A Risk-Oriented Model for Factor Rotation Decisions
We develop a factor rotation model that is suitably tailored to accommodate severe market conditions and largely to-date neglected phenomena, like factor crowdness, macroeconomic risks (concentration), and sudden factor reversals. Our model uses classification tree analysis on a number of fundamental factor characteristics as well as novel measures we develop through detailed risk attribution analysis of the factor. The model we propose provides significant value when applied in a single-factor setting. The outperformance of our model is even more pronounced when it is used in a dynamic multi-factor setting, where the risk/reward more than triples and the hit ratio improves by about 15% relative to the equally weighted model.