I'm currently writing a paper and need to regress the 22 days realized volatility of the following month on its GARCH estimate and the 126days realized volatility up to t=1
The paper im referring to describes its procedure as:
"first fit a GARCH process on the daily returns. then regress the future realized 22day return volatility(o22,t+1) on the Gauch estimate (o(march,t) and the 126 days volatility and a constant"
so i did the GARCH process on my Returns:
spec<-ugarchspec(variance.model = list(model="gjrGARCH"),mean.model = list(armaOrder=c(0,0))) fit<-ugarchfit(spec = spec,data = FF4F$WML) fit
the input data contains the daily log returns of my portfolio.
but how do i get the forecast for the next months volatility out of that ? or how do i get the daily forecasted values at least for every day of my data set, so i can form the realized 22 day volatility out of them?
do i have to use ugarchforecast or ugarchroll ?