I am reading Ernie Chan's blog post "Mean reversion, momentum, and volatility term structure". It says that

To be precise, if $z$ is the log price, then volatility, sampled at intervals of $\tau$, is $$\text{Volatility}(\tau)=\sqrt{\text{Var}(z_t-z_{t-\tau})}$$ where $\text{Var}$ means taking the variance over many sample times.

In general, we can write $$\text{Var}(z_t-z_{t-\tau})\propto \tau^{2H}$$ where H is called the "Hurst exponent", and it is equal to 0.5 for a true geometric random walk, but will be less than 0.5 for mean reverting prices, and greater than 0.5 for trending prices.

It looks correct intuitively, as for a trending price, the variance increases faster as the sample interval gets larger, and for a mean reverting price, the variance doesn't change that much accordingly.

Can anyone give some explanation (or proof) on that (and why it is equal to 0.5 for a true geometric random walk)? Thanks!


A geometric random walk looks something like $S_t=S_0\exp((\mu-\sigma^2/2)t+\sigma W_t)$, where $W_t$ is standard brownian motion. Thus $\log(S_t)$ has fluctuations $\sigma W_t$, and it is a fact that $W_{t}-W_{t-\tau}$ has standard deviation $\sqrt{\tau}$.

For a simpler explanation, see here. Effectively, $\log(S_t)\approx \log(S_{t-\tau})+e_{t-\tau}$, where $e_t$ is normally distributed with mean 0 and standard deviation $\sqrt{\tau}$.


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.