Stochastic Models in Finance

Solutions to Example Sheet 2

1. The random variable X takes values in {1, 2, 3, 4}.
A1 = {ω ∈ Ω : X(ω) = 1} = {ω1 , ω2 }
A2 = {ω ∈ Ω : X(ω) = 2} = {ω3 , ω4 }
A3 = {ω ∈ Ω : X(ω) = 3} = {ω5 , ω6 }
A4 = {ω ∈ Ω : X(ω) = 4} = {ω7 , ω8 }
{A1 , A2 , A3 , A4 } = P is a partition of Ω. So

F(X) ={φ, Ω, {ω1 , ω2 }, {ω3 , ω4 }, {ω5 , ω6 }, {ω7 , ω8 },

{ω1 , ω2 , ω3 , ω4 }, {ω1 , ω2 , ω5 , ω6 }, {ω1 , ω2 , ω7 , ω8 },
{ω3 , ω4 , ω5 , ω6 }, {ω3 , ω4 , ω7 , ω8 }, {ω5 , ω6 , ω7 , ω8 },
{ω1 , ω2 , ω3 , ω4 , ω5 , ω6 }, {ω1 , ω2 , ω3 , ω4 , ω7 , ω8 },
{ω1 , ω2 , ω5 , ω6 , ω7 , ω8 }, {ω3 , ω4 , ω5 , ω6 , ω7 , ω8 }},

2. Note A1 = {ω ∈ Ω : S(0) = 5} = Ω
So F0 = F(S(0)) = {φ, Ω}.
Note A11 = {ω ∈ Ω : S(0, ω) = 5, S(1, ω) = 6} = {ω1 , ω2 }
A12 = {ω ∈ Ω : S(0, ω) = 5, S(1, ω) = 4} = {ω3 , ω4 }
So F1 = {φ, Ω, {ω1 , ω2 }, {ω3 , ω4 }} and the corresponding partition is
P1 = {{ω1 , ω2 }, {ω3 , ω4 }}
Note
A111 = {ω : S(0, ω) = 5, S(1, ω) = 6, S(2, ω) = 7} = {ω1 }
A112 = {ω : S(0, ω) = 5, S(1, ω) = 6, S(2, ω) = 8} = {ω2 }
A123 = {ω : S(0, ω) = 5, S(1, ω) = 4, S(2, ω) = 6} = {ω3 }
A124 = {ω : S(0, ω) = 5, S(1, ω) = 4, S(2, ω) = 2} = {ω4 }

So
F2 ={φ, Ω, {ω1 }, {ω2 }, {ω3 }, {ω4 }, {ω1 , ω2 },
{ω1 , ω3 }, {ω1 , ω4 }, {ω2 , ω3 }, {ω2 , ω4 }, {ω3 , ω4 },
{ω1 , ω2 , ω3 }, {ω1 , ω2 , ω4 }, {ω1 , ω3 , ω4 }, {ω2 , ω3 , ω4 }}
The corresponding minimal partition contained in this σ−field is

P2 = {{ω1 }, {ω2 }, {ω3 }, {ω4 }}

The filtration {F0 , F1 , F2 }, Ft : t = 0, 1, 2 defined above, is generated by the stochastic

price process S(t, ω).

1
 Consider Y (t) = S 2 (t) + S(t). So Y (0) = 30

 42 ω = ω1 , ω2
Y (1, ω) =
20 ω = ω3 , ω4


56 ω = ω1






 72 ω = ω2


Y (2, ω) =
42 ω = ω3








6 ω = ω4

It is obvious that
{ω : Y (0) = 30} = Ω ∈ F0 , and {ω : Y (0) = a} = φ ∈ F0 for any a 6= 30. So Y (0) is
measurable with respect to F0 .
{ω : Y (1) = 42} = {ω1 , ω2 } ∈ F1
{ω : Y (1) = 20} = {ω3 , ω4 } ∈ F1 , {ω : Y (1) = a} = φ ∈ F1 for any a 6= 42, 20.
So Y (1) is measurable with respect to F1
{ω : Y (2) = 56} = {ω1 } ∈ F2
{ω : Y (2) = 72} = {ω2 } ∈ F2
{ω : Y (2) = 42} = {ω3 } ∈ F2
{ω : Y (2) = 6} = {ω4 } ∈ F2 , and {ω : Y (2) = a} = φ ∈ F1 for any a 6= 56, 72, 42, 6.
So Y (2) is measurable with respect to F2
2
Therefore, Y (t) is adapted to filtration {Ft }t=0

