I am facing the following unusual problem: $Z_t$ is a pure drift process of the form $$ dZ_t = \kappa(X_t - Z_t) dt $$ where $X_t$ is another bounded process. I am interested in computing / upper-bounding the probability that it crosses a threshold $\alpha$ before a given horizon $T$, i.e. $$\mathbb{P}[\exists t \in (0,T) \text{ s.t. }Z_t \leq \alpha] << \epsilon$$ which in my case is a rather unlikely event. The fact that $X_t$ is bounded allows me to compute a minimal time $t_0$ such that the event is possible. Furthermore I can upper-bound the following probability for all times $t$ $$ \mathbb{P}[Z_t \leq \alpha] $$ relatively well using moments inequalities. However I do not see how to bound the main probability of interest. Maybe this is not the right way to approach the problem.

By advance, thanks for your suggestions.

  • $\begingroup$ Your process $Z_t$ admits the explicit representation $e^{-\kappa t}\Bigl(Z_0 + \kappa \int_0^t e^{\kappa s} X_s \, ds\Bigr)$. So without knowing the distribution of $X_t$ or some particular properties (as bounded away from zero), I do not expect a general result. $\endgroup$ – Stephan Sturm Nov 20 '14 at 5:07
  • $\begingroup$ Thx. I have partial knowledge of the distribution of $X_t$ (the first $n$ moments), and $X_t \in (0,a)$. How would you proceed ? $\endgroup$ – user149575 Nov 20 '14 at 9:20
  • $\begingroup$ I assume by moments you mean the moments of the marginals for some fixed t? I doubt that this will help much without knowing the transition probabilities of $X$. Could you be more precise how your process $X$ is given? $\endgroup$ – Stephan Sturm Nov 20 '14 at 17:16

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.