DeePM: Regime-Robust Deep Learning for Systematic Macro Portfolio Management

2026-01-09

AI Summary

This study addresses the challenges in financial deep learning arising from asynchronous information flows, low signal-to-noise ratios, and abrupt shifts in market regimes that undermine strategy performance. To this end, the authors propose an end-to-end deep macro portfolio management system featuring three methodological innovations: a causal delay mechanism to handle asynchronous data, macroeconomic graph priors regularized by economic principles, and a differentiable worst-window penalty that approximates the entropic value-at-risk (EVaR) objective under distributionally robust optimization. The system further integrates lagged cross-sectional attention and explicit transaction cost modeling. In backtests across 50 futures markets from 2010 to 2025, the approach achieves net risk-adjusted returns twice those of conventional trend-following and passive benchmarks, and approximately 50% higher than the Momentum Transformer, while demonstrating robustness during both the CTA winter and the high-volatility post-pandemic period.

Technology Category

• Multiagent Systems: Mechanism Design
• Game Theory and Economic Paradigms: Adversarial Learning
• Machine Learning: Deep Neural Architectures and Foundation Models

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Economics, Online Markets and Human Computation: Sustainability of Web economics
• Security and Privacy: Large-scale security measurements

Abstract

We propose DeePM (Deep Portfolio Manager), a structured deep-learning macro portfolio manager trained end-to-end to maximize a robust, risk-adjusted utility. DeePM addresses three fundamental challenges in financial learning: (1) it resolves the asynchronous"ragged filtration"problem via a Directed Delay (Causal Sieve) mechanism that prioritizes causal impulse-response learning over information freshness; (2) it combats low signal-to-noise ratios via a Macroeconomic Graph Prior, regularizing cross-asset dependence according to economic first principles; and (3) it optimizes a distributionally robust objective where a smooth worst-window penalty serves as a differentiable proxy for Entropic Value-at-Risk (EVaR) - a window-robust utility encouraging strong performance in the most adverse historical subperiods. In large-scale backtests from 2010-2025 on 50 diversified futures with highly realistic transaction costs, DeePM attains net risk-adjusted returns that are roughly twice those of classical trend-following strategies and passive benchmarks, solely using daily closing prices. Furthermore, DeePM improves upon the state-of-the-art Momentum Transformer architecture by roughly fifty percent. The model demonstrates structural resilience across the 2010s"CTA (Commodity Trading Advisor) Winter"and the post-2020 volatility regime shift, maintaining consistent performance through the pandemic, inflation shocks, and the subsequent higher-for-longer environment. Ablation studies confirm that strictly lagged cross-sectional attention, graph prior, principled treatment of transaction costs, and robust minimax optimization are the primary drivers of this generalization capability.

Problem

• regime robustness
• macro portfolio management
• financial machine learning
• distributional robustness
• low signal-to-noise ratio

Innovation

• Directed Delay
• Macroeconomic Graph Prior
• Distributionally Robust Optimization
• Entropic Value-at-Risk (EVaR)
• Causal Sieve