I am trying to make sense of the Lévy Itô decomposition, in particular, of a note I have found regarding the size of the jumps.

From the Lévy decomposition we know that any Lévy process is a semimartingale, such that $$ X_t = b_t + \sigma B_t + \int_0^t\int_{\mathbb R} x \,\hat{N}(dt,dx) $$ where $B$ is a Brownian Motion started at $0$, and $$ \hat{N}(dt,dx) = \left\{ \begin{array}{ll} N(dt,dx) - \nu(dx)dt & \mbox{if } |x| < R \\ N(dt,dx) & \mbox{if } |x| \geq R \end{array} \right. $$ for some $R\in[0,\infty]$

What I am trying to make sense of is the following statement:

"If $R = 0$ then $\hat{N} = N$ everywhere, and if $R = \infty$ then $\hat{N} = \tilde{N}$ everywhere" (where $\tilde{N}$ is the compensated Poisson random measure).

Obviously, this make sense if one simply realizes how $\hat{N}$ is defined, but I would like to get a more intuitive idea. In other words, why when $R = \infty$ we have a martingale, and we have 'something else' if $R = 0$?

Thanks to anyone who bothers reading!


Basically, it's the fact that stochastic integrals with respect to a compensated random measure are martingales whereas stochastic integrals with respect to a non-compensated random measure are not martingales.

Just the same as for processes: The Poisson process is not a martingale, but the compensated (or centered) Poisson process is a martingale.


Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.