2026 Midterms
How forecasters see the 2026 House and Senate races today. These are not Govwatch predictions — we surface the ratings published by Sabato's Crystal Ball (UVA Center for Politics), which combine district fundamentals (partisan lean, incumbent strength, fundraising) with on-the-ground reporting. Ratings update weekly.
Ratings as of (26 days ago).
81 races rated by Sabato's Crystal Ball · 39 competitive
Districts not shown here are considered Safe (D or R) by Sabato's Crystal Ball and are not individually rated in the most recent update.
35 states rated · 7 competitive
Source. All ratings on this page come from Sabato's Crystal Ball, published by the UVA Center for Politics. The Crystal Ball is a long-running non-partisan election forecaster edited by Larry Sabato and Kyle Kondik. Govwatch does not produce its own forecasts.
How we ingest them. Sabato publishes their ratings as a single image each week. We download the image on Tuesday mornings, read it with a vision model, schema-validate the extracted ratings, and upsert into the database. The exact source image for each update is linked above.
What a rating means. “Safe” means very unlikely to flip. “Likely” means the favoured party has a clear edge but the race isn't a foregone conclusion. “Lean” means competitive but tilting. “Toss-Up” means genuinely uncertain. Ratings are revised in response to fundraising, candidate quality, scandals, and on-the-ground developments.
Not a probability. A rating is a categorical judgement, not a percentage. A “Toss-Up” race could plausibly break either way; it doesn't mean 50/50 odds.