Professional Documents
Culture Documents
Department of Mathematics
Technical University of Munich
Financial Mathematics 1
Aleksey Min, Henrik Sloot, and Ben Spies
Exercise sheet 7
The exercise sheet will be discussed in two groups of Zoom exercise sessions on December 8, 2021,
and in videos available on the Moodle page. For details, see the course’s Moodle announcements.
You should try to solve the exercises at home before the exercise. Grading bonus and homework
submission rules: https://www.moodle.tum.de/mod/page/view.php?id=1863336.
Exercise 7.1
Consider a two-period binomial model with P (0) = 100, u = 1.2, d = 0.8, and r = 5 %. We want to
price a European call option with maturity T = 2 and strike X = 90, whose payoff at maturity and price
process are denoted by C(2) and VC = {VC (t) : t ∈ {0, 1, 2}}, respectively.
a) Draw a binomial tree diagram with the values of the security at each node and determine the set of
all risk-neutral probability measures M.
Solution:
144 = 100 · u2
120 = 100 · u
80 = 100 · d
64 = 100 · d2
Since d < 1 + r < u, we can apply Theorem 3.27 and conclude that there exists a unique risk-neutral
martingale measure given by
Q̃ = Q̃(ω1 ), Q̃(ω2 ), Q̃(ω3 ), Q̃(ω4 )
such that
Q̃(ω1 ) = q̃ 2 ,
Q̃(ω2 ) = q̃ · (1 − q̃),
Q̃(ω3 ) = (1 − q̃) · q̃,
2
Q̃(ω4 ) = (1 − q̃) .
and therefore, the contingent claim is attainable. We use the formula from Theorem 4.3 to compute
4
h i h i 1 X
VC (0) = EQ̃ C̃(2) | I0 = EQ̃ C̃(2) = 2 C(2, ωi ) · Q̃(ωi )
(1 + r) i=1
1
= (54 · 0.3906 + 2 · 6 · 0.2344 + 0) ≈ 21.6827,
1.052 h i
VC (1) = P0 (1) · EQ̃ C̃(2) | I1 (ω)
h i
E C̃(2) | {ω1 , ω2 } ω ∈ {ω1 , ω2 },
Q̃
= P0 (1) · h i
E C̃(2) | {ω3 , ω4 } ω ∈ {ω3 , ω4 },
Q̃
1 C(2, ω1 ) · Q̃(ω1 ) Q̃(ω2 )
1+r Q̃(ω1 )+Q̃(ω2 )
+ C(2, ω2 ) · Q̃(ω1 )+Q̃(ω2 )
ω ∈ {ω1 , ω2 },
= ) )
1 C(2, ω3 ) · Q̃(ω3 Q̃(ω
+ C(2, ω4 ) · Q̃(ω )+Q̃(ω
4
ω ∈ {ω3 , ω4 },
1+r Q̃(ω3 )+Q̃(ω4 ) 3 4)
1 54 · 0.3906 0.2344
1.05 0.3906+0.2344 + 6 · 0.3906+0.2344 ω ∈ {ω1 , ω2 },
=
1 6· 0.2344 0.1406
1.05 0.2344+0.1406 + 0 · 0.2344+0.1406 ω ∈ {ω3 , ω4 },
(
34.6839 ω ∈ {ω1 , ω2 },
≈
3.5718 ω ∈ {ω3 , ω4 },
h i
VC (2) = P0 (2) · EQ̃ C̃(2) | I2 = C(2, ω) ∀ω ∈ Ω.
Exercise 7.2
Consider a two-period binomial model with P (0) = 100, u = 1.25, d = 0.8 and r = 10 %. Furthermore,
denote Y (t) = max {P (s) : s ≤ t}, t ∈ {0, 1, 2}.
a) Draw a tree diagram (see slide 151) and determine the risk-neutral probability for an upward move-
ment q̃.
Solution:
156.25 = 100 · u2
125 = 100 · u
80 = 100 · d
64 = 100 · d2
Using the same argumentation as in Exercise 1, we calculate the probability of an upward movement
as
1+r−d 1.1 − 0.8 2
q̃ = = = .
u−d 1.25 − 0.8 3
b) Write the elements of the sample set Ω and their corresponding probabilities under the risk-neutral
measure Q̃.
Solution: We have Ω = {ω1 , ω2 , ω3 , ω4 } = {(up, up) , (up, down) , (down, up) , (down, down)}. Again
using the argumentation from Exercise 1 and the probability of an upward movement calculated in
Part a, we get
Q̃ = Q̃(ω1 ), Q̃(ω2 ), Q̃(ω3 ), Q̃(ω4 )
2
= q̃ 2 , q̃ · (1 − q̃), (1 − q̃) · q̃, (1 − q̃)
4 2 2 1
= , , , .
