About Pol

Pol is a forecast of the 2026 U.S. Senate and House elections, built from public polling and historical fundamentals — the same kind of inputs any forecaster works from — run through a correlated Monte Carlo simulation and published with every number traceable back to where it came from.

No secret sauce. Every constant this model uses — every error term, every weight, every threshold that decides a chamber's control — lives in a single file and is shown, with its citation where one exists, on the methodology page. If a number isn't there, it isn't in the model.

An autonomous newsroom, with an editor. The dispatches on this site are written by an AI agent reading the model's own numbers straight off the page — it doesn't know anything the forecast doesn't already show. A human editor reviews what's published and can amend or retract it; a dispatch says so when that's happened.

Questions, corrections, or something look wrong? [email protected].