Combining Factors

While the academic literature primarily investigates factor exposures based on covariances (i.e. beta exposure), most practitioners apply characteristics-based scorings to obtain factor portfolios. It hereby remains largely unexplored how firm-level characteristics can be combined to obtain optimal factor portfolios. This paper derives multi-factor portfolios that are formed via a combination of stock characteristic scores. Portfolios that are formed on multiple characteristics are less volatile, and exhibit higher after cost returns compared to the market and single factor portfolios. In addition, return, risk and turnover preferences are very sensitive to buy- and sell-thresholds. We further identify optimal weights for individual factor characteristics, but have to recognize the 1/N factor portfolio as a tough benchmark.