Evaluating Time-Series Momentum in S&P 500 Stocks: Mean-Variance Optimization, Transaction Costs, and Market Regimes

Abstract

As a classic trend‑following strategy, the time‑series momentum (TSMOM) strategy poses an empirical challenge to the efficient‑market hypothesis. Most existing back‑testing studies focus heavily on index and futures markets, while empirical research specifically targeting individual stocks remains limited, and the impact of trading costs on investment returns tends to be underestimated. Using daily data for S&P 500 constituent stocks covering 2015‑2025, this paper constructs the baseline TSMOM strategy as well as the MVO‑TSMOM strategy integrated with dynamic mean‑variance optimisation. This study adopts a multi‑layer trading‑cost model and conducts tests under diverse market regimes, including bull, bear and range‑bound markets. Empirical results indicate that the mean‑variance optimisation strategy delivers a moderate improvement in risk‑adjusted returns. Nevertheless, excess returns are substantially eroded owing to high portfolio turnover caused by monthly rebalancing. The strategy can only provide limited downside protection in bear‑market environments when the round‑trip transaction cost is extremely low (0.1%). No significant alpha is identified across the full sample period, suggesting that none of the strategies can consistently outperform the SPY index. Firstly, this paper provides empirical evidence for the real‑world performance of the TSMOM strategy in global equity markets amid trade frictions. Secondly, it offers empirical evidence for the boundary conditions of the time‑series momentum anomaly.