Where we disagree most with the betting markets
The 13 races of 2026 where our final forecast and Polymarket or Kalshi are 10 points or more apart. In 5 of them we pick a different winner.
Disclosure:VotePredictor's owner holds positions on Kalshi in some of these races. This page compares forecasts; it is not a recommendation to bet.
The short version
- We pick a different winner from the markets in Georgia governor, Nevada governor, Kansas Senate, Alaska Senate, Iowa Senate.
- We pick the same winner but disagree on how likely in Florida governor, Texas governor, Vermont governor, Ohio governor, Mississippi Senate, South Carolina Senate, Florida Senate, Louisiana Senate.
- In all 13, the markets like the Democrat more than we do. So this is closer to one bet than 13: if the polls are underestimating Democrats this year, most of these go against us together.
Where we pick a different winner
Each cell names the candidate that source favors and their chance of winning. Market picks that differ from ours are in bold.
| Race | VotePredictor | Polymarket | Kalshi | Result |
|---|---|---|---|---|
| Georgia governor | Jackson 72% | Bottoms 71%now Bottoms 73% | Bottoms 69%now Bottoms 72% | Not called |
| Nevada governor | Lombardo 63% | Ford 76%now Ford 77% | Ford 74%now Ford 75% | Not called |
| Kansas Senate | Marshall 64% | Hamilton 63% | Hamilton 63%now Hamilton 64% | Not called |
| Alaska Senate | Sullivan 55% | not listed | Peltola 72%now Peltola 70% | Not called |
| Iowa Senate | Hinson 69% | Hinson 51%now Turek 55% | Turek 52%now Turek 53% | Not called |
Same winner, different confidence
Here everyone has the same favorite, and the disagreement is how sure to be. A market that has Donalds at 62% is saying the underdog wins the Florida governor race about one time in 3. We say about one in 32. It runs the other way in Ohio governor, where the markets are surer of the favorite than we are.
| Race | VotePredictor | Polymarket | Kalshi | Result |
|---|---|---|---|---|
| Florida governor | Donalds 97% | Donalds 62%now Donalds 63% | Donalds 63% | Not called |
| Texas governor | Abbott 89% | Abbott 62%now Abbott 63% | Abbott 61%now Abbott 63% | Not called |
| Vermont governor | Scott 96% | not listed | Scott 69%now Scott 68% | Not called |
| Ohio governor | Acton 65% | Acton 82%now Acton 84% | Acton 83% | Not called |
| Mississippi Senate | Hyde-Smith 99% | Hyde-Smith 87%now Hyde-Smith 86% | Hyde-Smith 86% | Not called |
| South Carolina Senate | Graham 80% | Graham 69%now Graham 70% | Graham 68% | Not called |
| Florida Senate | Moody 99% | Moody 88%now Moody 86% | not listed | Not called |
| Louisiana Senate | Letlow 98% | Letlow 88% | Letlow 96%now Letlow 97% | Not called |
Our numbers are the final call, frozen on November 2. Each market's number is its price saved at the same moment, and links to the market; where it has moved since, the price in our latest refresh (November 3) is beneath it. A market's price for a candidate is the crowd's chance they win: 69¢ is 69%.
Why the two disagree
Our model reads only the polls, each corrected for its pollster's record in past elections, and a fundamentals forecast built from the state's partisan lean, incumbency and the national environment. Markets can price things the model cannot see: early-vote turnout, a late scandal, candidate quality, money. They are also traders with their own leans, and a market with little volume moves on a few bets.
Georgia's governor race is one of the biggest gaps. Of the 11 polls in our final window, only two had Keisha Lance Bottoms ahead, and our pollster corrections moved a few of the Democratic-leaning ones toward Rick Jackson. The markets made her the favorite anyway.
Our record against the markets
On the 47 races of 2024 and 2025 that Polymarket priced, our walk-forward forecasts scored a Brier of 0.038 to its 0.059 (lower is better); on Kalshi's 22, 0.024 to 0.082. But races where we picked a different winner from the market were rare: 2 against Polymarket and 2 against Kalshi. That is far too few to say we beat the markets when we disagree, which is exactly the case this page is about. 2026 is the real test. See Polymarket and Kalshi for the full record.
This is a comparison of forecasts, not betting advice. Prediction markets can lose you money, and are not legal everywhere.