The Democrats are favored. A normal polling miss could change that
Forecasters and markets give the Democrats the House at 90% or more. Replay the polls' recent misses and the Senate, and often the House, slips away.
Polling Forecast
2 min read
Written in house from our poll data and our forecast.
Most coverage of the midterms now treats the result as settled. Nate Silver's Silver Bulletin gives the Democrats a 93% chance of winning the House and 75% for the Senate. The Kalshi betting market has 90% and 62%. Our own projection is more cautious, but it still has the Democrats ahead: 233 House seats and a 81% chance of the majority, and 53 Senate seats with a 66% chance of control.
All of those numbers rest on the polls, and the polls are never exactly right. The question every cycle is only which way they miss, and by how much.
What the last twelve elections say
We replay every past polling miss since 2002 on today's forecast. Each replay takes that year's average error in the final Senate and governor polls and moves every race in our projection by the same amount.
In five of the last six elections, the polls overstated the Democrats: by 2.3 points in 2024, 1.1 in 2022, 6.9 in 2020, 3.9 in 2016 and 3.3 in 2014. Only 2018 was close to level.
Repeat the 2014, 2016, 2020 or 2024 miss and the Republicans keep the Senate. Repeat 2014, 2016 or 2020 and they keep the House too. Even a repeat of 2022, a good year for the polls, leaves the Democrats with 51 Senate seats, where 51 are needed for control, and 229 House seats, against the 218 needed for a majority.
None of those were freak years. They are four of the last six elections. Pollsters have changed their methods after each miss, and the next miss has still come. Whether 2026 breaks the pattern will only be known on November 3.
The misses that went the other way
The record is not one-sided. In 2006, 2010 and 2012 the polls understated the Democrats, by 0.8, 1.7 and 2.4 points. Replay those years and the Democrats do better than we project: up to 245 House seats and 54 in the Senate. Across all 12 replays, the Democrats keep the House in 9 and the Senate in 8.
Our projection already allows for the usual miss. It takes about 1.4 points off the Democrats compared with the polls as they stand. A replay comes on top of that, so it is a test of how much a normal miss would matter, not a second forecast.
What it means
The Democrats are favored, and we say so. But a 81% chance is not 93%, and a 66% chance in the Senate is not far from a coin toss. It would not take anything unusual for election night to disappoint them, only for the polls to miss the way they did in 2014, 2016, 2020 or 2024.
Polling Forecast reads every poll in ten countries, corrects for how pollsters have missed before, and turns them into seats. Our forecasts, the polls and our methods are free to explore.
Elsewhere on Polling Forecast
353 seats? The mega-poll's Labour majority comes from a split right, not tactical voting
Run the mega-poll's own vote shares through our model and Labour still gets 358 seats. The difference from our 316 is a weaker Conservative Party and a stronger Restore Britain, not the tactical voting the coverage focused on.
Polls understated PiS in 31 of 32 final surveys since 2024
At four national votes since Tusk took office, final polls put PiS or its candidate too low. Our forecast makes no allowance for it.
Private polls understated Spain's PSOE at six of eight elections
104 of 142 private final polls since 2004 had the PSOE too low, but even a repeat of 2023 would probably leave the PP and Vox a majority.