Re: Maximization of ARMA-GARCH models


Thank you for your answer.
You're right, the question was about any model (hence the "similar" i
used.)

So what you're saying is that using a different distriburion for the
innovations/errors makes another model out of the first one ?

It would have been great if i could compare directly the ML function of
an ARMA(1,1)-GARCH(1,1) with Student-t distribution with an
ARMA(1,1)-GARCH(1,1) with a GED distribution.

It is true that comparing ARMA(1,1)-GARCH(1,1) with
ARMA(1,1)-APARCH(1,1) might not be evident, but i can't see why the
first example could not be compared directly (thus preferring the one
with the maximum value)

On Jan 24, 11:21 am, "David Jones" <daj...@xxxxxxxxx> wrote:
monnomiznog...@xxxxxxxxxxxx wrote:
Anyone to help ?

On Jan 22, 5:57 pm, monnomiznog...@xxxxxxxxxxxx wrote:
I was wondering if the "value" of the found maximum is relevant
when
using different types of distribution for the innovations in the
maximization process of an ARMA-GARCH (or similar) model for a time
series like an asset return.

For example, if the log maximum of the function is let's say 1500
when using Student-t distribution and 1600 when using a Hyperbolic
distribution, does this mean that the Hyperbolic distribution is a
better fit or is there another "criterium" that is more relevant
for
choosing the distribution ?This question has confused the issue by emphasizing "ARMA-GARCH". The
answer is the same for any models. Yes the likelihood can be used to
compare fitted models. Unfortunately there is no simple theory to tell
you how to do it. The main stuff in the literature goes under the
heading of "tesing separate families of hypotheses", A summary of the
approach is that you build a more extensive model that includes both
the models you have as special cases.

David Jones- Hide quoted text -- Show quoted text -

.

Relevant Pages

  • Re: Random values from Hyberbolic Distribution on unit interval [0,1]
    ... Thank you also for the suggestion regarding the Beta distribution. ... The parameters of the hyperbolic distribution (expressed in terms of the ... Note that lambda = 1 since this is the hyperbolic distribution. ...
    (sci.stat.math)
  • Re: Maximization of ARMA-GARCH models
    ... series like an asset return. ... when using Student-t distribution and 1600 when using a Hyperbolic ... This question has confused the issue by emphasizing "ARMA-GARCH". ...
    (sci.stat.math)
  • Re: Maximization of ARMA-GARCH models
    ... maximization process of an ARMA-GARCH model for a time ... using Student-t distribution and 1600 when using a Hyperbolic ...
    (sci.stat.math)