On Factor Model Construction

The academic literature identifies several persistent sources of cross-sectional return: value (Fama and French, 1992), momentum (Jegadeesh and Titman, 1993), quality (Novy-Marx, 2013), and low volatility (Baker, Bradley, and Wurgler, 2011). Each factor has periods of strong and weak performance. A strategy built on one factor alone is exposed to extended drawdowns when that factor underperforms.

We construct composite signals by combining multiple factor scores into a single ranking. The combination weights are not optimized on historical data. They are set equal or based on economic reasoning. Optimized weights tend to overfit to the sample period and degrade out of sample.

Each factor score is computed from standardized cross-sectional ranks, not raw values. Ranking removes the influence of outliers and makes signals comparable across factors with different units. A stock that ranks in the top decile for both value and momentum receives a higher composite score than one that ranks highly on only one dimension.

We test the composite against each individual factor using walk-forward validation. The composite must outperform its components on a risk-adjusted basis out of sample. If it does not, the combination adds noise rather than signal, and we discard it.

References

• Fama, E. F. and French, K. R. (1992). "The Cross-Section of Expected Stock Returns." Journal of Finance.
• Jegadeesh, N. and Titman, S. (1993). "Returns to Buying Winners and Selling Losers." Journal of Finance.
• Novy-Marx, R. (2013). "The Other Side of Value: The Gross Profitability Premium." Journal of Financial Economics.
• Asness, C., Moskowitz, T., and Pedersen, L. (2013). "Value and Momentum Everywhere." Journal of Finance.