U.S. midterm election betting volume surges; prediction markets wager on Democrats sweeping both chambers

Prediction markets on Tuesday showed growing expectations of a Democratic sweep of the U.S. Congress, with more than $1.48 billion wagered on the midterm elections as this new class of markets faces a fresh test. As polling stations began closing across the country, betting prices shifted further in favor of the Democrats. At about 19:00 ET on Tuesday (08:00 Beijing time on the 4th), prices aggregated by analytics firm BetInt showed a 98% probability that the Democrats would take the House, up 4 percentage points from intraday levels that afternoon. Markets were volatile, with the implied probability of Democratic control of the Senate rising as high as 88%, up 14 percentage points from earlier in the day, before falling back to around 80%. BetInt data showed that when the first polling stations closed at 18:00 ET (07:00 Beijing time on the 4th), about $2 million flowed into bets that the Democrats would win control of both chambers and prevail in competitive Senate races in Ohio, Texas and Maine as well as in some gubernatorial contests. Prediction markets rose to prominence in 2024, when they defied the direction of traditional polling and signaled that Trump would win the presidential election, building a large following among global media and investors. Such markets advertise their odds as highly accurate, but several high-profile misses in U.S. primaries and Brazil's presidential election show they are not infallible, and the Democratic-sweep conclusion implied by markets on Tuesday will be closely scrutinized. Analysts say the industry's reputation could suffer if midterm results diverge sharply from prediction-market odds, especially given record wagering. The largest platforms, Kalshi and Polymarket, also face pressure to guard against misconduct: several candidates had previously bet on elections in which they were running, and research by the Washington nonprofit Anti-Corruption Data Collective (ACDC) found that relatively small wagers can move political markets. "Tuesday really is a real test of this new generation of prediction markets," said Yesha Yadav, vice dean of Vanderbilt Law School, adding that the large number of congressional races this year would help fill out the industry's overall report card. Polymarket and Kalshi said they have strengthened controls and are well equipped to prevent misconduct. Kalshi bars politicians and campaign staff from participating in political betting. Both companies also said their markets self-correct quickly when large wagers move prices. Sean Patrick Maloney, chief executive of the Coalition for Prediction Markets, said the markets do not claim to be perfect but offer an unbiased way to forecast events "in a world full of chaotic and biased news sources." Research says such markets are well calibrated in scale Kalshi reported on Monday that total trading volume across all midterm bets reached $1.33 billion; according to ACDC data, betting on Polymarket's congressional markets totaled about $145 million. The nonprofit said more than 28,000 midterm-related markets have been opened to date, with a small subset of racesincluding Senate contests in Texas, Michigan, Maine and South Carolinaaccounting for most of the volume. For example, according to ACDC data, the Texas Senate race had attracted nearly $67 million in bets as of October 29. BetInt data showed that on Tuesday evening, the market implied a 75% probability that Democratic candidate James Talarico would defeat the state's Republican attorney general, Ken Paxton, up 3 percentage points from earlier in the day. In the Michigan Senate race, which has attracted more than $26 million in bets, BetInt data showed the market implied an 83% probability that Democrat Abdul El-Sayed would defeat Republican Mike Rogers, up 7 percentage points from earlier in the day. Academic research, including a 2025 Vanderbilt paper that analyzed nearly 2,500 markets in the 2024 election, has generally found that prediction markets are well calibrated. Calibration measures whether odds are reliable, while accuracy measures whether the market picked the winner. Joshua Clinton, a Vanderbilt political science professor who led the 2025 study, said that because most elections are not competitive, broad performance metrics may not reflect how markets perform in close races. Harry Crane, a Rutgers statistics professor, said a small group of sophisticated, profit-driven traders appears to account for a large share of prediction-market money and helps correct prices. Even the most sophisticated traders get it wrong, Clinton said. For example, prices on Kalshi and Polymarket at one point gave only a 20% probability that Brazil's President Luiz Inacio Lula da Silva would defeat his right-wing rival Flavio Bolsonaro, and Lula did win last month. Clinton said this may be because sophisticated traders overcorrected for the risk that "polls often understate right-wing candidates' appeal." Spokespeople for both companies said events with a 20% probability happen all the time, and political election upsets are no exception. Carlos Hardenberg, a portfolio manager at MCP Emerging Markets, said the lesson for investors is that these markets are useful reference indicators. "But they should not be confused with certainty."