Use AI to price derivatives

October 7, 2026

Connect Quantra to your AI agent and use natural language, screenshots and documents for your quantitative calculations and analysis.

Quantra has an MCP server. Add it to your AI agent (Claude, Cursor, Claude Code or any MCP client) and ask. Every number comes from the engine, every input is either yours or sourced and cited, and when you ask how something was computed the agent answers from the source code, file and line, with a link.

With most pricing systems you get a number but not how it was computed. The models, the conventions and the code are closed, so when a price differs from yours, explaining it means opening a ticket, reading a methodology document, or trusting it. Quantra is open source, so the agent can show you the exact code behind a number, and test an explanation by changing one input and pricing again.

Price a swap from a screenshot

Paste a screenshot of a trade and of its curve and ask for the price. The agent builds the curve, prices the trade and compares the result with the number on your screen.

Price a structured note from its term sheet

Attach a note's pricing supplement and ask "Can you price this as of the trade date?". The agent reads the terms, sources the market data from public publishers with a link for each value, prices the note and compares it with the issuer's estimated value. Then ask where the difference comes from, or which volatility model is used and where it is in the code.

In claude.ai, turn on web search and allow Claude in Chrome when it asks.

The public server runs on the demo machine, without authentication and rate limited. It is for trying things out, not for production use.