TradingAgents

Why we included this project

TradingAgents is a reference implementation for anyone building or experimenting with LLM-driven trading systems. It splits a trading workflow into a team of specialized agents: fundamental, sentiment, news, and technical analysts, plus bullish and bearish researchers who debate, a trader that composes their reports, and a risk-management layer feeding a final portfolio manager. That role-based structure is worth studying even if you never trade, because it shows how to orchestrate multiple LLM agents with distinct responsibilities and a clear decision pipeline. The project runs as a Python package or Docker container, works with providers from OpenAI and Anthropic to local Ollama, and includes backtesting and checkpointing so you can evaluate strategies without risking live capital. The authors explicitly frame it as a research tool rather than financial advice, so treat it as a framework to learn from and adapt, not a turnkey trading product.