The US presidential election in 2024 served as a big-bang moment for prediction markets. Two years later, their forecasts on the 2026 midterms are getting plenty of attention, but the underlying trading illustrates how the platforms have struggled to expand beyond a few booming categories.

In the final week before Election Day, turnover on US election-related markets on Kalshi Inc., the biggest prediction market, was roughly the same as it was in the comparable period before the 2024 vote, a Bloomberg News analysis of trading data shows. On Polymarket, Kalshi’s main rival, US election trading has shrunk significantly when adding up the activity on both its longstanding international exchange and its newer US exchange.

That lack of growth is notable compared to the overall volume on the platforms, which has multiplied by a factor of more than 20 since October 2024. Almost all of that expansion stemmed from trading on sports and crypto prices, while many smaller categories, including elections, have attracted relatively small and sometimes declining levels of activity.

Last month, politics accounted for just 1% of trading volume across Kalshi and Polymarket, down from two-thirds during the 2024 presidential election.

Much of Kalshi’s recent growth has also come from so-called combination bets, or parlays: high-risk, multi-leg trades that rely on a series of different outcomes across multiple events.

Limited trading interest in the midterms represents a problem for prediction markets. Advocates for the industry often point to election markets — and the forecasts they generate — as evidence that the platforms are offering socially useful information, rather than just another venue for betting. Exchanges are under pressure to demonstrate the broader public value of their platforms, as they battle dozens of states that have accused them of trying to circumvent state gambling laws.

“They need to prove to their investors that they have another business,” said Koleman Strumpf, a professor of economics at Wake Forest University in Winston-Salem, North Carolina. “They also need to prove to all these other people – regulators, judges – that the more socially valuable part of prediction markets is important.”The struggle to meaningfully expand election trading also leaves many of the forecasts displayed by the sites more vulnerable to manipulation, given the small sums it can take to change the prices — and thus the forecasts — coming from low-volume markets.

Almost all of the midterms-related markets offered on Polymarket and Kalshi have such low liquidity that it would take only a few thousand dollars to move their forecasts meaningfully, according to an analysis by the Anti-Corruption Data Collective published Monday. The collective is a non-profit organization of academics and researchers who investigate topics related to corruption and economic injustice.

“When you introduce tens of thousands of markets, but not all of them have national prominence or are able to attract betting volume, that opens up a lot of vulnerabilities where a savvy, potentially malicious actor could go in and bet a small amount of money and skew the odds in one way or another,” said David Szakonyi, a co-founder of the collective and a political science professor at George Washington University.

Kalshi spokesperson Jack Such called the collective’s findings “baseless,” stating the actual cost of such manipulation would be much higher and that its markets remain well-calibrated against such attempts. Major events like elections and sports also typically attract the majority of their volume very close to the event’s resolution, he said. Polymarket didn’t respond to a request for comment.

Both companies have said they have systems in place to monitor for suspicious trading. This month, Polymarket launched an investigation into trading on the Israeli election that bore some of the hallmarks of insider trading. Kalshi has previously banned politicians from its exchange who had bet on their own races.

There have been a number of previous incidents in which a candidate’s odds shifted suddenly due to a small amount of trading, creating concerns about the possibility that someone was trying to improperly influence the public perception of the race.

The 2024 surge in election trading came after a federal court decision, just a month before that vote, allowing Kalshi to list markets tied to US elections for the first time. Kalshi and Polymarket both made a late advertising blitz to consumers and quickly rose to prominence after trading on the platforms suggested that Donald Trump had higher odds of winning than many polls suggested.

Part of the decline in demand for political trading can be attributed to the lack of an immediate presidential vote to bet on this year, which in 2024 attracted nearly all election-related trading volume.

Trading activity during the current election cycle has been largely focused not on specific races, but instead on the broad question of which party will control the House and the Senate after the election. Market forecasts have broadly reflected expectations that Democrats will take both the House and the Senate, in line with most polling.

The race with the most turnover a week before the midterms was the Texas Senate, with about $79 million in volume on Kalshi as of Monday. That’s about a quarter of the $277 million Kalshi had recorded on its balance-of-power and congressional-control markets.

Beyond a handful of high-profile Senate races, trading activity falls off quickly. Only five House races attracted more than $1 million of trading volume on Kalshi as of Monday, the day before Election Day.

While the trading volumes on many races are relatively meager, the public focus on the odds on Kalshi and Polymarket has grown significantly. Both companies have struck deals with media outlets to include the forecasts in news coverage.

In addition to the worries around potential manipulation, prediction-market experts have expressed concern that many ordinary Americans seeing coverage of the forecasts may not understand how the forecasts differ from the underlying polls.

Trading in 2026 may also reflect a changed demographic among prediction-market users, compared to two years ago.

Polymarket was the largest venue during the 2024 election while being off-limits to American traders – though some still accessed the site using virtual private networks. That status quo has since flipped, with US-regulated Kalshi processing tens of billions of dollars in weekly notional volume more than Polymarket’s US and international venues, according to Bloomberg’s analysis of trading data.

“Now it’s a completely different picture,” said Kelly Ye, head of research at Talos.