How Different Are Alternative Beta Strategies?

Abstract

In this article, the authors use a quantitative approach to compare different alternative beta strategies, based on statistical relationships among their returns. Using correlations, principal component analysis, regression factor models, and minimum spanning tree graphs, they identify and quantify statistical closeness of these portfolios. The results show that when measured by return comovements and common systematic risk exposures, different alternative beta portfolios are, on average, quite close to each other. Surprisingly, in some cases, returns of portfolios with different strategic approaches can be more similar than those of two portfolios representing different variations of the same approach. Using a formal clustering technique, the authors show how to identify distinct clusters within a set of alternative beta portfolios. Given potential redundancy of alternative beta, their clusters can give a better diversified set of building blocks for multi-strategy allocations than individual strategies themselves. The authors build several portfolio allocations using clusters of alternative beta strategies as building blocks and compare individual strategy-based and cluster-based allocations, within both static and dynamic allocation frameworks. They find that the cluster-based allocations have a better risk–return profile with respect to the portfolios based on individual strategies.