ai-trader

Why we included this project

Most backtesting setups end up as a stack of notebooks you can't reproduce six months later. This project takes the opposite route: a config-driven Backtrader framework where backtests are defined in YAML files, so you can rerun a strategy after a parameter tweak and get the same result, not a different one. It ships over twenty strategies out of the box, from moving-average crossovers to adaptive models, and the CLI runs them against US stocks, Taiwan stocks, crypto, and forex without you writing glue code. The MCP server is the interesting part: an assistant like Claude can fetch data, list strategies, and trigger backtests through natural language, which makes the tool a practical bridge between quant research and LLM-driven workflows. If you prefer to stay in Python, the library API lets you call the same engine from your own scripts.