2
$\begingroup$

When calculating volatility (either using an SMA or EWMA approach), what impact does the window size have on the volatility estimate?

$\endgroup$

1 Answer 1

2
$\begingroup$

If volatility was constant, then the impact would have been negligible. Unfortunately (?) volatility is time varying. If by SMA you mean simple moving average then the impact is that once the window moves away from a period its impact on the volatility disappears. Suppose that you had a huge spike in volatility of FB on July 26 2018. If you have 6 months window then in February this spike will not contribute anything into the moving average. This can be an issue if you're interested in tails of return distribution. On the other hand you could argue that what happened long ago is irrelevant.

EWMA doesn't have the explicit window size, however it has an effective window size that is linked to its decay factor $e^{-\lambda t}$. So, in February the July spike won't suddenly drop out, but its contribution would have been gradually decreasing as it is further away from the current period.

$\endgroup$
2
  • $\begingroup$ Thank you so much. So based on that answer, EMWA is more appropriate as the window period affects the volatilty less due to the decay factor weighting more recent volatility higher? $\endgroup$
    – Mr.Rlover
    Commented Jul 27, 2018 at 18:22
  • $\begingroup$ Both SMA and EWMA are linear filters. You can tune EWMA to have effectively the same window size as SMA. the main difference is that old observations do not suddenly drop out, but gradually decay. so it can be more appropriate in some applications $\endgroup$
    – Aksakal
    Commented Jul 27, 2018 at 18:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.