Hi,

I was wondering why qq plots for my models with cond.dist = std suggest that standardised residuals have thinner tails than the student distribution with the estimated shape parameter. Then, I found a possible bug in .plot.garch.13, where I would suggest to delete the current definitions of 'skew' and 'shape' and include them into the if statements in the following way:

if (cond.dist == "qnorm")
    .qqDist(sres, dist = cond.dist)
if (cond.dist == "qstd" | cond.dist == "qged")
    .qqDist(sres, dist = cond.dist, nu = x@fit$par["shape"])
if (cond.dist == "qsnorm")
    .qqDist(sres, dist = cond.dist, xi = x@fit$par["skew"])
if (cond.dist == "qsstd" | cond.dist == "qsged")
    .qqDist(sres, dist = cond.dist, xi = x@fit$par["skew"], nu = x@fit$par["shape"])

The reason is that x@fit$params does not include values of the estimated coefficients but only of the (default) starting values.

Best regards,

Michal Miklovic




      ____________________________________________________________________________________

[[elided Yahoo spam]]
	[[alternative HTML version deleted]]