Agentic Trading Lab (ATL)
Agentic Trading Lab (ATL) is an open-source research and educational platform for studying trading systems powered by large language models. It turns trading ideas into experiments that researchers and students can inspect: configure an agent, run a historical backtest, examine its decisions and portfolio changes, and compare results with market baselines. The research focus extends beyond returns to risk, execution assumptions, and the behavior of agents across market conditions.
Research foundations
The project repository also contains the FinAgent Orchestration Framework, a separate research subsystem associated with the 2025 paper Orchestration Framework for Financial Agents: From Algorithmic Trading to Agentic Trading, coauthored by Xiao-Yang Liu. That work maps components of an algorithmic trading system to specialized agents, including planning, risk, portfolio management, execution, auditing, and memory. Its reported stock and cryptocurrency experiments belong to the paper’s specified experimental settings.
Experimentation and evaluation
The Lab provides a dashboard and interfaces for connecting external agents. Users can inspect trading decisions, simulated trades, equity curves, and evaluation metrics. Shared evaluation windows and buy-and-hold or index baselines support comparison across experiments. The external-agent interface allows researchers to supply their own decision logic while the Lab advances the market simulation and records results.
Ongoing development
Development aims to support a broader ecosystem of reusable agents, evaluation tools, and monitored trading workflows. Capabilities vary by operating mode: the current documentation describes historical backtesting and Alpaca paper-account monitoring; autonomous trading in that paper account is not yet enabled. The dashboard and orchestration research code are distinct components, and their capabilities should be assessed separately. Platform updates, implementation details, and access instructions are available through the resources below.