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

Paper: arXiv 2601.05975

Authors: Kieran Wood, Stephen J. Roberts, Stefan Zohren

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.

Complexity vs Empirical Score

• Math Complexity: 6.5/10
• Empirical Rigor: 8.5/10
• Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: DeePM presents a compelling end-to-end deep learning framework for macro portfolio management that effectively combines temporal attention, graph priors, and causal filtration. The paper is empirically robust, featuring extensive backtesting over 15 years on 50 futures contracts with explicit transaction costs and detailed ablation studies. Its primary innovation lies in mapping differentiable worst-window penalties directly to Entropic Value-at-Risk optimization.

Research Flowchart

  flowchart TD
    A["Research Goal: Robust Deep Learning for Macro Portfolio Management"] --> B["Inputs: 50 Diversified Futures & Daily Closing Prices (2010–2025)"]
    
    subgraph C ["DeePM Framework & Methodology"]
        C1["Directed Delay (Causal Sieve)"]
        C2["Macroeconomic Graph Prior"]
        C3["EVaR-based Worst-Window Utility"]
    end
    
    B --> C
    
    C --> D["End-to-End Optimization with Realistic Transaction Costs"]
    
    D --> E["Key Outcomes:<br/>• Net risk-adjusted returns ~2x over classical trend/passive benchmarks<br/>• Outperforms Momentum Transformer by ~50%<br/>• Structural resilience in 'CTA Winter' & post-2020 regimes"]

Cited References (98)

1. Robust Optimization (2009)
2. Robust Optimization (survey/overview) (2013)
3. Probability and Measure (1995)
4. Convex Optimization (2004)
5. Learning Models with Distributionally Robust Optimization (2021)
6. A Weak Convergence Approach to the Theory of Large Deviations (1997)
7. Semi-Supervised Classification with Graph Convolutional Networks (2017)
8. Temporal Fusion Transformers for Interpretable Multi-horizon Time Series Forecasting (2021)

1 of 98 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.