The Effect of Errors in Means, Variances, and Covariances on Optimal Portfolio Choice

What it asks

Michaud describes the problem; Chopra and Ziemba measure it. They compute the cost of errors in expected return, variance and covariance estimates separately. An error in expected return turns out to be an order of magnitude more expensive than the other two. That result explains why the following twenty years leaned mostly toward the covariance side.