
It o integral in Equation (1) is only guaranteed to be normally distributed if (t) and (t)
are deterministic functions of time. In this case
ln S(t) N
__
t
0
_
(s)
1
2

2
(s)
_
ds,
_
t
0

2
(s)ds
_
.
Note that this includes the less general case of a Geometric Brownian Motion with
(t) = and (t) = , i.e the drift and diusion coecient being a constant as it is
assumed in the Black-Scholes model.
4
Exercise 4.6
Let f(t, x) = S(0)e
x
. We have
f
t
= 0,
f
x
= f(x),

2
f
x
2
= f(x).
Now dene
X(t) =
_

1
2

2
_
t + W(t).
By the It o formula, the dierential of S(t) is given by
dS(t) = df(t, X(t))
= S(t)dX(t) +
1
2
S(t)dX(t)dX(t)
= S(t)dt
1
2

2
S(t)dt + S(t)dW(t) +
1
2

2
S(t)dt
= S(t)dt + S(t)dW(t).
Now let f(t, x) = x
p
such that
f
t
= 0,
f
x
= px
p1
,

2
f
x
2
= p(p 1)x
p1
.
Using the It o formula to compute the dierential of d (S
p
(t)) yields
d (S
p
(t)) = d (f(t, S(t)))
= pS
p1
(t)dS(t) +
1
2
p(p 1)S
p2
(t)dS(t)dS(t)
= pS
p
(t)dt + pS
p
(t)dW(t) +
1
2

2
p(p 1)S
p
(t)dt
=
_
+
1
2
(p 1)
_
pS
p
(t)dt + pS
p
(t)dW(t).
Exercise 4.7
(i) Let f(t, x) = x
4
. We have
f
t
= 0,
f
x
= 4x
3
,

2
f
x
2
= 12x
2
.
5
Using the It o formula to compute the dierential of d (W
4
(t)) yields
d
_
W
4
(t)
_
= 4W
3
(t)dW(t) + 6W
2
(t)dt
W
4
(t) = 4
_
T
0
W
3
(t)dW(t) + 6
_
T
0
W
2
(t)dt
Note that there is in general a constant of integration in the second equation. But
since W
4
(0) = 0, we omitted it directly.
(ii) Taking expectations on both sides and using that an Ito integral is a martingale that
starts at zero yields
E
_
W
4
(t)

= 4E
__
T
0
W
3
(t)dW(t)
_
+ 6E
__
T
0
W
2
(t)dt
_
= 6
_
T
0
E
_
W
2
(t)

dt = 6
_
T
0
tdt
= 6

1
2
t
2

T
0
= 3T
2
(q.e.d.)
(iii) Let f(t, x) = x
6
. We have
f
t
= 0,
f
x
= 6x
5
,

2
f
x
2
= 30x
4
.
Using the It o formula to compute the dierential of d (W
6
(t)) yields
d
_
W
6
(t)
_
= 6W
5
(t)dW(t) + 15W
4
(t)dt
W
6
(t) = 6
_
T
0
W
5
(t)dW(t) + 15
_
T
0
W
4
(t)dt
Taking expectations on both sides and using the result in part (ii) yields
E
_
W
6
(t)

= 6E
__
T
0
W
5
(t)dW(t)
_
+ 30E
__
T
0
W
4
(t)dt
_
= 15
_
T
0
E
_
W
4
(t)

dt = 15
_
T
0
3t
2
dt
= 15

t
3

T
0
= 15T
3
6
Exercise 4.8 (Solving the Vasicek equation)
(i) Let f(t, x) = e
t
x. We have
f
t
= f(t, x),
f
x
= e
t
,

2
f
x
2
= 0.
Using the It o formula, we can compute the dierential of d
_
e
t
R(t)
_
to be
d
_
e
t
R(t)
_
= df(t, R(t))
= e
t
R(t)dt + e
t
dR(t)
= e
t
R(t)dt + e
t
dt e
t
R(t)dt + e
t
dW(t)
= e
t
dt + e
t
dW(t) (2)
(ii) Integrating Equation (2) yields
e
t
R(t) = R(0) +
_
t
0
e
u
du +
_
t
0
e
u
dW(u)
= R(0) +

