Hi all;
I am trying to perform a rolling out-of-sample
regression on daily log return with volatility being estimated by GARCH(1,1)
model. I am using "roll" argument but whatever width of rolling window I chose,
it kept saying "Result too long".
> roll.fit = roll(FUN = garch, data = ""
width = 6, incr = 0, formula.mean =
hs.ts ~ 1, formula.var
= ~ garch(1, 1), save.list = c("coef"))
Problem in 0:n: Result too long
Is it a bug or anything wrong with my code? Please help.
Is there any other s-plus code to perform out-of-sample
analysis?
June
PS: S-Plus 6.2.1 for windows (academic site edition)
S+FinMetrics Professional Edition
Version 1.0 for Microsoft Windows : 2002