rates · SOFR · mcp
Rates Engine
Deterministic SOFR rates engine: curves, futures with convexity adjustment, swaps, hedging and options, plus USD/MXN. Every result returns the number and the evidence it rests on, with no AI at runtime.
Overview
A deterministic SOFR rates engine with the same conventions as the Optimization Engine. Every result returns the number and the evidence it rests on. The core depends on numpy, pandas and scipy, and nothing else. Built with Eduardo Ramos.
What it does
- Curve construction: discount, zero, par and forward, with monotone-convex interpolation, Nelson-Siegel and FOMC-step curves.
- SOFR futures with convexity adjustment; OIS, IRS and FRA pricing under collateral discounting.
- Key-rate and duration risk, and swap hedging with futures strips.
- Swaptions, caps and floors under Bachelier and Black, with a SABR volatility cube.
- A second currency: USD/MXN forwards and options, and a hedge-structure comparator.
- An MCP server, so agents call the engine as tools.
Status
In active development. Not on PyPI yet; the repository and the API reference are public.