Prediction markets are facing a high-profile test in the U.S. midterm elections, as platforms such as Kalshi and Polymarket seek to show that their forecasts can anticipate results. The companies have expanded their marketing and attracted more betting activity, while researchers and election analysts debate whether the markets offer a reliable signal beyond polls and projection models.
The platforms let users bet on the outcome of events. As wagers shift, the prices assigned to possible outcomes change, producing probabilities that reflect traders’ expectations. Elections have become a popular category, with campaigns attracting bets from both casual users and professionals.
Kalshi rented a large billboard in New York’s Times Square to advertise odds on which party will control the Senate. Polymarket released a television commercial promoting its midterm coverage. The companies are seeking new users and revenue from fees on bets as they try to establish a stronger profile.
More than $221 million has been traded in markets tied to this year’s midterms, compared with $92 million around the 2024 presidential contest, according to a report by the Anti-Corruption Data Collective, a Washington-based research group. The growth has also intensified scrutiny from lawmakers, including concerns about market manipulation and the use of inside information.
Evidence on predictive accuracy is mixed. A study of more than 1 million markets, published this year in Proceedings of the National Academy of Sciences, found that their probabilities were generally well calibrated: outcomes with a strong favorite usually occurred, but not always. Researchers caution that earlier findings may not fully describe today’s platforms, which have more users and more money flowing through them.
A Vanderbilt University study found that prediction markets signaled well ahead of polls that Donald Trump would defeat Kamala Harris in the 2024 presidential election. Polls, by contrast, showed Harris with a small but consistent lead, the researchers concluded. Brett Goldstein, a Vanderbilt research professor who worked on the study, said the markets could offer information closer to the truth, but stressed the need to watch them carefully.
The platforms have also made notable mistakes. During this year’s primaries, prediction markets gave Spencer Pratt, a conservative candidate for mayor of Los Angeles, a 75% chance of reaching the Nov. 3 general-election ballot. He finished third, nearly 30,000 votes behind the second-place candidate, and did not advance. Polls released before the primary were divided over whether Pratt or Nithya Raman would finish second; Raman took the spot.
Some pollsters and election forecasters argue that prediction markets mainly reflect the views of online bettors and often rely on existing polling. G. Elliott Morris, an election analyst who leads the polling and forecasting site FiftyPlusOne, said he sees little reason to follow the markets instead of direct measures of public opinion.
The disagreement is visible in Iowa’s Senate race. Polls show a close contest, while Morris and other modelers give Democrat Josh Turek an edge. Prediction markets, including Kalshi, Polymarket and PredictIt, favor Republican Ashley Hinson. In most other closely watched Senate races, the markets favor Democrats.
In Maine, Republican Senator Susan Collins, in office since 1997, is behind Democrat Troy Jackson in prediction markets. She also trailed in all but one high-quality poll conducted in October; that poll showed a tie. Collins faced a similar polling picture in the final four months of her 2020 reelection campaign, but ultimately defeated her Democratic opponent by more than eight percentage points, in a year when polls broadly underestimated Republican support.
The results of the midterms will give researchers a chance to assess how well the current generation of prediction markets performs as their reach and trading volumes grow. Their odds can shift rapidly as users place bets, but past performance shows that a market favorite is not guaranteed to win.
Sources: Folha Mercado.