I found this great post deriving the solution to the Merton Jump-Diffusion SDE

$$S_t = S_0\exp\left(\left(\mu - \frac{\sigma^2}{2}\right)t + \sigma W_t\right)\prod_{j=0}^{N_t}V_j$$

The first part of the equation (inside the exponent) just looks like Geometric Brownian Motion (GBM) to me, which I am ok with. However I am unsure about the product of $V$'s; it seems $N_t$ is a poisson process with intensity $\lambda$ and jump sizes $Y_j$ which are iid Normally distributed $Y_j ∼ N(\mu, \delta^2)$. But I don't know how to include these parameters in my simulation.

For GBM python code, I have followed the wiki page

import numpy as np
import math
dt = 1/252 # time increments
mu = 0.01 # stock price daily drift (1%)
sigma = 0.14 # stock price daily volatility (14%)
T = 400 # number of periods (dyas) to simulate
So = 100 # initial stock price
drift = mu - 0.5*sigma**2
diffusion = sigma*np.random.normal(loc=0, scale=math.sqrt(dt), size=(T,)) # normal distribution
S = np.exp(drift + diffusion)
S = So*S.cumprod() # GBM

I also found this post and this jupyter notebook; the latter has explicit code on the Merton jump model, however it is not very well commented and so I don't understand what is happening.

How can I interpret the product of $V$'s in the equation above, and how can I include these jumps in our simulation? (BONUS: preferably using numpy instead of explicit loops).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.