Testing the predictive power of ex-post efficient portfolios

Tested here is the hypothesis that portfolios selected from among ex‐post efficient assets will attain better results ex‐ante than by following the naive strategy of holding an equally weighted portfolio of all the assets. The tests are conducted using the returns of the one hundred mutual funds that were continuously in operation from 1959 to 1980. Ex‐post efficient funds are identified using nine investment decision rules (Stochastic Dominance and Mean‐Variance rules with and without riskless asset and the Geometric Mean rule). Ex‐ante performance is assessed in terms of terminal wealth and expected utility. Results indicate that over the twenty‐two years tested, significantly better performance could be attained ex‐ante by investing in mutual funds selected by ex‐post efficiency analysis using the distribution‐free Second‐ and Third‐Degree Stochastic Dominance tests with Riskless Asset as well as by the more traditional Mean‐Variance test with Riskless Asset. Excess returns from using ex‐post information exceed the substantial transaction costs incorporated in the analysis.