OpenFinGym: A Verifiable Multi-Task Gym Environment for Evaluating Quant Agents
The paper introduces OpenFinGym, a unified gym environment designed for evaluating quantitative‑finance agents across multiple interdependent tasks. It integrates forecasting, market generation, real‑time trading, and fraud detection within a single verification interface and includes tools to automate task creation from finance publications. The platform also offers a containerised runtime with a host‑side verifier to prevent train‑test leakage and enable scalable agent rollouts.
- ▪OpenFinGym consolidates several financial workflow stages—forecasting, strategy construction, risk management, and trading—into one execution environment.
- ▪The system provides an automated pipeline that converts quantitative‑finance research papers into executable task packages.
- ▪A containerised runtime with a verifier service ensures that agents cannot leak training data into testing phases while supporting large‑scale rollouts.
- ▪The platform includes a low‑latency paper‑trading engine and supports deferred‑resolution for long‑horizon forecasts.
- ▪Existing single‑task platforms may overstate agent competence, a limitation OpenFinGym aims to address.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2606.26350 (cs) [Submitted on 24 Jun 2026] Title:OpenFinGym: A Verifiable Multi-Task Gym Environment for Evaluating Quant Agents Authors:Kaicheng Zhang, Wen Ge, Lei Jiang, Weixin Yang, Jordan Langham-Lopez, Jialin Yu, Lukasz Szpruch, Hao Ni View a PDF of the paper titled OpenFinGym: A Verifiable Multi-Task Gym Environment for Evaluating Quant Agents, by Kaicheng Zhang and 7 other authors View PDF HTML (experimental) Abstract:Although large language model agents are increasingly applied to quantitative-finance workflows, their evaluation remains fragmented across isolated tasks, while the financial relevance of benchmark tasks is often overlooked.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv.org.
Discussion
0 commentsMore from arXiv.org
