Research map / Reinforcement learning / Offline RL
MetaTrader: Learning to Generalize RL Trading Policies Beyond Offline Data
Haochen Yuan, Minting Pan, Yunbo Wang, Siyu Gao, Xiaokang Yang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add transformation-based worst-case TD target (vs SAC TD) (ablation) | MetaTrader · CSI-300 online adaptation | cumulative return: 1.68 -> 1.84 | Improved |
| use transformation-based TD instead of min over 5 Q-networks (ablation) | MetaTrader · CSI-300 online adaptation | cumulative return: 1.49 -> 1.84 | Improved |
| use transformation-based TD instead of mean over 5 Q-networks (ablation) | MetaTrader · CSI-300 online adaptation | cumulative return: 1.39 -> 1.84 | Improved |
| add bilevel optimization in in-domain finetuning (ablation) | MetaTrader · CSI-300 online adaptation | cumulative return: 1.78 -> 1.84 | Improved |
| add bilevel optimization in in-domain finetuning (ablation) | MetaTrader · NASDAQ-100 online adaptation | cumulative return: 1.41 -> 1.58 | Improved |
| use transformed data during finetuning (ablation) | MetaTrader · CSI-300 online adaptation | cumulative return: 1.84 -> 0.84 | Hurt |
| add F3 downsampling transform only (ablation) | MetaTrader · CSI online adaptation (3 seeds) | cumulative return: 1.66 -> 1.73 | Improved |
| add F1 or F2 alone (ablation) | MetaTrader · CSI online adaptation (3 seeds) | cumulative return: 1.66 -> 1.69 (F1), 1.67 (F2) | No clear effect |
| combine F1+F2+F3 transformations (ablation) | MetaTrader · CSI online adaptation (3 seeds) | cumulative return: 1.66 -> 1.84 | Improved |
| replace StockFormer with MetaTrader | RL trading · CSI offline evaluation | cumulative return: 1.24 -> 1.44 | Improved |
Closest papers by meaning
- ROIDICE: Offline Return on Investment Maximization for Efficient Decision Making (NeurIPS 2024 · similarity 0.86)
- Robust Reinforcement Learning in Finance: Modeling Market Impact with Elliptic Uncertainty Sets (NeurIPS 2025 · similarity 0.86)
- Improving Offline RL by Blending Heuristics (ICLR 2024 · similarity 0.86)
- Augmented Lagrangian Risk-constrained Reinforcement Learning for Portfolio Optimization (Student Abstract) (AAAI 2025 · similarity 0.85)
- EarnHFT: Efficient Hierarchical Reinforcement Learning for High Frequency Trading (AAAI 2024 · similarity 0.85)
- R&D-Agent-Quant: A Multi-Agent Framework for Data-Centric Factors and Model Joint Optimization (NeurIPS 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
