I have estimated different models based on different distributions. Since they are not nested models of each other, I can't use LR tests. But how can I compare the models? Can I do something with the standardized residuals? Or with the skewness and kurtosis? Or can I plot the computed variance against the returns? And how can I interpret these results?