Skip to main content

​
Install the package

pip install quantcontext-mcp

​
Connect to Claude Desktop

Add this to your claude_desktop_config.json (find it at ~/Library/Application Support/Claude/ on macOS): 
{
  "mcpServers": {
    "quantcontext": {
      "command": "quantcontext"
    }
  }
}
Restart Claude Desktop. You’ll see quantcontext appear in the tools panel.

​
Connect to Claude Code

claude mcp add quantcontext -- quantcontext

​
Verify it’s working

Ask Claude:
“Screen S&P 500 for value stocks with PE under 15 and ROE above 12%”
Claude will call screen_stocks and return a ranked list of candidates with actual market data. The first call downloads and caches market data (~10s). Subsequent calls use the cache (under 1s for screening, 3–8s for backtesting).

​
Try the full workflow

1

Screen for candidates

“Find S&P 500 stocks trading below 12x earnings with return on equity above 15%”
2

Backtest the strategy

“Backtest a strategy that buys those value stocks, monthly rebalance, over the last 2 years”
3

Decompose the returns

“Run factor analysis on the backtest:where is the return coming from? Is the alpha real?”

​
Cache location

Data is cached at ~/.quantcontext/cache/. Total disk usage is ~50MB for full S&P 500 coverage.
The first call to any tool is slow because it downloads market data. After that, screening completes in under a second and backtesting in 3–8 seconds.