e
u

t
0
+
_
t
0
e
bu
dW(u)
= R(0) +

_
e
t
1
_
+
_
t
0
e
bu
dW(u)
Thus, R(t) is given by
R(t) = e
t
R(0) +

_
1 e
t
_
+ e
t

_
t
0
e
bu
dW(u)
This is the same expression as given in Example 4.4.10.
Note that the solution strategy employed is very common for stochastic dierential
equations with a geometric drift term. Let X
(
t) = (t)X(t)dt + . . .. Then by
computing the dierential of e
(t)t
X
(
t), we can remove the geometric drift.
Exercise 4.11
When dynamically replicating an mispriced option in the Black-Scholes model, we have to
decide upon which volatility to use to compute the hedge ratio. In this case, the market
7
uses a volatility of
1
to price the option at all times t [0, T]. Knowing that all model
assumption hold but with the actual volatility of the underlying diusion being
2
>
1
,
we will enter a long position in the underpriced european call option and dynamically
delta hedge it by shorting (t) =
c
S
shares of the underlying. But since the delta itself
depends upon the volatility, there are two possibilities to do this: Either we use the
market implied volatility (
1
) or the actual volatility (
2
) to compute the hedge ratio.
These two approaches will in general lead to dierent returns. As implicitly suggested by
the exercise, we will use the delta calculated based upon the implied volatility to hedge
the long position in the european call option.
Let f(t, x) = e
rt
x. We have
f
t
= rf(t, x),
f
x
= e
rt
,

2
f
x
2
= 0.
Apply Itos lemma to the dierential of the discounted portfolio value d (e
rt
X(t))
d
_
e
rt
X(t)
_
= df(t, X(t))
= re
rt
X(t)dt + e
rt
dX(t)
= e
rt
(rX(t)dt + dX(t)) . (3)
We furthermore note that the dierential of the call option price dc(t, S(t)) evolves
according to
dc(t, S(t)) =
c
t
dt +
c
S
dS(t) +
1
2

2
c
S
2
(dS(t))
2
=
_
c
t
+
1
2

2
2
S
2
(t)

2
c
S
2
_
dt +
c
S
dS(t). (4)
Plugging the given dierential of the portfolio value and the dierential of the call
option price in Equation (4) back into Equation (3) yields
d
_
e
rt
X(t)
_
= e
rt
_
rX(t)dt +
_
c
t
+
1
2

2
2
S
2
(t)

2
c
S
2
_
dt +
c
S
dS(t)

c
S
dS(t) + rX(t)dt rc(t, S(t))dt + rS(t)
c
S
dt

1
2
_

2
2

2
1
_
S
2
(t)

2
c
S
2
dt
_
= e
rt
_
c
t
+
1
2

2
1
S
2
(t)

2
c
S
2
+ rS(t)
c
S
rc(t, S(t))
_
dt.
8
We note that the call price c(t, S(t)) is a specic solution to the Black-Scholes PDE
where the volatility is equal to
1
c
t
+ rS(t)
c
S
+
1
2

2
1
S
2
(t)

2
c
S
2
= rc(t, S(t))
for the payo-specic terminal condition c(T, S(T)) = max(S(T)K, 0), S(T) 0 and
the boundary conditions c(t, 0) = 0, 0 t T and lim
S(t)
_
c(t, S(t))
_
S(t) Ke
r(Tt)
__
=
0, 0 t T. Using this result, it directly follows that d (e
rt
X(t)) = 0.
Exercise 4.13 (Decomposition of correlated Brownian motions
into independent Brownian motions)
The dierential form of B
1
(t) and B
2
(t) are given by
dB
1
(t) = dW
1
(t)
dB
2
(t) = (t)dW
1
(t) +
_
1
2
(t)dW
2
(t).
This allows us to express the dierentials of W
1
(t) and W
2
(t) in terms of the changes
in the correlated Brownian motions B
1
(t) and B
2
(t)
dW
1
(t) = dB
1
(t)
dW
2
(t) =
(t)
_
1
2
(t)
dB
1
(t) +
1
_
1
2
(t)
dB
2
(t). (5)
We have to show that these two processes are independent Brownian motions. While it
is obvious that W
1
(t) is a Brownian motion since it is just equal to B
1
(t), we have to check
if this also holds for W
2
(t). According to Levys theorem, any continuous martingale M(t)
with M(0) = 0 and quadratic variation equal to [M, M](t) = t (i.e. dM(t)dM(t) = dt) is a
Brownian motion. Furthermore, according to the extended Levy theorem, two Brownian
motions, with zero cross variation (i.e. dW
1
(t)dW
2
(t) = 0) are independent.
(i) Continuity:
The continuity of W
2
(t) follows from the continuity of the two Brownian motions
9
B
1
(t) and B
2
(t). To see this explicitly, we write down the integral form of Equation
(5) to get
W
2
(t) = W
2
(0)
_
t
0
(s)
_
1
2
(s)
dB
1
(s) +
_
t
0
1
_
1
2
(s)
dB
2
(s). (6)
Since an Ito integral is a continuous function of its upper limit of integration, W
2
(t)
is continuous as well.
(ii) Martingale property:
The martingale property of W
2
(t) directly follows from the absence of a drift term
in Equation (5) or its representation as an Ito integral in Equation (6).
(iii) Starting at Zero:
Given the information in the exercise, we cant show that W
2
(0) = 0. When in-
tegrating Equation (5), we have to add a constant of integration that might take
any value in R. Similarly, we could argue that in the equation for dB
2
(t), only the
dierential dW
2
(t) and thus its initial value does not matter.
(iv) Unit quadratic variation:
The quadratic variation of W
2
(t) can be computed as
dW
2
(t)dW
2
(t) =

