Suppose we have a sequence $(X_n), n \in \mathbb{Z_+}$ of random variables on $(\Omega, \mathcal{F}, \mathbb{P})$ and define $X^*_{n}(\omega)=\max_{k\leq n}X_k(\omega)$. I want to prove rigorously that $X_n^*$ is a random variable.

My main issue is that I'm not sure with respect to which $\sigma$-algebra we have to check the measurability - is it the same as the one - $(\mathcal{F}_n)$, which makes the sequence $(X_n)$ adapted? It doesn't seem so, but nevertheless, I know that the maximum of $2$ measurable functions is measurable (using the identity that $\max\{f,g\}=\frac{1}{2}(|f-g|+(f+g))$ and using standard results), however this doesn't generalize directly. If this is the true $\sigma$-algebra then I want to show that for a Borel-measurable $G$ we have

$$\{\omega: X_n^*(\omega)\in G\} \in \mathcal{F}_n$$

which gets a bit messy because of the maximums.

Hence the question: Is there a specific $\sigma$-algebra with respect to which the running maximal is measurable or is it the standard one?


  • $\begingroup$ It's the standard one, I would say - the ${\cal F}_n$ are increasing, correct? You might like to take a look at math.stackexchange.com/questions/1456350/… $\endgroup$ – peter a g Mar 19 '18 at 13:35
  • $\begingroup$ I would say use $(a,\infty]$, as that is more direct (from that link) - sups are the same as maxs if the index set is finite. $\endgroup$ – peter a g Mar 19 '18 at 13:43

When we have a probability space $(\Omega, \mathcal{F}, \mathbb{P})$ and a function $Y\colon \Omega\to\mathbb R$, we say that $Y$ is a random variable if for all Borel subset $B$ of the real line, the set $Y^{-1 }(B)$ belongs to $\mathcal F$. In particular, not filtration is involved.

If you know how to prove that the maximum of two random variables still is a random variable, then you can use an induction argument because $X_n^*=\max\left\{ X_{n-1}^*,X_n\right\}$.


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.