Smoothing Out Momentum and Reversal

We introduce new path-dependent constraints within a sequential portfolio optimization framework designed to reduce turnover in frequently rebalanced investment strategies, such as momentum and short-term reversal. This method classifies individual assets into distinct groups based on their attractiveness from signal and rebalancing perspectives, effectively managing the trade-off between anomaly-based predictability and the required trading volume for exploitation. These constraints function independently from the ℓ 1 portfolio turnover regularization, which manages reallocation at the aggregated portfolio level, proving more effective in enhancing net profitability. The combined turnover management mechanisms reduce the turnover of daily-rebalanced momentum and reversal portfolios by 95-99%, aligning closely with traditional monthly-rebalanced strategies. Furthermore, our method captures signals more promptly, resulting in more stable portfolios, a substantial reduction in maximum drawdown from 76-99% to 22-49%, and an improvement in risk-adjusted net returns by 38-149%, all under realistic transaction cost assumptions.