openfinclaw-quantitative-research

Install the OpenFinClaw CLI and submit a natural language query like "Backtest a 50/200 SMA crossover on SPY from 2015" using the deepagent research command or MCP tool. The service runs the backtest and returns metrics such as Sharpe ratio, max drawdown, and trade logs.

OpenFinClaw supports US equities, A-shares (沪深), Hong Kong stocks, cryptocurrency, and forex markets. It provides 60+ built-in analysis skills covering technical, fundamental, sentiment, risk, and factor analysis across these markets.

Yes, OpenFinClaw is MCP native and works with Claude Code, Cursor, VS Code, Windsurf, and over 20 AI agents. The install wizard auto-configures MCP for detected agents, or you can manually add the server to your MCP configuration file.

DeepAgent backtests can take 2-5 minutes for complex strategies. Check task status with the deepagent threads or status commands, and cancel a running task with the cancel command if needed.

Set the OPENFINCLAW_API_KEY environment variable with your fch_ prefixed key, or re-run the init wizard to store it in ~/.openfinclaw/config.json. The config file is auto-created with chmod 600 permissions.

The free tier has request rate limits that can trigger rate limit exceeded errors. You can upgrade at hub.openfinclaw.ai, add delays between batch requests, and cache results to stay within quotas.