Finding the Second Central Moment of a Martingale

177 Views Asked by At

Assuming that $X_0=a \in [0,1]$ where $X_n$ is a martingale and $$P(X_{n+1}=\frac{X_n}{2}|\textit F_n)=1-X_n,\>\>\>P(X_{n+1}=\frac{1+X_n}{2}|F_n)=X_n$$

How may we show that $\bf{E[(X_{n+1}-X_n)^2]=\frac{1}{4}E[X_n(1-X_n)}$

My first attempt was to rearrage the probabilities of each term so that for example

$$P(X_{n+1}=\frac{X_n}{2}|\textit F_n)=\frac{E[X_{n+1}|[\frac{X_n}{2}|\textit F_n]]}{X_{n+1}}$$ and then where $\frac{1}{2}X_n|\textit{F_n}=\frac{1}{2}X_{n-1}$ so that $$P(X_{n+1}=\frac{X_n}{2}|\textit F_n)=\frac{E[X_{n+1}|\frac{1}{2}X_{n-1}]}{X_{n+1}}=1-X_n$$

But frankly I'm just getting more and more confused on how to proceed whilst being slightly overwhelmed by each tidbit of intuition that arrives with an additional variable.

1

There are 1 best solutions below

3
On BEST ANSWER

It's easy demonstrate that $\mathbb{E}((X_{n+1}-X_n)^2) = \mathbb{E}(X_{n+1}^2 - X_{n}^2)$ (because $X_n$ is martingale). You can see that \begin{eqnarray} \mathbb{E}(X_{n+1}^2) &=& \mathbb{E}(\mathbb{E}(X_{n+1}^2|\mathcal{F}_n))\\ &=& \mathbb{E}(\mathbb{E}(X_{n+1}^2 1_{\{X_{n+1} = \frac{X_n}{2}\}}|\mathcal{F}_n)) + \mathbb{E}(\mathbb{E}(X_{n+1}^2 1_{\{X_{n+1} = \frac{1 + X_n}{2}\}} |\mathcal{F}_n)) \\ &=& \mathbb{E}(\frac{X_n^2}{4}\mathbb{P}(X_{n+1} = \frac{X_n}{2}|\mathcal{F}_n)) + \mathbb{E}((\frac{1+X_n}{2})^2\mathbb{P}(X_{n+1} = \frac{1+X_n}{2}|\mathcal{F}_n))\\ & =& \mathbb{E}(\frac{X_n^2}{4}(1-X_n) + (\frac{1+X_n}{2})^2X_n)\\ &=& \mathbb{E}(\frac{1}{4}X_n + \frac{3}{4}X_n^2) \end{eqnarray} then $\mathbb{E}(X_{n+1}^2 - X_{n}^2) = \mathbb{E}(\frac{1}{4}X_n + \frac{3}{4}X_n^2 - X_n^2) = \frac{1}{4}\mathbb{E}(X_n(1-X_n))$