Residual Momentum with Factor-Adjusted Stock Returns

Summary

The document presents a monthly long-short residual-momentum strategy for U.S. equities. It describes ranking stocks by risk-adjusted residual returns over the prior 12 months, skipping the latest month, and buying the strongest decile while shorting the weakest. Residuals are estimated from a rolling 36-month regression on market, size, and value factors. The stated portfolio is equal-weighted and rebalanced monthly, with a universe based on major U.S. exchanges and larger firms.

The code is a QuantConnect implementation that builds monthly price histories, estimates factor returns, fits regressions, and forms long and short holdings. It includes implementation changes to the stated universe, uses a capped coarse universe, applies leverage, and defines a custom fee model. The document supplies code rather than reported backtest results, so it offers no evidence of realized performance. Its implementation also appears to use regression intercepts as residual returns and does not visibly apply the stated one-month skip when ranking; these choices warrant review before treating it as a faithful reproduction. Shorting, leverage, turnover, and model estimates add risks.

Key ideas

• The strategy ranks stocks by residual momentum after adjusting returns for market, size, and value factors.
• It forms a monthly long-short portfolio from the top and bottom deciles of the ranked universe.
• The documented signal uses a 12-month residual-return history estimated from rolling 36-month regressions.
• The code changes the stated universe and includes leverage, short positions, and a custom transaction-fee model.
• The source provides implementation code but no backtest results, and some code choices may not match the described signal.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.