9 9 9 9
c) Consider a knock-in barrier option with exercise price X = 90, barrier B = 110 and maturity T = 2;
its payoff is given by:
D1 (2) = max {P (2) − X, 0}1[Y (2)≥B] .
Determine the payoff D1 (2, ω) for all ω ∈ Ω and use Theorem 4.3 (see slide 163) to calculate the
option price at time t = 0.
Solution: First, note that the path-maximum in the different states is given by
d) Use Theorem 3.28 (see slide 156) to determine the probability Q̃(Y (2) ≥ P (0)ui ), i ∈ {0, 1, 2}, and
use the result to price a digital barrier option with payoff at maturity T = 2 given by D2 (2) =
1[Y (2)≥B] for B = 110.
Solution: Note that we have d = u−1 and Y (2) ∈ P (0) · ui , i ∈ {0, 1, 2} = {100, 125, 156.25}, so
Since the market is complete, the digital barrier option is attainable and can be priced as follows:
h i 1
VD2 (0) = P0 (0) · EQ̃ D̃2 (2) | F0 = · EQ̃ [D2 (2)]
1.12
1 1
= 2 · EQ̃ 1{Y (2)≥110} = 1.21 · Q̃(Y (2) ≥ 110)
1.1
1 1 2
= · Q̃(Y (2) ≥ 125) = · ≈ 0.5510.
1.21 1.21 3
Exercise 7.3
Consider the two-period financial market model from Exercise 6.2 with a riskless interest rate r = 0.
Then, the set of all risk neutral probability measures is given by:
n 0 o
M = 1/6 1/6 2/9 4/9 .
Construct a hedging strategy for a European put option with strike X = 2 and maturity T = 2.
Solution: Note that |M| = 1, i.e. by Theorem 4.7, the financial market is complete. The payoff of the
put option is
Put(2) = max {X − P1 (2), 0} = max {2 − P1 (2), 0}.
We first evaluate the value of the put option at the different time points. Obviously,
1 7
Put(2, ω1 ) = 0, Put(2, ω2 ) = , Put(2, ω3 ) = 1, Put(2, ω4 ) = .
2 4
1 1 1 2 4 7 13
VPut (0) = EQ̃ [Put(2) | F0 ] = ·0+ · + ·1+ · =
6 6 2 9 9 4 12
VPut (1) = P0 (1) · EQ̃ [Put(2) | F1 ](ω) = EQ̃ [Put(2) | F1 ](ω)
(
EQ̃ [Put(2) | {ω1 , ω2 }], ω ∈ {ω1 , ω2 }
=
EQ̃ [Put(2) | {ω3 , ω4 }], ω ∈ {ω3 , ω4 }
1 1
0 · 1 6 1 + 1 · 1 6 1 , ω ∈ {ω1 , ω2 }
+6 2 +6
= 6
2
6
4
1 · 2 9 4 + 7 · 2 9 4 , ω ∈ {ω3 , ω4 }
9 + 9
4 9 + 9
(
1
, ω ∈ {ω1 , ω2 }
= 43
2 , ω ∈ {ω3 , ω4 }.
Now let ϕ be a predictable hedging strategy. We want to solve the following equations:
At time t = 2, this results in the following equations, depending on the four different states:
5
0 = ϕ0 (2, ω1 ) + ϕ1 (2, ω1 ) · ,
2
1 3
= ϕ0 (2, ω2 ) + ϕ1 (2, ω2 ) · ,
2 2
1 = ϕ0 (2, ω3 ) + ϕ1 (2, ω3 ) · 1,
7 1
= ϕ0 (2, ω4 ) + ϕ1 (2, ω4 ) · .
4 4
Since the hedging strategy is predictable, it furthermore satisfies the four equations
5
ϕ0 (2, ω1 ) = ϕ0 (2, ω2 ) = , ϕ0 (2, ω3 ) = ϕ0 (2, ω4 ) = 2,
4
1
ϕ1 (2, ω1 ) = ϕ1 (2, ω2 ) = − , ϕ1 (2, ω3 ) = ϕ1 (2, ω4 ) = −1.
2
At time t = 1, the equation to satisfy is
Note, that a hedging strategy also must be self-financing; for this, we check that
1 ! 5 1 1
= ϕ0 (2, ω1 ) + ϕ1 (2, ω1 )2 = − · 2 =
4 4 2 4
3 ! 1 1 3
= ϕ0 (2, ω1 ) + ϕ1 (2, ω1 ) = 2 − 1 = .
