[R-SIG-Finance] Include only distinct lags into a GARCH specification (rgarch)

Johannes Lips johannes.lips at googlemail.com
Sat May 7 16:46:15 CEST 2011


I am trying to fit a GARCH-model to an hourly price series with 24 
values each day.
Therefore I got relatively high autocorrelation in the data, especially 
for the lag=24. I don't want to fit a GARCH with 24 AR-Terms but rather 
would add one distinct external regressor for the lagged variable.

So far I tried the following:
	LaggedDiffHourlyPriceSeries <- lag(DiffHourlyPriceSeries,-24)

	spec.hourly.lagged.gjr <- ugarchspec(variance.model = 	list(model = 
"gjrGARCH", garchOrder = c(1,1), external.regressors = 
LaggedDiffHourlyPriceSeries, distribution.model = "std"))

	fit.hourly.lagged.gjr <- ugarchfit(DiffHourlyPriceSeries, spec = 
spec.hourly.lagged.gjr,  fit.control = list(scale = 1),solver = "solnp", 
solver.control = list(outer.iter=500,inner.iter=1000))

That obviously won't work because of the NAs in the whole data set.
A little bit of the whole dataset
could be found here:

I would be glad if someone could point me to a feasible solution!

Thanks in advance


More information about the R-SIG-Finance mailing list