[R-SIG-Finance] [rugarch package] SIgn Bias Test

[alexios ghalanos]

1. Actually, in the paper it is the squared standardized innovations
(z^2) against the non-standardized innovations (residuals) multiplied by
the sign.
2. "fitted" is not what is defined as 'y(t)-mi(theta, x(t))'. This is
called "residuals".
3. The vignette has a typo. The actual test (.signbiasTest) in the
rugarch-tests.R file uses the residuals not the standardized residuals
for the right hand side.

-Alexios

On 24/05/2014 20:44, Zirael wrote:
> Hi,
> 
> I have a question about the regression used in the Sign Bias test of Engle
> and Ng. In the original paper the regression is for squared standardized
> residuals but not against their non-squared values (as it is described in
> the "introduction" to rugarch package) - in the paper it is against the
> innovations (y(t)). Is it the same?
> 
> Or should we even rather use here the "fitted" values of innovations
> (y^(t)= y(t)-mi(theta,x(t)), because in fact the standardized residuals
> come grom ARMA-GARCH model, not GARCH itself, as it is in the original
> paper od Engle and Ng. 
> 
> Many thanks for any help. 
> 
> _______________________________________________
> 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.
> 
>