A note on the out-of-sample performance of resampled efficiency

Abstract

The concept of resampled efficiency (RE) is debated both in academia as well as among practitioners. For supporters of RE the litmus test seems to be out-of-sample performance. While Markowitz and Usmen have shown that RE outperforms a Bayesian alternative, the present study is able to reverse their results. The key is to understand that Bayesian methods are literally impossible to test out-of-sample. For every distribution, a prior will be found that will outperform resampling (and vice versa). Equally, for every prior, a distribution will be found where resampling outperforms. The fact that one method outperforms another for a given set of data means little. In the absence of theory, investors do not know when one method will outperform the other, as they do not know the true distribution.