Skip to main content

​
1. Pick the Server

Use the MCP Server Catalog to choose the server for the workflow.
GoalStart here
Load OHLCV data or indicatorsDataloader Server
Compute volatility estimatesVolatility Scaling Lab Server
Fit statistical factor modelsStatistical Factor Models Server
Fit regime-switching JumpModelsJump Models Server
Generate ExtremeHurst signalsParallax ExtremeHurst Server

​
2. Add It to Your Client

{
  "mcpServers": {
    "volatility-scaling-lab": {
      "url": "https://volatility-scaling-lab-production.up.railway.app/mcp/qca-open-loop-volatility"
    }
  }
}
See MCP Client Setup for all servers and generic HTTP examples.

​
3. Run a Small Tool Call

Ask your MCP client to call compute-ewma-volatility-series with this payload: 
{
  "request": {
    "ticker": "SPY",
    "start_date": "2020-01-01",
    "halflife": 252,
    "annualize": true
  }
}

​
4. Handle Large Outputs

If a tool returns a result_id, use the result/artifact helpers documented in Shared MCP Tools. The usual path is preview, fetch if needed, then save as an artifact when the output should be downloaded.

​
5. Next Steps

Browse servers

Compare all MCP endpoints and choose the right one for each workflow.

Client setup

Configure Cursor, Claude Desktop, and generic streamable HTTP clients.