Contextual Fundamentals, Models, and Active Management

Abstract

Applying a multifactor alpha model across a diverse range of stocks is a popular approach to forecast securities' expected returns. This approach assumes that one single return-generating equation provides adequate alpha forecasts - that is, one-size-fits-all. In this paper, we extend prior empirical research by introducing an alternative alpha-modeling approach. Our approach presents a parsimonious way of modeling securities individually in order to capture the idiosyncratic return behavior pertaining to different security contexts.

Our investment objective is information ratio maximization, through optimal alpha factor weights. Our modeling technique demonstrates the importance of different factor categories (cheapness, quality, and sentiment) that vary significantly across various security contexts. These contexts span typical dimensions of risk characteristics (value, growth, or earnings variability). We illustrate how practitioners can apply our technique in their security selection process, and document to what extent our approach improves the ex post information ratio when compared with a one-size-fits-all approach.

Keywords: active management, contextual analysis, and equity alpha model

JEL Classification: D40, G12