Optimality of Naive Investment Strategies in Dynamic Mean-Variance Optimization Problems with Multiple Priors

We study dynamic mean-variance optimization problems with multiple priors. We introduce two types of multiple priors, the priors for expected returns and the priors for covariances. Our framework suggests that the global minimum-variance portfolio is optimal when the investor strongly doubts the correctness of the estimated expected returns, and the equally weighted portfolio is optimal when the investor strongly doubts the correctness of the estimated covariances. From the back tests, we find that for some data sets, the strategy that invests in the global minimum-variance portfolio or the equally weighted portfolio considering the market condition is more efficient than the other mean-variance efficient portfolios