[R-SIG-Finance] ARMA(1,1)-GARCH(1,1) rolling estimation

[Matthieu Stigler]

Should "AA.day.forecast.data" contain the series oh one step ahead 
forecasts? If yes, you need to index it also!

AA.day.forecast.data[i-start+1] <- garch.pred[,1]

Btw, if you are doing it with a loop, you might have a look on package 
foreach, which is likely to make your loop contruction easier, and 
especially can run it easily on parallel cores...

Mat

PS: try when posting to give a code fully reproducible!

mam3xs a écrit :
> Hi all,
>
> I got the daily stock return data from 2005 - 2008, calculated from HF 
> minute data. (Thanks to Jeff and Josh).
>
> Now, I set 05 - 07yr as the carlibration period for estimating the 
> parameters of ARMA(1,1)-GARCH(1,1) model, aqnd leave 08 for backtesting.
>
> So I use the return data observations from 1:760 (yr 05-07) to 
> estimate the volatility on 2nd-Jan-08 (the position 761),
> then use the return data from 1:761 to estimate the vol on 3rd-Jan-08...
> 1:762...
> 1:763....
> ...
> ...
> and so on...
>
> My codes are as follow:
>
> start=length(AA.day.rn['2005-01::2007-12'])
> end = length(AA.day.rn)
>
> AA.day.forecast.data = array(0, dim=(end-start))
>
> for (i in start:end) {
>
>     garch.fit <- garchFit(formula = ~arma(1,1)+garch(1,1), data = 
> AA.day.rn[1:i])
>     garch.pred <- predict(garch.fit, n.ahead=1)
>     AA.day.forecast.data = garch.pred[,1]
>
> }
>
>
> Following my design, there should be 193 forecast data, however, when 
> I checked the object "AA.day.forecast.data", it gives the whole length 
> of the original data( 953 observations).....
> Also, I am not too sure that after the loop function of garch.fit, 
> will garch.pred catch the right estimation result?
>
> Many thanks.
> Mam
>
> P.S Sample data AA.day.rn is attached with this email.
> ------------------------------------------------------------------------
>
> _______________________________________________
> R-SIG-Finance at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
> -- Subscriber-posting only. If you want to post, subscribe first.
> -- Also note that this is not the r-help list where general R questions should go.