in the video is specified that GARCH (p = 1, q = 1) is the best for returns, but higher q may be good for different time series. What kind of time series. Thanks
It really depends what you’re measuring. You could be measuring the rainfalls over a specific geographical region over time, or the results from a clinical trial of a new drug. In both those cases you can have time series data, which might (I don’t know if it will) exhibit qualities best described by a higher-level GARCH model.
Hi Chef, ok…so the best way is always to check the p-values and perform the LLR test. Right?Thanks in advance
Hey Luigi, I’d even focus more on the AIC in those cases. Just find whichever one has the lowest information criterion, while having significant coefficients. (That’s because the Log-likelihood is incorporated in the computation of the AIC)