Midterm prediction markets are more confident about Democrats than Decision Desk HQ’s forecast model, but less so than Silver Bulletin’s, with Election Day on November 3. On the morning of October 29, prices on Kalshi and Polymarket put Democratic control of the House near 92% and the Senate near 67%. Decision Desk HQ’s model, by contrast, had 80% and 58%.
Key facts (as of October 29, 2026)
- Averaging Kalshi and Polymarket, the DefiRate tracker put Democrats at about 92% for the House and 67% for the Senate on October 29.
- Decision Desk HQ’s model gives Democrats an 80% chance in the House and 58% in the Senate (updated October 28).
- Silver Bulletin’s model gave Democrats 95% for the House and 76.4% for the Senate in its October 22 update, Newsweek reported.
Democrats win control of the House and Senate, networks project
This article explains what the numbers mean. It is not betting advice, and none of these figures are results.
What are midterm prediction markets?
A prediction market is an exchange where people buy and sell contracts tied to an event. A contract pays $1 if the event happens and nothing if it doesn't. So a price of 67 cents is read as the crowd's estimate of a 67% chance.
A poll, by comparison, asks a sample of voters whom they plan to back. Fox 35 Orlando summed up the gap neatly: polls measure what voters say, while markets reflect what traders expect to happen. A forecast model sits between the two, since it turns polls and other data into odds.
How do the markets compare with forecast models?
All three sources agree on the direction: Democrats are favored to win the House, and the Senate is closer. However, they disagree on how close. For the numbers on the eve of the vote, our roundup of the final House and Senate forecasts compares them side by side.
| Source (date) | Democratic House chance | Democratic Senate chance |
|---|---|---|
| Kalshi + Polymarket average (Oct 29) | 92% | 67% |
| Decision Desk HQ model (Oct 28) | 80% | 58% |
| Silver Bulletin model (Oct 22) | 95% | 76.4% |
Market average via DefiRate; Silver Bulletin figures via Newsweek. On the Senate, DefiRate listed Kalshi at 66.3% and Polymarket at 67.5%, and it notes that Kalshi's figure is built from two contracts.
The models also differ from each other. Silver Bulletin's average outcome was a 52–48 Democratic Senate, while Decision Desk HQ projects 51 Democratic seats. In other words, the spread between two models can be as wide as the spread between a model and a market.
Where do midterm prediction markets and models disagree most?
The biggest gaps show up in individual Senate races. Here is Kalshi's Democratic price on the morning of October 29 next to Decision Desk HQ's Democratic chance in the same race.
| Senate race | Kalshi (Dem) | DDHQ model (Dem) | Gap |
|---|---|---|---|
| Texas | 69% | 57% | 12 points |
| Ohio | 68% | 55% | 13 points |
| Alaska | 69% | 61% | 8 points |
| Maine | 62% | 56% | 6 points |
| Michigan | 71% | 69% | 2 points |
| Iowa | 48% | 46% | 2 points |
Kalshi prices via DefiRate; DDHQ figures from its Senate forecast page.
Iowa and Michigan are where the two come closest. In Iowa, both lean slightly Republican. Our Texas Senate race guide and Ohio Senate race guide cover the two contests with the widest gaps.
Why don't midterm prediction markets match the models?
There are a few well-known reasons, and none of them prove one side right.
- Speed. Markets move within minutes of news. Models change when new polls and data come in.
- Inputs. Decision Desk HQ says its race forecasts blend fundamentals (30%), polls (45%) and market prices (25%), then run a million simulations. Traders can weigh anything, including hunches and campaign news.
- Who takes part. A market reflects the people trading on it, not a random sample of the public. Officials quoted by Votebeat also worry that prices could be manipulated.
- Contract design. Kalshi's Senate price is assembled from two contracts, DefiRate notes, so the platforms are not always pricing exactly the same question.
History offers some support for markets. Economists Paul Rhode and Koleman Strumpf found that US election betting between 1868 and 1940 forecast results well, Votebeat noted. Even so, the 2026 markets are far larger and newer than anything in that record.
Why are election officials paying attention?
Votebeat reported in August that local officials worry about public trust if early returns differ from what markets expected. Los Angeles County's top election official said that happened in a June primary there, which led to suspicion about the count.
The money is real. An NBC News analysis cited by Votebeat counted nearly $200 million traded on midterm outcomes on Kalshi and Polymarket. Meanwhile, a Pew Research Center review found most states have some law against betting on elections, and Delaware County, Pennsylvania, has banned its election workers from using prediction markets.
For daily model numbers, see our midterm election forecast page, and our generic ballot explainer for the national polling picture.
How to watch election night live
Markets will swing all night as states report, so it helps to watch the real count. Our guide to watching midterm results live lists every option.
The C-SPAN live stream carries speeches without commentary, while the NewsNation live stream and Fox News will run full results coverage. Following from the UK or Europe? See how to watch the US midterms in the UK and Europe.
Coming soon: the NewsLive streaming app. We're building one subscription for live international news on Android, Roku, Fire TV and desktop. The service is under development. See what the NewsLive app will offer and join the early-access list.
Frequently asked questions
Are midterm prediction markets more accurate than polls?
There is no settled answer. Historical research found old election betting markets did well, but markets and polls have both missed results before. They measure different things, so many analysts read them side by side.
What does a 67% chance mean?
It means the outcome is expected about two times in three, not that it is certain. A 33% outcome happens often, so a favorite can still lose.
Who is favored to win the Senate?
Democrats are favored in both the markets and the main models, by margins from 58% to about 76% as of late October. Our Elections section tracks each race.