2
(t)
1
2
(t)
dt +
1
1
2
(t)
dt 2

2
(t)
1
2
(t)
dt
=
1
2
(t)
1
2
(t)
dt = dt (q.e.d.).
and it follows that [W
2
, W
2
](t) = t.
(v) Zero cross variation:
Using dB
1
(t)dB
1
(t) = dt and dB
1
(t)dB
2
(t) = (t)dt, the instantaneous cross varia-
10
tion between W
1
(t) and W
2
(t) can be computed as
dW
1
(t)dW
2
(t) =
(t)
_
1
2
(t)
dt +
(t)
_
1
2
(t)
dt = 0.
This establishes that W
1
(t) and W
2
(t) have zero cross-covariance. An alternative
proof of this result directly computes dB
1
(t)dB
2
(t)
dB
1
(t)dB
2
(t) = (t)dt +
_
1
2
(t)dW
1
(t)dW
2
(t).
This equation is only equal to (t)dt as given in the exercise, when
_
1
2
(t)dW
1
(t)dW
2
(t) = 0.
Since (1, 1), the square root term can never be equal to zero and we conclude
that dW
1
(t)dW
2
(t) = 0.
Exercise 4.15 (Creating correlated Brownian motions from inde-
pendent ones)
(i) Similar to Exercise 4.13, we use Levys theorem to show that B
i
(t) is a Brownian
motion for each i = 1, . . . , m. Since W
j
(0) = 0 for each j = 1, . . . , d, it directly
follows from the denition of B
i
(t) that B
i
(0) = 0. Furthermore, due to its repre-
sentation as a sum of Ito integrals, we obtain both the continuity and the martingale
property. It remains to show that B
i
(t) has quadratic variation [B
i
, B
i
](t) = t or
equivalently that dB
i
(t)dB
i
(t) = dt. In dierential form, B
i
(t) is
dB
i
(t) =
1

i
(t)
d

j=1

ij
(t)dW
j
(t).
Using the independence of the d-dimensional vector of Brownian motions, we get
dB
i
(t)dB
i
(t) =
1

2
i
(t)
d

j=1

2
ij
(t)dt =

d
j=1

2
ij
(t)

d
j=1

2
ij
(t)
dt = dt (q.e.d.).
11
(ii) Again using independence we get
dB
i
(t)dB
k
(t) =
1

i
(t)
k
(t)
d

j=1

ij

kj
(t)dt =
ik
(t)dt.
Exercise 4.16 (Creating independent Brownian motions to repre-
sent correlated ones)
In dierential form and using matrix notation, we have dB(t) = A(t)dW(t) or dW(t) =
A
1
(t)dB(t). We use the multidimensional Levy theorem (a straight forward extension
of Theorem 4.6.5) to show that W(t) is a vector of independent Brownian motions. We
need to check the following conditions:
Martingale property:
Since we can represent W(t) in terms of an integral constant plus and Ito integral
W(t) = W(0) +
_
t
0
A
1
(u)dB(u) W
i
(t) = W
i
(0) +
m