3. The value process


 H0 (1)B(0) + H1 (1)S(0), t=0
V (t) =
H0 (t)B(t) + H1 (t)S(t), t ≥ 1.


So
V (0) = H0 (1) + 5H1 (1)
V (1, ω) = H0 (1)(1 + r) + 6H1 (1) ω = ω1 , ω2
V (1, ω) = H0 (1)(1 + r) + 4H1 (1) ω = ω3 , ω4
V (2, ω) = H0 (2)(1 + r)2 + 7H1 (2) ω = ω1
V (2, ω) = H0 (2)(1 + r)2 + 8H1 (2) ω = ω2
V (2, ω) = H0 (2)(1 + r)2 + 6H1 (2) ω = ω3
V (2, ω) = H0 (2)(1 + r)2 + 2H1 (2) ω = ω4

2
Gain process

 rH0 (1) + H1 (1). ω = ω1 , ω2
G(1, ω) =
rH0 (1) − H1 (1), ω = ω3 , ω4


rH (1) + H1 (1) + (r2 + r)H0 (2) + H1 (2), ω = ω1


 0

rH0 (1) + H1 (1) + (r2 + r)H0 (2) + 2H1 (2), ω = ω2
G(2, ω) = 2
 rH0 (1) − H1 (1) + (r2 + r)H0 (2) + 2H1 (2),
 ω = ω3
rH0 (1) − H1 (1) + (r + r)H0 (2) − 2H1 (2), ω = ω4

The discounted value process V ∗ (t) = V (t)/B(t).

So V ∗ (0) = H0 (1) + 5H1 (1)
 6
 H0 (1) +
 H1 (1) , ω = ω1 , ω2
∗ 1+r
V (1, ω) =
 H (1) + 4 H (1) ,

ω = ω3 , ω4
0 1
1+r

 7
 H0 (2) + H (2)
2 1
, ω = ω1
(1 + r)




8



 H0 (2) + (1 + r)2 H1 (2) , ω = ω2


V ∗ (2, ω) =
6
H0 (2) + H1 (2) , ω = ω3


(1 + r)2







 H0 (2) + 2
H1 (2) , ω = ω4

(1 + r)2

The discounted gain process

 1 − 5r
 H1 (1) , ω = ω1 , ω2
1+r

∗
G (1, ω) =
 − 1 + 5r H (1) ,

ω = ω3 , ω4
1
1+r

 1 − 5r 1 − 6r
 H 1 (1) + H (2),
2 1
ω = ω1
1 + r (1 + r)




1 − 5r 2 − 6r



 1 + r H1 (1) + (1 + r)2 H1 (2), ω = ω2


G∗ (2, ω) =
1 + 5r 2 − 4r
− H1 (1) + H1 (2), ω = ω3


(1 + r)2



 1+r

 − 1 + 5r H1 (1) − 2 + 4r H1 (2),


ω = ω4

1+r (1 + r)2

3
4. If H is a self-financing trading strategy, then
V (0) + G(t)

t
X t X
X N
= V (0) + H0 (u)(B(u) − B(u − 1)) + Hn (u)(Sn (u) − Sn (u − 1))
u=1 u=1 n=1

t
X t
X X N
t X t X
X N
= V (0) + H0 (u)B(u) − H0 (u)B(u − 1) + Hn (u)Sn (u) − Hn (u)Sn (u − 1)
u=1 u=1 u=1 n=1 n=1 n=1

t
X t
X
= V (0) + V (u) − V (u − 1)
u=1 u=1

= V (t).

