1
$\begingroup$

If $Y_n=\min\{M_n,7\}$ and $\{M_n\}$ is a martingale wrt ${X_n}$, show that ${Y_n}$ is a supermartingale wrt ${X_n}$

I tried doing cases for $M_n<7$ and for $M_n>7$, but I couldn't get that $E[M_{n+1}-M_n|X_{\le n}]\le 0$

$\endgroup$
1
$\begingroup$

The function $f(x)=\min\{x,7\}$ is concave, hence by Jensen's inequality for conditional expectation $$ \mathbb{E}[Y_{n+1}|\mathcal{F}_n]=\mathbb{E}[f(X_{n+1})|\mathcal{F}_n]\leq f(\mathbb{E}[X_{n+1}|\mathcal{F}_n])=f(X_n)=Y_n$$

$\endgroup$
  • $\begingroup$ I don't know what Fn stands for. $\endgroup$ – ak87 Apr 12 '16 at 2:05
  • $\begingroup$ $\mathcal{F}_n$ is whatever filtration that $X_n$ is a martingale with respect to. So it looks like $\mathcal{F}_n=\sigma(X_1,\dots,X_n)$ in your case. $\endgroup$ – carmichael561 Apr 12 '16 at 2:07
  • $\begingroup$ ok. I think I understand. So, "filtration" is the sequence ${X_i}$ of fair bets, and by using Jensen's inequality, $\mathbb{E}[Y_{n+1}-Y_n|\mathcal{F}_n]\le 0$ since I can put $Y_n$ inside the expectation function. Thanks - and correct me if I'm wrong please. $\endgroup$ – ak87 Apr 12 '16 at 2:37
  • 1
    $\begingroup$ The filtration is the information you know at time $n$, which you could just replace by $X_{\leq n}$ for your problem. And my argument shows that $\mathbb{E}[Y_{n+1}-Y_n|\mathcal{F_n}]\leq 0$. $\endgroup$ – carmichael561 Apr 12 '16 at 2:40

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.