0
$\begingroup$

Start by defining the Symmetric Random Walk:

$$ M_t = \sum_{i=1}^{t}X_i, ~~ \text{with}~X_0=0 $$

where $X_i$ is equal to 1 or -1 with $p=(1-p)=0.5$.

Consider $t > s$, the variance of its increments is:

\begin{equation} \begin{split} VAR(M_t - M_s) & = VAR \Big( \sum_{i=s+1}^tX_i\Big)~\text{by indepence of} ~ X_i \\ & = \sum_{i=s+1}^t VAR(X_i) = t-s \end{split} \end{equation}

Now define the Scaled Symmetric Random Walk:

$$ W^{(n)}(t) = \begin{cases} \dfrac{1}{\sqrt{n}}M_{nt}, & \mbox{if }~t=\dfrac{m}{n}~\in~ \mathbb{N} \\ \mbox{linear interpolation}, & \mbox{if } t~\in~ \Big( \dfrac{m}{n}, \dfrac{m+1}{n} \Big) \end{cases} $$

Now suppose $~t > s~$ and $~t = l/n~$ and $~s = m/n~$ are integers, now how is possible to obtain the following variance?

$$ VAR(W^{n}(t) - W^{n}(s)) $$

I tried this way:

\begin{equation} \begin{split} VAR(W^{n}(t) - W^{n}(s)) & = VAR\Big( \dfrac{1}{\sqrt{n}} (M_l - M_m) \Big) \\ & = \dfrac{1}{n}VAR(M_l - M_m) \end{split} \end{equation}

Then by applying the passages of the normal random walk reported above, I obtain:

$$ VAR(W^{n}(t) - W^{n}(s)) = \dfrac{1}{n} (t-s) $$

But I know that the correct formula should be $t-s$. Where is my error?

$\endgroup$
0
$\begingroup$

You forget an $n$ in the index. After "I tried this way", second equivalence, you should apply:

$$ M_l - M_m = M_{tn} - M_{sn} $$

$\endgroup$
  • $\begingroup$ No because it is $M_{tn}$ that becomes $M_{l}$ since $t = l/n$. $\endgroup$ – Archimede Aug 30 '17 at 14:14
  • $\begingroup$ Okay then the problem is later, see above. $\endgroup$ – Gijs Aug 30 '17 at 14:25

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.