Growing the Efficient Frontier on Panel Trees

TL;DR

This paper introduces P-Trees, a new class of tree-based models for analyzing asset returns that improve the efficient frontier, outperform benchmark models, and balance complexity with interpretability.

Contribution

P-Trees generalize high-dimensional sorting with economic guidance, constructing test assets and factors that enhance asset pricing and portfolio performance.

Findings

P-Trees significantly improve the efficient frontier over standard test assets.

P-Trees' portfolios outperform popular factor models in pricing and investment.

P-Trees achieve high out-of-sample Sharpe ratios with sparse, interpretable models.

Abstract

We introduce a new class of tree-based models, P-Trees, for analyzing (unbalanced) panel of individual asset returns, generalizing high-dimensional sorting with economic guidance and interpretability. Under the mean-variance efficient framework, P-Trees construct test assets that significantly advance the efficient frontier compared to commonly used test assets, with alphas unexplained by benchmark pricing models. P-Tree tangency portfolios also constitute traded factors, recovering the pricing kernel and outperforming popular observable and latent factor models for investments and cross-sectional pricing. Finally, P-Trees capture the complexity of asset returns with sparsity, achieving out-of-sample Sharpe ratios close to those attained only by over-parameterized large models.