[R-SIG-Finance] ARMA(m,n)-APARCH(p,q)

[kriskumar at earthlink.net]

The estimation is by MLE and so completely depends on the optimizer you specify. 

I would take the loglikelihood function and try with a global optimizer like SA or diff evolution. The DE works with a population of solns and so you can get a distibution of the parameters and get a sense if there are multiple local minima.

HTH

Best
Krishna




Sent from my BlackBerry® wireless handheld  

-----Original Message-----
From: "Citta Francesco" <fcitta at inwind.it>
Date: Sun, 19 Nov 2006 15:10:18 
To:<r-sig-finance at stat.math.ethz.ch>
Subject: [R-SIG-Finance] ARMA(m,n)-APARCH(p,q)

Dear Listers,
I am working with cac40 time series data (from 03-01-2000 to 08-11-2006). I
estimated, with successful, a simple ARMA(6,0)-GARCH(1,1) but I think tha it
is better an ARMA(6,0)-TGARCH(1,1).
For this I used the garchFit() function , setting:

garchFit(formula.var=~arma(6,0),formula.var=~aparch(1,1),series=cac,delta=2,
cond.dist="dnorm",include.delta=FALSE)

I obtain a strange value of gamma1 (1) ,I think that it wrong, the expected
value would be near 0.12, and the ARCH sign is positive instead of negative.
How can solve this problem?
Are you able to say me where I mistaked?
Best regards.

_______________________________________________
R-SIG-Finance at stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance