When Factor Timing Makes Sense

Abstract

In this paper we investigate when it makes sense for portfolio managers to implement factor timing in their quantitative investing, that is we outline necessary circumstances under which the benefits of factor timing (measured by the improvement in Sharpe ratio and the skewness of the returns) outweighs the challenges associated with development and implementation of factor timing. In particular, we mathematically show that factor timing for a single strategy does not yield substantial improvements unless either (1) the Sharpe ratio of the strategy is orders of magnitude different across states, or (2) the signal used for factor timing can accurately predict when the strategy will deliver negative returns (and the portfolio manager is willing to go short the strategy at that time). On the other hand, using simulation, we provide evidence that for a multi-factor portfolio (containing more than 10 factors) allocating risk based on instantaneous correlations between the factors at the beginning of each time period improves performance above the passive approach of allocating risk based on the long-term factor correlations.