The notes say “An advantage of optimization compared with stratified sampling is that optimization takes into account the covariances among the factors used to explain the return on stocks” By taking the covariances among the factors in a multiple regression, isn’t it a multi-collinearity problem (two or more independent variables are correlated)?