Advancing Style Analysis and Risk Modeling by Incorporating Model Uncertainty with Bayesian Model Averaging

Abstract

Investors have increasing interests in sophisticated yet transparent analytic tools to handle model uncertainty, tail risk and market dynamics. This paper demonstrates how macroeconomic factor models, based on Bayesian model averaging (BMA), can help address the challenges in some specific investment analytic tasks from three perspectives: (1) selecting risk factors and estimating risk factor exposure in risk allocation, (2) modeling the dynamic exposure of multiple asset classes to systematic risk; and (3) forecasting risk/return profiles of hedge fund investment strategies.