Less is More: Ranking Information, Estimation Errors and Optimal Portfolios

We propose a novel approach that mitigates the crippling effects of parameter uncertainty and estimation errors on the out-of-sample performance of mean-variance optimized portfolios. Rather than relying on exact forecasts, we argue investors should base optimizations on ranking or grouping information, which reduces the informational content of input parameters and, in turn, eliminates outliers caused by estimation errors. The resulting portfolios are better diversified, less concentrated, and exhibit statistically and economically significant improvements in Sharpe ratios relative to both plug-in mean-variance and naively diversified portfolios.