Growing the Efficient Frontier on Panel Trees

We develop a new class of tree-based models (P-Trees) for analyzing (unbalanced) panel data using economically guided, global (instead of local) split criteria that guard against overfitting while preserving interpretability. To generalize security sorting and better estimate the efficient frontier, we grow a P-Tree top-down to split the cross section of asset returns to construct stochastic discount factors and test assets under the MVE framework, visualizing (asymmetric) nonlinear interactions among firm characteristics (and with macroeconomic states). When applied to U.S. equities and especially when boosted, P-Trees significantly advance the efficient frontier relative to those constructed with established factors and common test assets. Diversified P-Tree test portfolios exhibit significant unexplained alphas against benchmark factor models. P-Trees also outperforms most known observable and latent factor models in pricing cross-sectional returns, delivering transparent trading strategies, and generating risk-adjusted investment outcomes. Beyond asset pricing, our framework offers a more interpretable and computationally efficient alternative to recent machine learning and AI models for analyzing panel data through goal-oriented, high-dimensional clustering.