Forecasts of the 2026 midterm elections from polls, prediction markets, academic models and the PLSC 2219 class model, updated daily.
A dash means no forecast. “Withheld” means we collect the number but cannot publish it yet. The last column is the range of House margins among the forecasts in each source.
| Source | House margin | House seats | P(D House) | Senate seats | P(D Senate) | Internal spread |
|---|---|---|---|---|---|---|
| Polling | D+8.3 | 233 | 85% | 52 | 69% | D+7.8 to D+9.1 |
| Markets | D+8.5 | 234 | 91% | 51 | 64% | D+8.2 to D+8.6 |
| Academic | D+7.9 | 233 | 68% | 50 | 37% | D+1.8 to D+15.8 |
| Class model | D+9.5 | 237 | 95% | 50 | 43% | — |
| Average of the sources · 3 sources | D+8.2 | 233 | 81% | 51 | 57% | D+7.9 to D+8.5 |
Probabilities are the chance of Democratic control. The Senate needs 51 seats. House seats are out of 435 and Senate seats out of 100. The average of the sources counts each source once and leaves out the class model and any withheld number.
Chance of a Democratic win in each competitive seat. Each mark is one source; the grey band runs from the lowest to the highest, and the spread column is the gap in points. Race polls give a margin rather than a chance, so their hollow marks show the chance that margin implies, using the polling model's Senate race error of 5.4 points.
Expected Democratic margin in each competitive seat, in points (cut off at 30). Each mark is one source; the grey band runs from the lowest to the highest, and the spread column is the gap in points. Markets price a chance rather than a margin, so their hollow marks show the margin that would give that chance, using the polling model's Senate race error of 5.4 points. A market price above 95% or below 5% is left out of this view, because near certainty a price stops tracking the margin.
Competitive seats only. All races are on the contests page.