j=1
_
t
0
a
1
ij
(u)dB
j
(u)
it follows that each element W
i
(t) of this vector is a martingale (Theorem 4.3.1(iv)).
Continuity:
By Theorem 4.3.1(i), It o integrals have continuous sample paths.
Starting at zero:
Similar to Exercise 4.13(iii), we cant show that W(0) is a vector of zeros. In the
rst point, we could choose the integration constant W(0) arbitrarily as only the
change dW(t) is relevant in the denition of B(t). However, the exercise is only
asking for the existence of a vector of Brownian motions W(t). Thus, by setting
W(0) = 0 we can nd such a vector.
Unit quadratic variation and zero cross variation:
We need to show that each W
i
(t) has unit quadratic variation but zero cross variation
with any other W
j
(t) for j = i or equivalently that dW(t)dW
T
(t) = I where I is
the mm identity matrix. This can be easily done using matrix notation. We get
12
dW(t)dW
T
(t) = A
1
(t)dB(t)
_
A
1
(t)dB(t)
_
T
= A
1
(t)dB(t)dB
T
(t)A
T
(t)
= A
1
(t)C(t)A
T
(t)
= A
1
(t)A(t)A
T
(t)A
T
= I (q.e.d.).
Exercise 4.17 (Instantaneous correlation)
(i) We rst note that
B
1
(t
0
+ ) B
2
(t
0
+ ) = B
1
(t
0
) B
2
(t
0
) +
_
t
0
+
t
0
d (B
1
(s)B
2
(s)) .
By Corollary 4.6.3, we have for the dierential of the product of the two Brownian
motions
d (B
1
(t)B
2
(t)) = B
2
(t)dB
1
(t) + B
1
(t)dB
2
(t) + dB
1
(t)dB
2
(t)
= B
2
(t)dB
1
(t) + B
1
(t)dB
2
(t) + dt.
Thus,
B
1
(t
0
+ ) B
2
(t
0
+ ) = B
1
(t
0
) B
2
(t
0
)+
_
t
0
+
t
0
B
2
(s)dB
1
(s)+
_
t
0
+
t
0
B
1
(s)dB
2
(s)+
and
E[ B
1
(t
0
+ ) B
2
(t
0
+ )| F (t
0
)] = B
1
(t
0
) B
2
(t
0
) +
where we used that the two integrals are independent of the ltration F (t
0
) and
that by Theorem 4.3.1(iv) It o integrals are martingales. Finally,
13
E[ (B
1
(t
0
+ ) B
1
(t
0
)) (B
2
(t
0
+ ) B
2
(t
0
))| F (t
0
)]
= E[ B
1
(t
0
+ ) B
2
(t
0
+ )| F (t
0
)] E[ B
1
(t
0
+ ) B
2
(t
0
)| F (t
0
)]
E[ B
1
(t
0
) B
2
(t
0
+ )| F (t
0
)] +E[ B
1
(t
0
) B
2
(t
0
)| F (t
0
)]
= B
1
(t
0
) B
2
(t
0
) + B
2
(t
0
) E[B
1
(t
0
+ )]
B
1
(t
0
) E[B
2
(t
0
+ )] + B
1
(t
0
) B
2
(t
0
)
= . (q.e.d.)
(ii) The means are
M
i
() = E[ X
i
(t
0
+ ) X
i
(t
0
)| F (t
0
)]
=
_
t
0
+
t
0

i
du
=
i
(q.e.d.).
Where we again used that the Riemann integral is deterministic and that the It o
integral is a martingale with zero expected value.
The variances are
V
i
() = E
_
(X
i
(t
0
+ ) X
i
(t
0
))
2

F (t
0
)

M
2
i
()
= E
_
__
t
0
+
t
0

i
du +
_
t
0
+
t
0

i
dB
i
(u)
_
2
_

2
i

2
=
__
t
0
+
t
0

i
du
_
E
__
t
0
+
t
0

i
dB
i
(u)
_
+E
_
__
t
0
+
t
0

i
dB
i
(u)
_
2
_
=
_
t
0
+
t
0

2
i
du
=
2
i
(q.e.d.).
Here, we used that the increment in the Ito processes is independent of the ltration
F (t
0
) in the second step. We can take the deterministic time integral out of the
expectation in the third step and again reply on the martingale property of the Ito
integral in the fourth step to eliminate this term. We further use the It o isometry
14
(Theorem 4.3.1(v)) to transform the squared Ito integral into a Riemann integral in
the fourth step.
The covariance is
C() = E[ (X
1
(t
0
+ ) X
1
(t
0
)) (X
2
(t
0
+ ) X
2
(t
0
))| F (t
0
)] M
1
()M
2
()
= E
___
t
0
+
t
0

