Harvesting the seasons of the size anomaly

This paper employs the DSN portfolio sorting procedure introduced by Lambert et al. (J Banking Finance 114:105811, 2020) to factor size characteristics into returns. The US size anomaly boils then down to a pure seasonal effect, fully supporting the “tax-loss-pruning” hypothesis. We build a long-short calendar trading strategy, easily reproducible by an asset manager, being long the Small-minus-Big (SMB) portfolio in January (or in Q1), staying in cash in Q2 and Q3, and shorting SMB in Q4. The strategy achieves a mean yearly return close to 11% from 1963 to 2019. It remains steady over time, across a variety of subperiods, and resists to the detection of false discoveries. The abnormal returns of the long-short calendar trading strategy withstands realistic transaction costs and short sales limitations.