On the other hand, if above holds, then

N
X
V (u − 1) = Hn (u)Sn (u − 1) + H0 (u)B(u − 1), n = 2, 3, . . . , T
n=1

Let t = u − 1, then

N
X
V (t) = H0 (t + 1)B(t) + Hu (t + 1)Sn (t), t = 1, 2, . . . , T − 1
n=1

By definition of self-financing, H is a self-financing strategy.

5. If H is a self-financing strategy, then

N
X
V (t) = H0 (t + 1)B(t) + Hn (t + 1)Sn (t)
n=1

So
V ∗ (t) = V (t)/B(t)
N
X Sn (t)
= H0 (t + 1) + Hn (t + 1)
n=1
B(t)
N
X
= H0 (t) + Hn (t + 1)Sn∗ (t)
n=1

4
 Similar to question 4, if H is self-financing, then
t
X t X
X N
∗ ∗ ∗ ∗ ∗
V (0) + G (t) = V (0) + H0 (u)(B (u) − B (u − 1)) + Hn (u)(S ∗ (u) − S ∗ (u − 1))
u=1 u=1 n=1

t
X N
X
∗ ∗
= V (0) + (H0 (u)B (u) + Hu (u)S ∗ (u))
u=1 n=1
t
X N
X
− (H0 (u)B ∗ (u − 1) + Hn (u)S ∗ (u − 1))
u=1 n=1

t
X t
X
∗ ∗
= V (0) + V (u) − V ∗ (u − 1)
u=1 u=1

= V ∗ (t)

On the other hand, if V ∗ (0) + G∗ (t) = V ∗ (t), from above computations, we know

N
X
∗ ∗
V (u − 1) = H0 (u)B (u − 1) + Hn (u)Sn∗ (u − 1) u = 2, 3, . . . , T,
n=1

i.e.
N
X
V ∗ (t) = H0 (t + 1)B ∗ (t) + Hu (t + 1)Sn∗ (t), t = 1, 2, . . . , T − 1.
n=1

Therefore, H is a self-financing trading strategy.

6. Let F = {B1 , B2 , . . . , Bm } be the partition corresponding to the σ−field F. As X is

measurable with respect to F. So on each Bi , X = xi is a constant. So for ω ∈ Bi

E[XY |F](ω) = E[XY |Bi ]

= E[xi Y |Bi ]
= xi E[Y |Bi ]

But XE[Y |F](ω) = xi E[Y |Bi ] if ω ∈ Bi .

Therefore, E[XY |F] = XE[Y |F].
So if X is measurable with respect to F. Then from proposition 1 in Section 2

E[XY ] = E[E[XY |F]]

= E[XE[Y |F]]

follows.

5
7. (a)⇒(b). Assume X is a martingale with respect to the filtration {Ft }. That is to say

E[Zt+s |Ft ] = Zt for all s, t ≥ 0.

Take s = T − t, then

E[ZT |Ft ] = Zt , t = 0, 1, . . . , T − 1.

(b)⇒(c). E[Zt+1 |Ft ] = Zt follows from (b). But as Zt is measurable with respect to
Ft , so
E[Zt |Ft ] = Zt .

So E[Zt+1 |Ft ] = E[Zt |Ft ], i.e.

E[Zt+1 − Zt |F] = 0.

(c)⇒(a). If s = 0, then E[Zt+s |Ft ] = Zt is obvious. Assume s > 0 and (c) holds.
Then
E[Zt+s |Ft ]
=E[Zt+s − Zt+s−1 + Zt+s−1 − Zt+s−2 + . . . + Zt+1 − Zt + Zt |F]
=E[Zt+s − Zt+s−1 |Ft ] + E[Zt+s−1 − Zt+s−2 |Ft ]
+ . . . + E[Zt+1 − Zt |Ft ] + Zt
=E[E(Zt+s − Zt+s−1 |Ft+s−1 )|Ft ]
+E[E(Zt+s − 1 − Zt+s−2 |Ft+s−2 )|Ft ]
+ . . . + E[Zt+1 − Zt |Ft ] + Zt
=Zt .