1
du +
_
t
0
+
t
0

1
dB
1
(u)
___
t
0
+
t
0

2
du +
_
t
0
+
t
0

2
dB
2
(u)
__

2
=
__
t
0
+
t
0

1
du
_
E
__
t
0
+
t
0

2
dB
2
(u)
_
+
__
t
0
+
t
0

2
du
_
E
__
t
0
+
t
0

1
dB
1
(u)
_
+E
___
t
0
+
t
0

1
dB
1
(u)
___
t
0
+
t
0

2
dB
2
(u)
__
= E
___
t
0
+
t
0

1
dB
1
(u)
___
t
0
+
t
0

2
dB
2
(u)
__
In order to solve the remaining expectation, we rst dene for t t
0
I
i
(t) =
_
t
t
0

i
dB
i
(u) dI
i
(t) =
i
dB
i
(t)
By the It o product rule, we have
d (I
1
(t)I
2
(t)) (t) = I
2
(t)dI
1
(t) + I
1
(t)dI
2
(t) + dI
1
(t)dI
2
(t)
or
I
1
(t)I
2
(t) = I
1
(t
0
) I
2
(t
0
) +
_
t
t
0
d (I
1
(u)I
2
(u))
=
_
t
t
0
I
2
(u)dI
1
(u) +
_
t
t
0
I
1
(u)dI
2
(u) +
_
t
t
0

2
du.
Thus,
C() = E[I
1
(t
0
+ ) I
2
(t
0
+ )] =
_
t
0
+
t
0

2
du =
1

2
. (q.e.d.).
Consequently, we have for the correlation between the increments of the It o processes
C()
_
V
1
()V
2
()
=

1

2
1

2
2

2
= .
15
(iii) We have
M
i
() = E[ X
i
(t
0
+ ) X (t
0
)| F (t
0
)]
= E
__
t
0
+
t
0

i
(u)du +
_
t
0
+
t
0

i
(u)dB
i
(u)

F (t
0
)
_
= E
__
t
0
+
t
0

i
(u)du

F (t
0
)
_
.
We rst note that by Denition 1.2.1, the real constant M can be considered a
(trivial) random variable as well. Here {M B} = , i.e. we do not learn anything
about the particular outcome by observing M. We now apply the dominated
convergence theorem (Theorem 1.4.9) to the sequence of random variables
I() =
1

_
t
0
+
t
0

i
(u)du
that is bounded in absolute value by M. We have
lim
0
1

M
i
() = lim
0
E[ I()| F (t
0
)]
= E
_
lim
0
I()

F (t
0
)
_
= E
_
lim
0
1

[I (0) + I

(0) + o ()]

F (t
0
)
_
= E[
i
(t
0
)| F (t
0
)]
=
i
(t
0
) (q.e.d.).
Here, we used a Taylor series expansion of the integral around = 0 where I(0) = 0
and I

(0) =
i
(t
0
).
(iv) Similar to the hint, we dene for t > t
0
Y
i
(t) =
_
t
t
0

i
(u)dB
i
(u)
and get
16
D
ij
() = E[ (X
i
(t
0
+ ) X
i
(t
0
)) (X
j
(t
0
+ ) X
j
(t
0
))| F (t
0
)] M
i
()M
j
()
= E
__
Y
i
(t
0
+ ) +
_
t
0
+
t
0

i
(u)du
__
Y
j
(t
0
+ ) +
_
t
0
+
t
0

j
(u)du
_

F (t
0
)
_
M
i
()M
j
()
= E[ Y
i
(t
0
+ ) Y
j
(t
0
+ )| F (t
0
)] +
__
t
0
+
t
0

j
(u)du
_
E[ Y
i
(t
0
+ )| F (t
0
)]
+
__
t
0
+
t
0

i
(u)du
_
E[ Y
j
(t
0
+ )| F (t
0
)]
= E[ Y
i
(t
0
+ ) Y
j
(t
0
+ )| F (t
0
)]
Here, we followed the same main steps as in the derivation in (ii). We now apply
the Ito product rule again to compute the remaining expectation and get
D
ij
() = E
__
t
0
+
t
0
Y
j
(u)dY
i
(u) +
_
t
0
+
t
0
Y
i
(u)dY
j
(u) +
_
t
0
+
t
0
dY
i
(u)dY
j
(u)

