NFL pick'em tools

In development

An MCP server that gives a language model live access to rating systems, market lines, and my pick history for pick'em strategy, plus a mobile app for win-total predictions and pick logging.

Illustration of predicted win totals compared with market lines

Two related side projects, built mostly to learn the Model Context Protocol (MCP) and agent tooling on a problem I care about.

The MCP server

Gives a language model live access to rating systems, market lines, line movement, and my own pick history, so it can help with against-the-spread picks and track which signals have actually been accurate.

The app

A free mobile app for season-long win-total predictions. It combines several team rating systems into an ensemble, weighted by how well each has predicted in the past, and lets anyone log their own picks.

Status

In development, with Python for the server and React Native and Expo for the app.

Built with Python, MCP, React Native, Expo.