Early this week, election poll forecasting signals failed in the same race. Kalshi had Abdul El-Sayed at 98% to win Michigan's Democratic Senate primary on $8 million of volume, and Polymarket at 99%, per DeFi Rate on August 5.
Four different polls fielded in the final 10 days by Emerson College, Mitchell Research, Zenith Research, and Tavern Research put El-Sayed's lead between 11 and 19 points. He beat Haley Stevens by one point, 48.5% to 47.5%. The exchanges were more confident than the polls, and the polls were wrong by more.
We're tracking every election through the 2026 midterms right here as the political scene ramps up toward the 2028 Presidential Race. Here's the latest movement in polls and the live markets at Kalshi and Polymarket.
Democrats Favored in U.S. House Midterm Election Markets
On the general shape of the election, the two signals agree. Silver Bulletin's August 4 average has the generic ballot at Democrats +6.4, and Trump's net approval hit –20.6 on July 30, an all-time low across both terms. On August 6, Kalshi prices Democratic House control at 84% on $20.3 million of volume, and Polymarket at 87%.
Iowa is a clean race-level agreement. Emerson College, polling between August 2 and 4, 2026 among 712 likely voters at +/– 3.7 points, found Ashley Hinson ahead of Josh Turek 48% to 45% for the Senate and Rob Sand ahead of Zach Lahn 48% to 43% for governor. Kalshi has Hinson at 61% as of August 6, which is what a three-point August lead converts to once polling error and three months of campaign are priced in. Both signals also favor Sand, with Kalshi at 63%.
Where Polls and Markets Disagree
Michigan is the widest gap, which opened the moment the primary closed. EPIC-MRA, polling July 24 to 31 among 600 likely voters, had Republican Mike Rogers ahead of El-Sayed 46% to 43%. Mitchell Research, July 28 to 29, had them tied at 45%. Glengariff Group, July 22 to 24, had Rogers ahead by 10. Not one of the three has the Democrats favored. Kalshi's party contract has them at 58%, and no poll has been conducted since El-Sayed became the nominee, so the exchange is pricing an unpolled matchup.
Texas is the same disagreement closing, with the market moving toward the polls rather than away. Kalshi had Ken Paxton at 59.2% on August 2 and at 51% on August 6, about eight points of movement in four days on $7.4 million of volume. Every poll fielded since mid-July has James Talarico ahead, including Texas Southern University's Barbara Jordan Center, which polled July 27 to 30 among 1,200 likely voters and had Talarico up 47% to 45%. The two signals still disagree on who is ahead, just by less than they did a few days ago.
Wisconsin is the reverse, a repricing with no polling behind it. Mandela Barnes withdrew from the Democratic gubernatorial primary on July 30, 2026, leaving Francesca Hong the favorite for the August 11 nomination. Kalshi had Republicans at 46% on August 1, 53% on August 3, and 60% on August 6, a 14-point move in five days, with Polymarket at 58%. No general-election poll has been fielded since, and the most recent, RMG Research, fielded July 20 to 24 among 800 registered voters, had Hong ahead of Tom Tiffany, 44% to 43%.
What Our Models Predict
The statistical forecasters do not settle it, because they disagree with each other more than either disagrees with the exchanges. Decision Desk HQ, as reported by The Hill on July 27, had Democrats at 60% for the House and Republicans at 60% for the Senate. G. Elliott Morris's FiftyPlusOne, updated August 3, had Democrats at 85% for the House and 55% for the Senate. That is a 15-point spread on Senate control, with Kalshi splitting the difference at Democrats 45%, five points above Decision Desk HQ's implied figure and 10 below FiftyPlusOne.
None of these discrepancies are new. Joshua Clinton and TzuFeng Huang of Vanderbilt University, examining more than 2,500 political markets and $2.4 billion in transactions in the final five weeks of 2024, found that PredictIt beat chance in 93% of its markets, compared with 78% on Kalshi and 67% on Polymarket.
Rajiv Sethi of Barnard College found Polymarket held up on the 2024 presidential race, but fared poorly down-ballot, especially in congressional races, which he attributes to thinner volume. Depth within a venue is the variable, and the Michigan primary is what a thin book looks like when it is certain.
What to Watch Next
The Wisconsin primary on Tuesday, August 11, 2026, where Kalshi prices Democrat Francesca Hong at 90%, is the largest scheduled repricing on the calendar. Next is the first Michigan poll of El-Sayed against Rogers, which would give the exchange something current to argue with.
