Regime-based Factor Allocation: An Empirical Study of Regime-Bbsed Factor Timing Strategies in the US Stock Market

Abstract

This thesis investigates whether regime-based factor timing strategies can outperform a static diversified multi-factor model. It combines business cycle regimes with factor investing in the US stock market using data from 1998-2021. The business cycle is split into four regimes, Recovery, Expansion, Slowdown, and Contraction. This paper analyzes factor performances in these regimes to find the optimal factor allocation in each regime. The paper found the Size and Value factors to be the factors with the best performance under the Recovery regime. Momentum was the best performing factor in both the Expansion and Slowdown regimes, whereas the factors, Quality and Minimum Volatility, showed superior returns in the Contraction stage. Two regime-based factor timing strategies have been constructed in this paper, namely the Regime-Based Mean-Variance Model and the Regime-Based Factor Model. Both models showed significant annual outperformance relative to the static Equally Weighted Model by 0.95 and 2.55 percentage points, respectively. Taking transaction costs into account for the Regime-Based Factor Model, the annualized return was reduced by 0.31 percentage points. Even after accounting for transaction costs, this dynamic model still showed significant outperformance relative to the static model.