Abstract

Using simulations controlling for the ability to time the equity, bond, and money markets, we compare daily and monthly performance measures. Our main results highlight the joint importance of the fictitious timer’s trading frequency and the data sampling frequency for estimation. Specifically, daily timing measures are superior to those estimated monthly for daily timers, but inferior for occasional or monthly timers. Global measures show more robustness to differences in trading and data sampling frequencies. Finally, conditional measures do not improve upon unconditional ones, and results are similar for performance detection versus ranking.

You do not currently have access to this article.