Evaluating active investing with generic trading reactions

Abstract

We evaluate the performance of rules using past information to generate daily trading signals. Assuming generic trading reactions to buy and sell signals, we derive an analytic excess return that isolates commissions, interests, the impact of trading timing, and that of the benchmark's choice. The result is useful in dealing with data snooping through leverage and benchmark tweaking. We illustrate the empirical implications by examining trend‐following performance across Dow Jones Industrial Average (1927–2016) and an international sample of major equity indexes and blue‐chip stocks (1980–2016). The results show substantial, fading, non‐persistent and highly methodology‐sensitive excess returns.