Prediction markets called the Democratic sweep, but missed consequential races

Prediction markets correctly anticipated that Democrats would take both chambers of Congress in the 2026 midterms, and their prices generally identified winners in the largest contests. But several results exposed the limits of markets whose odds can be shaped by sparse trading, concentrated wagers and late shifts in information.

Kalshi traders had been predicting Democratic Senate control since April, when they put the chance of a Democratic “blue tsunami” at 55%. As reported in our Nov. 3 coverage, market odds moved close to certainty on a Democratic congressional sweep after polls began closing.

The platforms also called several prominent Democratic wins. Kalshi and Polymarket favored James Talarico in Texas, Troy Jackson in Maine, Abdul El-Sayed in Michigan and Sherrod Brown in Ohio, all of whom won. In Ohio’s governor’s race, Kalshi put Amy Acton’s chance of beating Vivek Ramaswamy at 82%, while Polymarket put it at 83%.

Fox Television Stations said the markets identified the winner in nearly every major contest covered by their final forecasts, while polls generally captured the direction of races but underestimated the size of several Democratic victories. Georgia’s gubernatorial contest was the clearest polling miss: late polls had Rick Jackson ahead by one to three points, or tied, but Keisha Lance Bottoms won by about eight points to become the state’s first Black woman governor.

The market record was not clean. Semafor said Kalshi and Polymarket missed Republican Gov. Greg Abbott’s loss in Texas, as well as Democratic Senate victories in Iowa and South Carolina. Annie Andrews defeated Darline Graham in South Carolina despite markets giving Graham roughly a 70% chance, while Democrats also won House races in Indiana, Ohio and Missouri where their odds had been only 25% to 40%.

Accounts of the Texas governor’s race differ on the final pricing. KFGO’s report described odds favoring Abbott, whereas Fox reported that Kalshi assigned Gina Hinojosa a 62% chance of defeating him before Abbott conceded. In Alaska, markets favored Mary Peltola in the Senate race, but the outcome remained unresolved under the state’s ranked-choice system.

The platforms’ mechanics help explain both their appeal and their vulnerability. Users buy and sell event-linked contracts, so prices express collective expectations and can react quickly to new returns. More than $4 million was wagered in the minutes after the first polls closed, mostly on Democrats; BETINT identified a $951,198 Polymarket wager on a Democratic House win as the largest such late bet.

Low-liquidity contests remain a particular concern. House-race markets often draw less trading and can therefore be less reliable. The Anti-Corruption Data Collective found that 88% of 18,000 active congressional markets could be moved by 10 cents with a single wager under $1,000, while about 23% of 167 markets holding more than $100,000 recorded price swings of at least 30 cents.

The volume figures nevertheless show how far election betting has expanded. Semafor estimated that trading on down-ballot elections reached $1.8 billion, nearly five times the 2024 level, although sports and cryptocurrency contracts remained larger on the major platforms. Separately, KFGO reported that Kalshi’s election-market volume exceeded $2.5 billion in the 2026 cycle, up from $1.3 billion in 2024. Quartz put total wagers on the midterms across prediction markets at more than $1.48 billion.

The companies’ growing visibility has extended beyond trading. CNN, CBS News, Dow Jones and CNBC formed partnerships with Kalshi or Polymarket to incorporate market information into election coverage, even as critics questioned the use of gambling-style odds in political reporting. Kalshi and Polymarket have added surveillance and transparency tools aimed at detecting unusual trading, and their US operations are governed by Commodity Futures Trading Commission rules addressing manipulation and insider trading.

Kalshi said candidates favored by its markets have won roughly nine times out of 10 in US elections since 2024. A Washington Post review cited by CNBC offered a more limited measure: three-quarters of 268 candidates assigned a 70% to 80% winning probability by Kalshi or Polymarket during the two months before this year’s primaries ultimately won.