Skip to content

marginal model with AR-t-GARCH model

3 messages · Xiaochen Sun, Joe W. Byers, Diethelm Wuertz

Original
Xiaochen Sun
# 
Dear list,
 
I am now dealing with the time series data from electricity market, regarding the marginal modelling I wonder how to fit data with AR-t-GARCH model in R or S-plus.
 
Many thanks.
 
With regards
Mc
Joe W. Byers
#
Xiaochen Sun wrote:
MC

I think you will have to define your own LL function and call optim to 
fit the AR t Garch model.  I'm not sure it is in fseries.  I am working 
on a GED distribution Garch myself for commodities.  The documentation 
on rmetrics website for Garch modelling has a example of the LL function 
and the call to optim to fit the data as well as the code to calc SE, 
Tstats, and Pvalues of the parameter estimates.

Are you fitting daily, hourly or other data? and what power market?

Good luck
Joe
1 day later
Diethelm Wuertz
#
Joe W. Byers wrote:
Use:

require(fSeries)

garchFit(formula.mean = ~arma(1, 0), formula.var = ~garch(1, 1), series 
= x, cond.dist ="dstd")

for an AR(1)-t-GARCH(1,1) model

Important: You have to use the standardized t-distribution dstd(), the 
usual dt() gives wrong results!

DW