Prediction Markets Boast 98% Accuracy in Midterms as Experts Advise Caution

Polymarket stated that 98.4 percent of the resolved midterm races went to the favorite on its platform the day before the election, and CBC News reports that Kalshi made a similar claim, stating its favored candidates won or led in 98 percent of races before polls closed on Election Day. Industry observers note that while these figures are not inherently incorrect, they require important context. Valerie Cross, an analyst who monitored the platforms on election night, explained that these statistics demonstrate how the markets effectively incorporated “early returns, exit polls and county counts quickly, and mostly moved in the right direction”. However, Cross cautioned that “it’s easy to overread” these high accuracy claims.

Despite the self-reported milestones from the platforms, some media outlets and forecasting experts are pushing back against the narrative that prediction markets have entirely replaced traditional polling models.

Reporting on the phenomenon, Quartz noted that “an analysis by The New York Times found that prediction markets fared about the same as election forecasts through most of the midterm campaign”. While the markets gained an edge in the final days as static polls became outdated, Quartz highlighted that “outside experts urged caution” when interpreting the companies’ sweeping victory laps.

Furthermore, traditional forecasters have fiercely defended their own methodologies. According to Quartz, Nate Silver stated that “his model and others showed the Democratic wave sooner and more aggressively” than the prediction markets did.

The debate highlights the evolving and complex nature of political forecasting. Prediction markets rely on financial stakes to drive price discovery, with traders risking real capital on the outcomes. This dynamic theoretically forces quicker adjustments to breaking news, such as late-breaking scandals or early voting returns, than traditional public opinion polls can manage.

Yet, evaluating the platforms purely on their final accuracy claims might not tell the whole story of the 2026 cycle. As Quartz reported, experts pointed out that these accuracy figures “include only resolved markets, which may be easier to predict, and that many races were fairly certain going into Election Day”. Furthermore, an economics professor cited by Quartz gave the markets a “B,” noting that they “pointed the right way but underestimated how large the Democratic wave would be”.

As the political landscape continues to shift, the tension between traditional forecasters and market-based platforms will likely intensify. While the markets were undoubtedly fast to react on election night, the broader consensus, as summarized by Quartz, remains that these platforms “did no better than election forecasts for most of the race”.

Sources Used:

• CBC News: https://www.cbc.ca/news/business/prediction-markets-midterm-elections-9.7411622
• Quartz: https://qz.com/kalshi-polymarket-98-accuracy-midterm-races-110626