2 2 2 2
D iscreteDistribution (
10 supp = s * exp ((0: n ) * log ( u ) + ( n - (0: n )) * log ( d )) ,
prob = dbinom (0: n , n , q )
)
r _ dt <- r * dt
u _ dt <- exp ( sigma * sqrt ( dt ))
10 d _ dt <- 1 / u _ dt
Solution:
s - E ( S to ckP ri ceD is tri but i o n (s , r _ dt , u _ dt , d _ dt , n ) / (1 + r _ dt )^ n )
Now, consider a digital call option D, i.e., with payoff D(T ) = 1{P1 (T )≥k} for some k > 0.
d) Create a plot for n ∈ {1, . . . , 1000} to show that the arbitrage-free price of the digital call option
with k = s0 converges to !
log (s0 /k) + r − 21 σ 2 T
Φ − √ ,
σ T
where Φ is the cumulative distribution function of the standard normal distribution. Create a second
√
plot to show that the convergence happens at rate O(1/ n), i.e., there exists n0 , m ∈ N s.t. the error
terms n , n ∈ N, fulfil
|n | ≤ K ∀n ≥ n0 .
Solution:
k <- s
g <- function (x , k , r _ dt , n ) {
ifelse ( x > k , 1 , 0) / (1 + r _ dt )^ n
}
5
df <- tibble (
n = seq _ len (1 e3 ) ,
dt = ! ! tt / n ,
u _ dt = exp ( ! ! sigma * sqrt ( dt )) ,
10 d _ dt = 1 / u _ dt ,
r _ dt = ! ! r * dt ) % >%
rowwise () % >%
mutate (
Binomial = E ( S toc kP ric eD ist rib ut ion ( ! !s , r _ dt , u _ dt , d _ dt , n ) ,
15 fun = g , k = ! !k , r _ dt = r _ dt , n = n ) ,
Limit = pnorm (
0,
mean = log ( ! ! s / ! ! k ) + ( ! ! r - 0.5 * ! ! sigma ^2) * ! ! tt ,
sd = ! ! sigma * sqrt ( ! ! tt ) ,
20 lower . tail = FALSE ) *
exp ( - ! ! r * ! ! tt ) ,
Error = Limit - Binomial )
(
25 df % >%
ggplot ( aes ( x = n , y = Binomial )) +
geom _ line () +
geom _ hline ( aes ( yintercept = Limit [[1]]) , linetype = " dotted " ) +
scale _ x _ continuous ( name = " n " , trans = " log10 " ) +
30 scale _ y _ continuous ( name = ~ E [ tilde ( Q )]( tilde ( P )[1]( T ))) +
theme _ minimal ()
) / (
df % >%
ggplot ( aes ( x = n , y = Error * sqrt ( n ))) +
35 geom _ line () +
geom _ hline ( yintercept = 0 , linetype = " dotted " ) +
scale _ x _ continuous ( name = " n " , trans = " log10 " ) +
scale _ y _ continuous ( name = ~ epsilon [ n ] * sqrt ( n )) +
theme _ minimal ()
40 )
e) Create a plot for n ∈ {1, 3, . . . , 999} to show that the convergence in d) on the odd subsequence
happens at rate O(1/n).
Solution:
(
df % >%
filter ( n %% 2 == 1) % >%
ggplot ( aes ( x = n , y = Binomial )) +
5 geom _ line () +
geom _ hline ( aes ( yintercept = Limit [[1]]) , linetype = " dotted " ) +
scale _ x _ continuous ( name = " n " , trans = " log10 " ) +
f) Create a plot for n ∈ {1, 2, . . . , 1000} to show that the arbitrage-free price of the digital call option
with k = 1.1 · s0 converges to
!
log (s0 /k) + r − 12 σ 2 T
Φ − √ .
σ T
Create a second and third plot to show that the convergence of the odd subsequence happens at
√
rate O(1/ n), but not at rate O(1/n).
Solution:
k <- s * 1.1
df <- tibble (
n = seq _ len (1 e3 ) ,
5 dt = ! ! tt / n ,
u _ dt = exp ( ! ! sigma * sqrt ( dt )) ,
d _ dt = 1 / u _ dt ,
r _ dt = ! ! r * dt ) % >%
rowwise () % >%
10 mutate (
Binomial = E ( S toc kP ric eD ist rib ut ion ( ! !s , r _ dt , u _ dt , d _ dt , n ) ,
fun = g , k = ! !k , r _ dt = r _ dt , n = n ) ,
Limit = pnorm (
0,
15 mean = log ( ! ! s / ! ! k ) + ( ! ! r - 0.5 * ! ! sigma ^2) * ! ! tt ,
sd = ! ! sigma * sqrt ( ! ! tt ) ,
lower . tail = FALSE ) *
exp ( - ! ! r * ! ! tt ) ,
Error = Limit - Binomial )
20
(
df % >%
ggplot ( aes ( x = n , y = Binomial )) +
geom _ line () +
25 geom _ hline ( aes ( yintercept = Limit [[1]]) , linetype = " dotted " ) +
scale _ x _ continuous ( name = " n " , trans = " log10 " ) +
scale _ y _ continuous ( name = ~ E [ tilde ( Q )]( tilde ( P )[1]( T ))) +
theme _ minimal ()
) / (
30 df % >%
filter ( n %% 2 == 1) % >%