OpenFinGym: A Verifiable Multi-Task Gym Environment for Evaluating Quant Agents

2026-06-24

AI Summary

Existing quant agent evaluation platforms suffer from task isolation and a lack of financial relevance, making it difficult to assess agents’ generalization and decision-making capabilities within multi-stage financial workflows. This work proposes a unified multi-task training and evaluation framework that encompasses core components such as forecasting, market simulation, real-time trading, and fraud detection. It introduces a novel pipeline that automatically converts quantitative finance research papers into executable task bundles. The system integrates a containerized runtime environment with data leakage prevention, a delayed settlement mechanism, a low-latency paper-trading platform, and support for both supervised fine-tuning and reinforcement learning-based post-training. This architecture enables scalable, verifiable end-to-end evaluation, significantly enhancing the ability to characterize agents’ generalization and decision robustness in realistic financial scenarios.

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. Yet financial workflows are inherently multi-stage, spanning interdependent tasks such as forecasting, strategy construction, risk management, and trading. Existing platforms typically focus on a single task, and can therefore overstate agent competence and fail to reveal weaknesses in generalization, real-market interaction, and financially meaningful decision-making. We introduce OpenFinGym, a unified gym environment for quantitative-finance agent development that covers forecasting, market generation, real-time trading, and fraud detection under a single execution and verification interface. OpenFinGym additionally provides an automated task-construction pipeline that turns quantitative finance publications into executable task packages; a containerised runtime with a host-side verifier service that supports scalable agent rollouts and prevents runtime train-test leakage; a paper trading engine with a low-latency data-stream design; deferred-resolution support for long-horizon and event-market forecasts; and integration for SFT and RL post-training

Problem

• quantitative finance
• multi-task evaluation
• agent benchmarking
• financial workflows
• verifiable environment

Innovation

• multi-task reinforcement learning
• quantitative finance agents
• verifiable evaluation environment
• automated task construction
• train-test leakage prevention