I am using squared return as a proxy to calculate volatility, however i'm not sure whether to use raw return or percentage return. Under raw return all return estimates are below 1, however under percentage return there is a mix of return greater than 1 and less than 1. Percentage return below 1 would end up as a volatility figure less than the percentage return itself, on the other hand percentage return above 1 would end up as a volatility figure greater than percentage return.

My question is: Doesn't this pose a problem when calculating volatility in that there is an over estimation when the return is above 1?

I am going to use the data to fit an ARMA-GARCH model, would there be any difference if I used percentage or absolute values?


1 Answer 1


For GARCH family models the standard approach is to use log returns:

r(t) = log( p(t)/p(t-1) )

The magnitude is not so important as trying to make the time series stationary.


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.