F (t
0
)
_
= E
__
t
0
+
t
0

ij
(u)
i
(u)
j
(u)du

F (t
0
)
_
.
We now apply the same limit argument as in (iii) and start by dening the sequence
of random variables
I
ij
() =
1

_
t
0
+
t
0

ij
(u)
i
(u)
j
(u)du
that is bounded in absolute value by M
3
. We get
lim
0
1

D
ij
() = lim
0
E[ I
ij
()| F (t
0
)]
= E
_
lim
0
I()

F (t
0
)
_
= E
_
lim
0
1

[I(0) + I

(0) + o()]

F (t
0
)
_
= E[
ij
(t
0
)
i
(t
0
)
j
(t
0
)| F (t
0
)]
=
ij
(t
0
)
i
(t
0
)
j
(t
0
) (q.e.d.).
(v) We showed this already implicitly in the derivation of the means and covariances.
We have
17
D
ij
() = E[ I
ij
()| F (t
0
)]
= E[ I(0) + I

(0) + o()| F (t
0
)]
= E[
ij
(t
0
)
i
(t
0
)
j
(t
0
) + o()| F (t
0
)]
=
ij
(t
0
)
i
(t
0
)
j
(t
0
) + o()
=
_

_
V
i
() =
2
k
(t
0
) + o() for i = j = k
C() = (t
0
)
1
(t
0
)
2
(t
0
) + o() for i = j
(q.e.d.).
(vi) Finally,
lim
0
C()
_
V
1
()V
2
()
= lim
0
(t
0
)
1
(t
0
)
2
(t
0
) + o()
_
(
2
1
(t
0
) + o()) (
2
2
(t
0
) + o())
= lim
0
(t
0
)
1
(t
0
)
2
(t
0
)
_

2
1
(t
0
)
2
2
(t
0
)
+ o()
= (t
0
) (q.e.d.).
Exercise 4.18
(i) Let f(t, x) = exp
_
x
_
r +
1
2

2
_
t
_
. We have
f
t
=
_
r +
1
2

2
_
f(t, x),
f
x
= f(t, x),

2
f
x
2
=
2
f(t, x).
It os lemma yields
d(t) = df(t, W(t))
=
_
r +
1
2

2
_
(t)dt (t)dW(t) +
1
2

2
(t)dt
= r(t)dt (t)dW(t) (q.e.d.).
(ii) By It os product rule (Corollary 4.6.3) we get
18
d ((t)X(t)) = (t)dX(t) + X(t)d(t) + dX(t)d(t)
= r(t)X(t)dt + (t)(t)( r)S(t)dt + (t)(t)S(t)dW(t)
(t)X(t)dW(t) r(t)X(t)dt (t)(t)S(t)dt
= (t) ((t)S(t) X(t)) dW(t). (7)
In the last step, we have used the fact that = r. Equation (7) shows that
(t)X(t) is a martingale, since all dt-terms vanish in the dierential. Equivalently, we
could express (t)X(t) by a single It o integral. Since all It o integrals are martingales,
so is (t)X(t)
(t)X(t) = (0)X(0) +
_
t
0
(s) ((s)S(s) X(s)) dW(s).
(iii) First, note that
(0) = exp {W(0)} = 1.
Now let (t) be an adapted portfolio process such that, X(T) = V (T), i.e. the
nal value of the trading strategy is equal to the nal value of the F(T) measurable
random variable V (T). As has been shown in (ii), (t)X(t) is a martingale, i.e.
E[(t)X(t)|F(s)] = (s)X(s), s t.
We obtain
X(0) = (0)X(0) = E[(T)X(T)] = E[(T)V (T)] .
Exercise 4.19
(i) By Levys theorem it is sucient to show that B(t) is a continuous martingale start-
ing at zero with unit quadratic variation in order to establish that it is a Brownian
19
motion. We rst note that B(t) is an Ito integral with B(0) = 0. By Theorem 4.3.1,
It o integrals are continuous martingales. The quadratic variation of B(t) is
(dB(t))
2
= (sign (W(t)) dW(t))
2
= sign
2
(W(t)) (dW(t))
2
= dt
and it follows that it is a Brownian motion.
(ii) We have
d (B(t)W(t)) = B(t)dW(t) + W(t)dB(t) + dB(t)dW(t)
= B(t)dW(t) + W(t)sign (W(t)) dW(t) + sign (W(t)) dt
and since B(0)W(0) = 0 we get
B(t)W(t) =
_
t
0
B(s)dW(s) +
_
t
0
W(s)sign (W(s)) dW(s) +
_
t
0
sign (W(s)) ds
Since Ito integrals are martingales it follows that
E[B(t)W(t)] = E
__
t
0
sign (W(s)) ds
_
=
_
t
0
E[sign (W(s))] ds = 0.
In the last step, we have used that with probability one half, sign (W(t)) = 1 for
t > 0.
(iii) Let f(t, x) = x
2
. We have
f
t
= 0,
f
x
= 2x,

