Multi-factor style rotation – an empirical study in the US stock market

ABSTRACT

To evaluate the rotation characteristics of different styles, this study employs the design of experiments (DOE) to systematically combine factor weights to generate stock selection models that express various stock selection styles. This study uses the S&P 500 constituent stocks as the stock pool. At the beginning of each month, we select the highest-scoring 4% stocks with a weighted scoring method and hold them until the end of the month. The back-test period goes from 1999/Jan to 2020/June, a total of 21.5 years (258 months). The results show that (1) many stock selection models mixing value style with growth style have high synergy effect. (2) many models have positive autocorrelations according to the Durbin–Watson (DW) statistics of each model’s time series of monthly excess rate of return. (3) the candidate models of the style timing strategy should include a value style factor, a growth style factor, and a two-factor model with them.

KEYWORDS:

• Style rotation phenomenon
• style timing strategy
• multi-factor model
• design of experiments

Disclosure statement

No potential conflict of interest was reported by the author(s).