[R-SIG-Finance] RUGARCH Question

[alexios ghalanos]

Hi,

Not sure what you mean by "halts". In any case, if you try searching the 
methods available for a uGARCHfit object (?'uGARCHfit-class') you will 
find one called 'convergence' which returns the solver convergence code 
for the estimated object (where convergence >= 0 means it has converged 
for most solvers).

There is also the possibility that the solver does converge but to a 
local solution (rare but it does happen) in which case it is also 
possible that the estimated hessian at that point cannot be inverted 
resulting in a warning.

You might find the 'multispec' and 'multifit' methods of value in your 
strategy since they do try to catch such problems and are also setup for 
parallel processing.

Finally, for purely arfima models, try the 'autoarfima' function which 
can also search for non-sequential ARMA orders via complete parameter 
space enumeration.

Regards,
Alexios

On 14/03/2013 02:15, Ranjodh Singh wrote:
> Hi,
>
> I have a rather quick question regarding the rugarch package.
>
> I am attempting to run several ARMA(ar,ma)-GARCH(p,q) models on a single
> set of data whilst varying the values of ar,ma,p and q. For some
> combinations, the GARCH routine does not converge. This is understandable.
>
> The dilemma I am facing is that R halts when convergence is not achieved.
> I would like it output something like "convergence not achieved" and move
> on to the next combination of parameters.
>
> I am hoping that someone else may have asked this question before or its
> part of the official documentation and all someone has to do is to point me
> in that direction.
>
> Thanks in advance.
>
> RJS
>
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> R-SIG-Finance at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
> -- Subscriber-posting only. If you want to post, subscribe first.
> -- Also note that this is not the r-help list where general R questions should go.
>