2
f
x
2
= 2.
Using the It o formula to compute the dierential of dW
2
(t) yields
dW
2
(t) = 2W(t)dW(t) + dt (q.e.d.)
20
(iv) We have
d
_
B(t)W
2
(t)
_
= B(t)dW
2
(t) + W
2
(t)dB(t) + dB(t)dW
2
(t)
= 2B(t)W(t)dW(t) + B(t)dt + W
2
(t)sign (W(t)) dW(t)
+2W(t)sign (W(t)) dt.
Integrating both sides yields
B(t)W
2
(t) =
_
t
0
2B(s)W(s)dW(s) +
_
t
0
B(s)ds +
_
t
0
W
2
(s)sign (W(s)) dW(s)
+
_
t
0
2W(s)sign (W(s)) ds.
We take expectation on both sides and again use the martingale property of the It o
integral to get
E
_
B(t)W
2
(t)

= E
__
t
0
B(s)ds +
_
t
0
2W(s)sign (W(s)) ds
_
(8)
In order to evaluate the rst Riemann integral, we set f(t, x) = tx such that f
t
(t, x) =
x and f
x
(t, x) = t. We apply the It o formula to obtain the dierential of tW(t) as
d (tB(t)) = B(t)dt + tdB(t).
Integrating and rearranging yields
_
T
0
B(t)dt = TB(T)
_
T
0
tdB(t) =
_
T
0
(T t)dB(t).
The last term is an It o integral of a deterministic integrand and thus normally
distributed with zero mean and variance
_
T
0
(T t)
2
dt =
1
3
(T t)
3

T
0
=
1
3
T
3
.
It follows that the rst term in Equation (8) is zero and we get
E
_
B(t)W
2
(t)

= E
__
t
0
2W(s)sign (W(s)) ds
_
= E
__
t
0
2 |W(s)| ds
_
21
It is obvious, that the expected value is strictly positive for t > 0 since the integrand
is non-negative and almost surely strictly positive. If B(t) and W(t) are independent,
then by Theorem 2.2.5 any two function f (B(t)) and g (W(t)) are independent. Now
let g (B(t)) = B(t) and f (W(t)) = W
2
(t) then by Theorem 2.2.7 independence
implies that E[B(t)W
2
(t)] = E[B(t)] E[W
2
(t)]. But since E[B(t)] = 0 we would
require E[B(t)W
2
(t)] = 0. We have shown above that E[B(t)W
2
(t)] > 0 and thus
B(t) and W(t) are not independent.
Exercise 4.20 (Local time)
(i)
f

(x) =
_

_
1 if x > K
undened if x = K
0 x < K
f

(x) =
_

_
0 if x = K
undened if x = K
(ii)
f(W(T)) = f(W(0)) +
_
T
0
I
{W(t)>K}
dW(t) (9)
Using that by Theorem 3.3.2(i), W(T) N(0, T), we can compute the expected
value of the right hand side as
22
E[f(W(T))] = E
_
(W(T) K)
+

=
_

Tx K
_
+
N

(x)dx
=
_

K

T
_

Tx K
_
N

(x)dx
=

T
_

K

T
xN

(x)dx KN
_

T
_
=
_
T
2
_

K

T
xe

x
2
2
dx KN
_

T
_
=
_
T
2
e

x
2
2

T
KN
_

T
_
=
_
T
2
e

K
2
2T
KN
_

T
_
Since the It o integral is a martingale starting at zero, it has zero expected value and
we get for the right hand side
E[f(W(0))] +E
__
T
0
I
{W(t)>K}
dW(t)
_
= E
_
(W(0) K)
+

= (0 K)
+
= 0.
This shows that Equation (9) does not hold.
(iii) The answer to this question is pretty much given in the hints already.
f

