SKILL ISSUE

Regimetryx: An Open-Source Reinforcement Learning Financial Market Simulator -Problem Statement Financial markets are constantly changing. A trading strategy that performs well under normal market conditions may fail during periods of high volatility, sudden price movements, market crashes, or changing market trends. However, developing and evaluating trading agents across these different market conditions is difficult for students, independent researchers, and developers who lack access to sophisticated financial research infrastructure. Existing backtesting approaches can also produce misleading results when they fail to account for transaction costs, execution constraints, risk exposure, or the differences between historical market conditions and simulated environments. Comparing trading agents fairly is challenging when they are tested using different assumptions, datasets, or evaluation procedures. There is a need for an accessible, open-source research platform that enables users to build, train, test, and compare algorithmic trading agents under controlled and reproducible market conditions. ~Regimetryx aims to address this problem by providing a configurable financial market simulation and experimentation environment where reinforcement learning agents and traditional trading strategies can be evaluated under identical conditions.** The platform will support multiple market regimes, realistic trading costs, standardized agent interfaces, quantitative performance metrics, and stress-testing scenarios to help researchers understand not only how well an agent performs, but also when, why, and under what conditions it fails. The platform will initially include baseline trading strategies and a Proximal Policy Optimization (PPO) reinforcement learning agent, with support for external agents through standardized interfaces. Its evaluation framework will measure returns, volatility, maximum drawdown, and risk-adjusted performance across normal and stressed market scenarios. The ultimate goal is to democratize quantitative trading research by making systematic experimentation, fair comparison, and reproducible evaluation accessible to students, developers, researchers, and educational institutions without requiring real-money trading. -Expected Outcome A working, open-source platform that allows users to configure market scenarios, run multiple trading agents, compare their performance and risk, reproduce experiments, and identify the market conditions under which different strategies succeed or fail. The platform is intended for research and educational purposes. It will use simulated or historical market data and will not execute real-money trades.