How Our 2026 Forecast Compares

Where Who Shows Up agrees and disagrees with other forecasters, raters and prediction markets, and how the method would have done in past elections.

As of November 3, 2026: Democrats win the House in 98% of 20,000 simulations (about 234 seats; 218 needed) and the Senate in 85% (about 53 seats). Expected national House vote: D+7.5. Democrats win about 21 of 36 governor's races.

Compare · Updated

How our forecast compares with the major forecasters

Where we agree with the major raters, modelers and prediction markets, and where we don't. None of them feed into Who Shows Up.

"Consensus" is the middle rating across the outlets listed below. A disagreement is at least a step and a half apart on the Safe–Likely–Lean–Toss-up scale, in a race someone rates as competitive.

Senate

Governors

Prediction markets

Each mark is a race with a PredictIt market, colored by the side our forecast favors. Both scales run from a sure Republican win (R 100%) through a coin flip (50/50) to a sure win for the other side (D 100%). On the dashed line, we and the market agree. Above it, our forecast leans further toward the Democratic side than the market does; below it, further toward the Republican side. In Nebraska's Senate race the other side is independent Dan Osborn.

Races where our forecast and the market differ by 10 points or more, naming the side each one favors. PredictIt's prices come from its free data feed, credited to PredictIt, and are converted to chances: each price is the midpoint between the best offers to buy and sell, rescaled so a race's prices add up to 100% (PredictIt's fees push them a little over). Markets with too few trades to set a clear price are left out, which is why only some House races appear. PredictIt limits how much each trader can bet, so its prices can differ from larger markets. We don't use Kalshi or Polymarket because their terms don't allow their prices to be republished or collected automatically.

Backtest: how our method would have done against the pros

Our past-year runs read their race polls from FiveThirtyEight's complete poll lists, using only polls finished by each date. "Competitive" means at least one side rated the race below Safe. Past ratings are as listed on Wikipedia's election pages on each date, where every outlet's column is dated and cited; FiveThirtyEight's probabilities come from its archived forecast data. Only races both sides rated are compared.