Processus de Lévy en Finance - Laboratoire de Probabilités et ...
Processus de Lévy en Finance - Laboratoire de Probabilités et ...
Processus de Lévy en Finance - Laboratoire de Probabilités et ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
1.4. PRICING EUROPEAN OPTIONS 55<br />
Since the bound ˜g n (v) is increasing on (−∞, 0) and <strong>de</strong>creasing on (0, ∞),<br />
∆ 3 N−2<br />
∑<br />
sup<br />
∂ 2<br />
∫ L/2<br />
24π ∣∂v 2 e−ivk ζ T (v)<br />
∣ ≤ ∆3<br />
12π ˜g 2(0) + ∆2<br />
24π<br />
m=0<br />
−L/2<br />
⎧<br />
2∑<br />
≤ ∆2 C<br />
⎨<br />
3−l<br />
l∑<br />
12π (3 − l)! ⎩ 2∆ |k − rT | j ∑l−1<br />
|k − rT | j<br />
+<br />
j!<br />
j!<br />
≤ ∆2<br />
6π<br />
l=0<br />
2∑<br />
l=0<br />
j=0<br />
j=0<br />
{<br />
C (<br />
3−l<br />
∆ + π )<br />
e |k−rT | + log<br />
(3 − l)! 2<br />
˜g 2 (v)dv<br />
∫ L/2<br />
−L/2<br />
(<br />
L<br />
2 + √<br />
L 2<br />
4 + 1 )<br />
dv |k − rT |l<br />
+<br />
1 + v2 l!<br />
}<br />
|k − rT | l<br />
l!<br />
.<br />
∫ L/2<br />
−L/2<br />
⎫<br />
dv<br />
⎬<br />
√<br />
1 + v 2 ⎭<br />
To compl<strong>et</strong>e the proof of (1.32) it remains to substitute this bound and a similar bound for<br />
∆<br />
∑<br />
∣ ∣ 3 N−2 ∣∣<br />
24π m=0 sup ∂ 2<br />
e −ivk ζ Σ ∣∣<br />
∂v 2 T (v) into (1.34).<br />
The Simpson rule (cf. [32]) is <strong>de</strong>fined by:<br />
∫ x+2h<br />
x<br />
f(ξ)dξ = 1 h(f(x) + 4f(x + h) + f(x + 2h)) + R,<br />
3<br />
where R = 1<br />
90 h4 f (4) (x ∗ )<br />
for some x ∗ ∈ [x, x + 2h]. The proof of (1.33) can be carried out in the same way as for the<br />
trapezoidal rule.<br />
Example 1.1. L<strong>et</strong> us compute the truncation and discr<strong>et</strong>ization errors in the Merton mo<strong>de</strong>l<br />
(1.16) with param<strong>et</strong>ers σ = 0.1, λ = 2, δ = 0.1, µ = 0 for options with maturity T = 0.25.<br />
Taking Σ = 0.1, we obtain the following values of the coeffici<strong>en</strong>ts C k and C Σ k :<br />
k C k Ck<br />
Σ<br />
1 0.0871 0.05<br />
2 0.0076 0.0025<br />
3 0.0024 2.85 · 10 −4<br />
4 3.29 · 10 −4 1.88 · 10 −5<br />
5 1.82 · 10 −4 2.99 · 10 −6<br />
With 4048 points and the log strike step d equal to 0.01, the truncation error is extremely small:<br />
ε T = 2 · 10 −59 , and the discr<strong>et</strong>ization error for at the money options is giv<strong>en</strong> by ε D = 0.0013<br />
for the trapezoidal rule and by ε D = 3.8 · 10 −5 for the Simpson rule. Since the call option price<br />
in this s<strong>et</strong>ting (S = K = 1, r = q = 0) is C = 0.0313, we conclu<strong>de</strong> that the Simpson rule gives<br />
an acceptable pricing error for this choice of param<strong>et</strong>ers.