# how to compute daily skewness of S&P daily return timeseries under no other more high - frequency time series?

As we all know , return time series marked features: fat tail or negative skewness and peakedness. For a similar problem of variance computation, we can compute variance by garch model and other derivatives.In other words,Does we predict skewnenss like predicting variance by garch model and other derivatives ? does the existing model for skewness exist ? According to modern fiancial theory(investors like high return ,high skewness or positive skewness ,dislike variance ,kurtosis),we can deduce$\frac{ \partial {\text{variance}}}{\partial{\text{skewness}}}<0$ and$\frac{ \partial {\text{variance}}}{\partial{\text{kurtosis}}}>0$.and kurtosis ? Two seeming useful links:

Any comments are appreciated

-
anybody here?any links or thougts can be post. – Jacky Zhang May 26 '14 at 7:34

## 1 Answer

skewness and kurtosis can both be predicted by models similar to garch and its derivatives. For more details: a new published paper here

-