Uncertainty-Adjusted Sorting for Asset Pricing with Machine Learning

2026-01-02

AI Summary

This study addresses a key limitation in traditional asset pricing—portfolio construction based solely on point forecasts, which ignores estimation uncertainty at the individual asset level. The paper proposes a novel approach that explicitly incorporates asset-specific uncertainty into machine learning–driven asset ranking by replacing point predictions with uncertainty-adjusted prediction intervals for expected returns. Integrating multiple machine learning models with uncertainty quantification techniques, the method remains robust even under partially misspecified or imperfect uncertainty estimates. Empirical results demonstrate that this strategy significantly enhances portfolio performance, particularly when used with complex predictive models, primarily through a substantial reduction in portfolio volatility.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Stochastic Optimization
• Multiagent Systems: Multiagent Systems under Uncertainty

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

Machine learning is central to empirical asset pricing, but portfolio construction still relies on point predictions and largely ignores asset-specific estimation uncertainty. We propose a simple change: sort assets using uncertainty-adjusted prediction bounds instead of point predictions alone. Across a broad set of ML models and a U.S. equity panel, this approach improves portfolio performance relative to point-prediction sorting. These gains persist even when bounds are built from partial or misspecified uncertainty information. They arise mainly from reduced volatility and are strongest for flexible machine learning models. Identification and robustness exercises show that these improvements are driven by asset-level rather than time or aggregate predictive uncertainty.

Problem

• asset pricing
• machine learning
• estimation uncertainty
• portfolio construction
• prediction uncertainty

Innovation

• uncertainty-adjusted sorting
• machine learning
• asset pricing
• prediction bounds
• portfolio construction