This paper compares the excess of return of the dynamic investment model for four classes of estimators of the mean returns (the simple historic or sample estimator, the James-Stein estimator, the Bayes-Stein estimator and a CAPM based estimator) in the Spanish asset market. The sample estimator is clearly affected by estimation error. The three other estimators are the principal methods have been used to deal with estimation risk in the means, and we conclude that none of them significantly outperform the sample estimator. Furthermore to that, monthly portfolio rebalancing does not seem to be significantly better than more passive strategies.

Keywords: expected return estimation, risk estimation, portfolio management.