8.
(NOTE: The following method is a more tedious method to compute martingale
measure. You can use the decomposition method to decompose the multiperiod
model to single period models, then it is easy to find the martingale measure
that way. I strongly recommend to try the latter method.)
Assume Q is a martingale measure. Then
 
B(t)S(t + s)
EQ Ft = S(t) for t, s ≥ 0 t+s≤2
B(t + s)

First if t = 0, F0 = {φ, Ω}. So EQ (·|F0 ) = EQ . From this we obtain

t = 0, s = 1, 8(Q(ω1 ) + Q(ω2 )) + 4(Q(ω3 ) + Q(ω4 )) = 5(1 + r) (1)
2
t = 0, s = 2, 9Q(ω1 ) + 6Q(ω2 ) + 6Q(ω3 ) + Q(ω4 ) = 5(1 + r) (2)

6
Secondly, if t = 1, F1 = {φ, Ω, {ω1 , ω2 }, {ω3 , ω4 }}, so

Q(ω1 )
P {S(2) = 9|{ω1 , ω2 }} =
Q(ω1 ) + Q(ω1 )

Q(ω2 )
P {S(2) = 6|{ω1 ω2 }} =
Q(ω1 ) + Q(ω2 )

Q(ω3 )
P {S(2) = 6|{ω3 ω4 }} =
Q(ω3 ) + Q(ω4 )

Q(ω4 )
P {S(2) = 3|{ω3 ω4 }} =
Q(ω3 ) + Q(ω4 )

So for ω ∈ {ω1 , ω2 }
   
B(1)S(2) 1 9Q(ω1 ) 6Q(ω2 )
S(1, ω) = EQ F1 (ω) = +
B(2) 1 + r Q(ω1 ) + Q(ω2 ) Q(ω1 ) + Q(ω2 )

From this we obtain

9Q(ω1 ) + 6Q(ω2 ) = 8(1 + r)(Q(ω1 ) + Q(ω2 )) (3)

Similarly for ω = {ω3 , ω4 }

   
B(1)S(2) 1 6Q(ω3 ) + 3Q(ω4 )
S(1, ω) = EQ F1 (ω) =
B(2) 1+r Q(ω3 ) + Q(ω4 )

From this we obtain

6Q(ω3 ) + 3Q(ω4 ) = 4(1 + r)(Q(ω3 ) + Q(ω4 )) (4)

From (3), (4) we can easily obtain

2 + 8r
Q(ω1 ) = Q(ω2 ) (5)
1 − 8r
1 + 4r
Q(ω3 ) = Q(ω4 ) (6)
2 − 4r
Substituting (5) and (6) to (1) we have
24 12
Q(ω2 ) + Q(ω4 ) = 5(1 + r) (7)
1 − 8r 2 − 4r
Substituting (5) and (6) to (2) we obtain

24(1 + r) 12(1 + r)
Q(ω2 ) + Q(ω4 ) = 5(1 + r)2
1 − 8r 2 − 4r

7
which is equivalent to (7).
Substituting (5) and (6) to Q(ω1 ) + Q(ω2 ) + Q(ω3 ) + Q(ω4 ) = 1 we obtain

3 3Q(ω4 )
Q(ω2 ) + =1 (8)
1 − 8r 2 − 4r

(1 + 4r)(1 − 8r)
(7) − (8) × 4 leads to Q(ω2 ) = .
12
Subsequently we have
(1 − 2r)(3 − 5r)
Q(ω4 ) =
6
(1 + 4r)(1 + 5r)
Q(ω1 ) =
6
(1 + 4r)(1 − 2r)
Q(ω3 ) =
12
This Q is a martingale measure. As there are no arbitrage opportunities if and only
if there is a risk neutral probability measure. So in this model, there are no arbitrage
opportunities.

DUAN/NKU