36 reputation
bio website
visits member for 2 years, 11 months
seen May 15 '12 at 16:01

comment Is there an optimal covariance one would want forecasts to have?
Yes, I am suggesting saving these historical forecasts and looking at the covariance of the series. It is not necessary to specifying the model that generates the expected returns at all, I'm simply asking whether adding any particular covariance, and specifically expected market covariance, is provably good or inconsequential. I am not suggesting a VAR model specifically at all since transforming the covariance via the stated equation does not add any dependency on lagged values.