40%
Prediction markets
Kalshi & Polymarket Dem prices, mildly de-vigged in log-odds space.
30%
Polls, quality-weighted
Recency decay, sample & LV/RV weights, pollster quality tiers, ±2.5 pt partisan-sponsor adjustment, shrunk toward the ratings prior.
20%
Expert ratings
Cook Political Report and Sabato's Crystal Ball, plus Inside Elections as a fourth rater for Senate races (and Silver's label when no Silver % is published).
10%
Silver Bulletin
FLIPR win probability where published. Components combine in log-odds, with a candidate-quality adjustment.
How to read it: “Dem win %” is the model's probability the Democratic candidate (or, in NE-SEN, independent Dan Osborn's coalition side as coded in the workbook) wins that race. Chamber totals come from 25,000 simulated elections using a hierarchical Student-t copula — national and same-state polling errors move together, with fat tails for wave nights. Probabilities are model estimates as of Oct 31, 2026, not results. Oct 31 update: Inside Elections joined as a fourth Senate rater — TX-SEN and OH-SEN moved Toss-up → Tilt D; IA-SEN and KS-SEN moved Tilt R → Toss-up; SC-SEN and NE-SEN moved Lean R → Tilt R; MS-SEN moved Solid R → Likely R (LA-SEN also moved, but is not among the 116 competitive races tracked here). The Senate majority chance is 72.4% (52.0 expected seats, 80% range 48–55). Every race, probability, favorite, rating, and chamber total was cross-checked against the v7 workbook and the v5.1 model rerun. The Electorate scenario slider applies a uniform national swing to every race — each race's win probability is shifted in probit space (σ = 7.5 for House, 6.0 for Senate and governor) — and reruns the same hierarchical Student-t simulation in your browser (6,000 seeded draws; national shock 0.25, state shock 0.20, idiosyncratic 0.55), so seat totals, 80% ranges, and majority chances update live. Presentation redesigned in the clean, rows-first style of Decision Desk HQ election pages; all data and model behavior are unchanged.