So we are trying to forecasted portfolio return and deviation of a historic set of investments, we have different lengths of data for different investments. My boss calculated the covariance using the all the time periods the two shared, not only time periods available for all data.
I think this could cause an issue, can any one confirm this?