Attention-Enhanced Reinforcement Learning for Dynamic Portfolio Optimization

2025/10/07 by Pei Xue, Xue, Pei, Yuanchun Ye +1
Decision Sciences · Economics, Econometrics and Finance · #91G10 (Primary) 68T05 #91G60 (Secondary) #Computational Engineering #FOS: Computer and information sciences #Finance #Financial Markets and Investment Strategies #Stock Market Forecasting Methods #and Science (cs.CE)

paper · pdf · doi:10.48550/arxiv.2510.06466

openalex publication_date 2025/10/07 · openalex created_date 2025/10/18 · openalex updated_date 2026/07/28

Abstract

We develop a deep reinforcement learning framework for dynamic portfolio optimization that combines a Dirichlet policy with cross-sectional attention mechanisms. The Dirichlet formulation ensures that portfolio weights are always feasible, handles tradability constraints naturally, and provides a stable way to explore the allocation space. The model integrates per-asset temporal encoders with a global attention layer, allowing it to capture sector relationships, factor spillovers, and other cross asset dependencies. The reward function includes transaction costs and portfolio variance penalties, linking the learning objective to traditional mean variance trade offs. The results show that attention based Dirichlet policies outperform equal-weight and standard reinforcement learning benchmarks in terms of terminal wealth and Sharpe ratio, while maintaining realistic turnover and drawdown levels. Overall, the study shows that combining principled action design with attention-based representations improves both the stability and interpretability of reinforcement learning for portfolio management.

Citations

• Deep Reinforcement Learning for Quantitative Trading
• Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting
• Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting
• Deep Reinforcement Learning for Automated Stock Trading: An Ensemble Strategy
• Temporal Fusion Transformers for Interpretable Multi-horizon Time Series Forecasting
• Proximal Policy Optimization Algorithms
• Learning to trade via direct reinforcement
• Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency
• Common risk factors in the returns on stocks and bonds
• Generalized autoregressive conditional heteroskedasticity
• A five-factor asset pricing model

Related

• A Deep Reinforcement Learning Framework for Dynamic Portfolio Optimization: Evidence from China's Stock Market
• Risk-Aware Deep Reinforcement Learning for Dynamic Portfolio Optimization
• Model-based Deep Reinforcement Learning for Dynamic Portfolio Optimization
• A novel multi-agent dynamic portfolio optimization learning system based on hierarchical deep reinforcement learning
• Portfolio Optimization with 2D Relative-Attentional Gated Transformer