Heteroscedasticity - Detection 

Thank you so much.

Thanks Kian, for your question and sorry for my delayed response. Residuals are not part of this regression inquiry. In this example, they only serve the purpose of helping us determine if we have the problem of heteroscedasticity. To your other question, in any regression study, you, the researcher, should begin by identifying your Y variable because that is the subject of your investigation. According to the CAPM, stock return (Y) is a function of risk, measured by beta (X). Hope this helps :-)

No you can't do log, unless you transform all your data somehow to force them all to be positive.

DW test is for serial correlation. I'm not sure it can be use for heteroscedasticity, unless there's a variation of it.