Dear alexios and other R users, I have the returns data on a firm named "ret" having 2767 observations total. I want the one day ahead rollingt forecasts starting from 1001th day and model is to be refit every 5 days. So my first 5 forecasts (i.e., for the 1001,1002,1003,1004 nd 1005th day) will be based on first 1000 observations, next 5 forecasts(i.e., for the 1006 to 1010'th day) will be based on first 1005 observations and then next 5 days forecast will be based on first 1010 observations and this process continues. I used following code spec= ugarchspec(variance.model= list(model = 'gjrGARCH') ,mean.model=list(armaOrder= c(0,0),include.mean= FALSE, archpow=2)) forecast<-ugarchroll(spec=spec,data=mkt_ret,forecast.length=1767,refit.every=5, refit.window="recursive") mkt_fcst<- as.data.frame(forecast) My questions are 1) I suppose the f_sigma in mkt_fcst dataframe is forecasted volatality but is this the true forecast which I need according to above description of my needed forecast 2) If this f_sigma is the right forecasted vol then why i got only the forecasts till 2765 days because next two days forecasts could be based on the data till 2765 observations. 3) Please suggest me how to formulate the right code for my needed forecast 4) I will do assymetric DCC also using second series on market index (mkt_ret) nd would need the DCC forecasts too in the same manner using rmgarch. -- View this message in context: http://r.789695.n4.nabble.com/forecasting-with-expanding-window-tp4634076.html Sent from the Rmetrics mailing list archive at Nabble.com.
forecasting with expanding window
1 message · naval