Composite Alpha Factor and Portfolio Rebalance on Factor Tilts

Abstract

Quantitative fund managers often combine factors to build “composite alpha factor” in their portfolio construction process. In this paper we use several factors to explain stock returns, and then calculate both expected return and risk of portfolio based on those factors. First, we calculate optimal alpha weights for composite alpha factor by static solution. In this calculation, we use covariance matrix of factor returns and correlation of factor exposures in addition to expected values of factor returns. Second, supposing time variability of factor exposures, we calculate optimal alpha weights for composite alpha factor and optimal rebalance volume for factor tilt portfolio by dynamic solution under existence of trading costs. Next, using historical data in Japanese equity market, we show numerical example of composite factor from three factors by static solution. And we show how much we have to adjust portfolio for typical factors by dynamic solution when we consider time variability of factor exposures. Our results suggest that we should use dynamic solution when we adopt factors whose exposures change fast.