Seasonal Effects and Other Anomalies

Abstract

We revisit a series of popular anomalies: seasonal, announcement and momentum. We comment on statistical significance and persistence of these effects and propose useful investment strategies to incorporate this information. We investigate the creation of a seasonal anomaly and trend model composed of the Sell in May (SIM), Turn of the Month (TOM), Federal Open Market Committee pre-announcement drift (FOMC) and State Dependent Momentum (SDM). Using the total return S&P 500 dataset starting in 1975, we estimate the parameters of each model on a yearly basis based on an expanding window, and then proceed to form, in a walk forward manner, an optimized combination of the four models using a return to risk optimization procedure. We find that a real-time optimized strategy of the aforementioned four market anomalies produced 9.10% annualized returns with 5.2% volatility and a Sharpe ratio of 0.84. This strategy exceeds the Sharpe ratio of Buy-and-Hold in the same period by more than 100%. Furthermore, the strategy also adds value to the previously published market-timing models of Hull and Qiao (2017) and Hull, Qiao, and Bakosova (2017). A simple strategy which combines all three models more than doubles the Sharpe ratio of Buy-and-Hold between 2003-2017. The combined strategy produces a Sharpe ratio of 1.26, with annualized returns of 18.07% and 13.26% volatility. We publish conclusions from the seasonal trend and anomaly model in our Daily Report.

Keywords: equity premium, seasonal effects, anomalies, forecasting, predictability, market timing, asset returns, tactical asset allocation