(x) =
_

_
0 if x K
1
2n
n(x K) +
1
2
if K
1
2n
x K +
1
2n
1 if x K +
1
2n
f

(x) =
_

_
0 if x < K
1
2n
n if K
1
2n
< x < K +
1
2n
0 if x > K +
1
2n
23
(iv) Similar to (iii), there is not too much to add to the hints. We rst note that the
intervals of the piecewise function denition have limits
lim
n
_
0, K
1
2n
_
= [0, K)
lim
n
_
K
1
2n
, K +
1
2n
_
= K
lim
n
_
K +
1
2n
,
_
= (K, ).
Thus,
lim
n
f
n
(x) =
_

_
0 if x K
x K if x K
= (x K)
+
lim
n
f

n
(x) =
_

_
0 if x < K
1
2
if x = K
1 if x > K
. (10)
(v) Let M(T) = max
0tT
W(T) be the maximum of a xed Brownian motion path
during the time interval [0, T]. By assumption, M(T) < K and by denition
L
K
(T) = lim
n
n
_
T
0
I
{
W(t)
(
K
1
2n
,K+
1
2n
)}
dt.
Now, from M(T) < K it follows that there exists an n
0
such that K
1
2n
> M(T) for
all n n
0
. Consequently, for all n n
0
, the indicator function inside the integral
evaluates to zero for all t [0, T] and we get L
K
(T) = 0.
(vi) Taking expectation on both sides and using that by Theorem 4.3.1(iv) the It o integral
is a martingale with zero expected value yields
E[L
K
(T)] = E
_
(W(T) K)
+

> 0,
24
where the inequality follows from our result in (ii). Although it is not immediately
obvious from the nal result, it is enough to note that by the normal distribution of
W(T) values of W(T) > K have a strictly positive probability for all nite levels of
K and thus the expectation is strictly positive as well. Now,
P{L
K
(T) = 0} = 1 E[L
K
(T) = 0]
and just showed that the implication does not hold. Thus, by contradiction we
cannot have that P{L
K
(T) = 0} = 1.
Exercise 4.21 (Stop-loss start-gain paradox)
(i) The stop-loss start-gain strategy involves holding a long position in one share when-
ever the stock price strictly exceeds the strike price and holding no position other-
wise. The crucial point is, that buys and sells are not made at the same price and
thus the hedger incurs a cost every time he enters and unwinds the stock position.
To make it clear, assume that we by one stock whenever the price level K + is
crossed from below (and we had no position before) and sell one stock whenever the
price level K is crossed from above (and we were long one stock before). Building up
and subsequently unwinding the stock position thus yields to a loss of . Since the
mathematical model of a geometric Brownian motion is a process with continuous
sample paths, can theoretically made arbitrarily small. However, smaller values of
lead to more frequent transactions.
Real world stock prices however dont take values in R but are usually quoted in full
USD-cents. Furthermore, they dont move continuously but jump. Thus, each time
the stock price crosses the strike price the hedger encounters a friction of at least
one USD-cent. Furthermore, the strategy ignores transaction cost that can become
very signicant for high number of buys and sells in the stock.
(ii) Taking expectation and using that by Theorem 4.3.1(i) the Ito integral is a martin-
gale with zero expected value yields E[X(T)] = 0. However,
25
E
_
(S(T) K)
+

=
_

_
S(0) exp
_

1
2

2
T +

Tx
_
K
_
+
N

(x)dx
=
_

ln
(
S(0)
K
)

1
2

T
_
S(0) exp
_

1
2

2
T +

Tx
_
K
_
N

(x)dx
> 0.
From the second equation and the properties of the normal distribution it is obvious
that the integrand is non-negative everywhere and that it takes strictly positive
values for some values of x. Thus, the expected value is strictly positive as well. In
conclusion
P
_
X(T) = (S(T) K)
+
_
< 1,
i.e. there are some paths where the stop-loss start-gain strategy does replicate the
option payo (i.e. all paths where the maximum stock price over the interval [0, T]
is strictly less than K). However, since the expected value of the payo of the
strategy does not agree with the expected value of the option payo, we do not have
X(T) = (S(T) K)
+
almost surely. This also follows from the result on the local
time in Exercise 4.21. Since a Brownian motion spends a non-zero time at any level
with positive probability, so does it at the strike K. But once the strike is crossed,
the hedger incurs a cost and the exact replication fails.
26