Given a probability space $(\Omega,\mathcal{F},P)$ and a process $X_{t}$ defined on it. We consider the natural Filtration generated by the process $\mathcal{F}_{t}=\sigma (X_{s}:s\leq t)$. Let $\tau$ be a stopping time. The corresponding $\sigma$-Algebra of the stopping time $\tau$ is given by \begin{align} F_{\tau}=\left\{A\in \bigcup_{t\geq 0} F_{t}: A\cap\{\tau \leq t\}\in \mathcal{F}_{t} \forall t\geq 0\right\} \end{align} Now i am considering the requirements under which \begin{align} \mathcal{F}_{\tau}=\sigma (X_{s\wedge \tau }:s\geq 0) \end{align} holds.

Shiryaev "Optimal Stopping Rules" Stochastic Modelling and Applied Probability 8 in Springer 2008 makes the assumptions, that $\Omega$ has to be sufficiently rich, that means for all $t\geq 0$ and $\omega \in \Omega$ there exists $\omega'\in\Omega$ such that \begin{align} X_{s}(\omega')=X_{s\wedge t}(\omega) \forall s\geq 0 \end{align} holds; see Theorem 6 of Shiryaev. Then $\mathcal{F}_{\tau}=\sigma (X_{s\wedge \tau }:s\geq 0)$ should hold. Shiryaev shows this for arbitrary measurable space, and measurable process $X$ fullfilling this condition.

Also D. Stroock, S. Varadhan "Multidimensional Diffusion Processes" Lemma 1.3.3 proofs it, where here $\Theta$ is the space of continuous trajectories from $[0,\infty)$ to $\mathbb{R}^{d}$. In the proof Stroock also uses the assumption of the sufficiently richness of $\Omega$ under $X$, but he mentions it incidental, as if it would be clear for continuous $X$.

I guess, that this property also holds for cadlag processes $X$ i.e. fullfilling \begin{align} \mathcal{F}_{\tau}=\sigma (X_{s\wedge \tau }:s\geq 0) \end{align} But i dont know, how it is fullfilled.


The "saturation" property assumed by Shiryaev is indeed clear for $\Omega$ consisting of the space of cadlag paths from $[0,\infty)$ to $\Bbb R^d$, and $X_t(\omega)=\omega(t)$ for $t\ge0$ and $\omega\in\Omega$. For, given $t\ge 0$ and $\omega\in\Omega$, define $\omega'$ to be the path "$\omega$ stopped at time $t$"; that is, $\omega'(s):=\omega(s\wedge t)$ for $s\ge 0$. This path $\omega'$ is clearly an element of $\Omega$, and $X_{s\wedge t}(\omega) =\omega(s\wedge t) =\omega'(s)=X_s(\omega')$ for each $s\ge 0$.

| cite | improve this answer | |

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.