Prediction markets are poised to further complicate a confusing election-information system, leaving courts, state governments, and election officials scrambling to catch up.
More than $197 million has been wagered on markets predicting midterm-election results, according to an NBC News analysis from July. The Anti-Corruption Data Collective reported that betting on elections has surpassed $133 million and could balloon to $1.6 billion for the 2026 cycle if it mirrors the pace of 2024.
As questions about polling’s limits abound, some observers are looking toward prediction markets to forecast races. The growing popularity of the markets worries election workers who say the shifting markets could compound voter confusion.
“Prediction markets are a lot like forecasters, where they provide people with a level of certainty that doesn’t exist,” Democratic pollster Jeff Horwitt told National Journal. “When we’re in a situation where confidence in all institutions are at lows, I don’t think it’s particularly helpful for the industry.”
A smattering of recent prediction-market forecasts appeared not to align with the eventual results.
Abdul El-Sayed was the overwhelming favorite to clinch the Democratic nomination in the Michigan Senate race, peaking with 98 percent odds. In the end, he won by about 1 point, or roughly 15,000 votes. In Wisconsin, markets had Francesca Hong at 95.9 percent odds to win the Democratic primary for governor, but she lost to Milwaukee County Executive David Crowley by 3,600 votes.
Experts caution about conflating prediction markets and polls. Polling averages had shown El-Sayed favored by double digits, while odds for his victory reached the 90s.
“The poll tells us the vote share; the prediction market tells us how likely [someone] is to win,” said Koleman Strumpf, an economics professor at Wake Forest University. “They’re not exactly the same thing.”
“If you’re not really paying attention to what [a prediction market] really means, and then the result is the opposite, it does, understandably, cause you to doubt anything related to this, including polls that are out there,” Horwitt said.
The discrepancy between the market and the eventual election results in the Los Angeles mayoral race spurred unsubstantiated claims of election fraud. Kalshi told influencer partners to remove conspiratorial posts, and Polymarket pulled back on partnerships with some creators spreading falsehoods.
“The confidence and participation of voters in elections is being impacted by disruptive activity, misinformation, and just a lack of understanding of how the process works,” Los Angeles County Registrar Dean Logan said during a press call in July.
Logan said his communications team had to answer questions both about polls and prediction-market odds that had indicated a different result than the eventual primary results.
“I don’t know that we can draw a clear line specifically to the prediction markets, but I think we definitely had a situation here where early election returns differed from market expectations, and that resulted in suspicion and questioning of normal ballot processing and canvassing procedures in California,” Logan said.
Horwitt said prediction markets, fundamentally, are financially interested, while pollsters try to help their clients make informed decisions by giving them information about an electorate.
Markets and polls do not operate in a vacuum. Markets tend to shift based on news events, including polls showing one candidate or another leading.
“If [buyers] think that’s a good price, then they take the contract,” said Steve Ruddock, a gambling-industry analyst and consultant. “If enough people start taking the opposite side of a contract, the line will ebb and flow.”
Ruddock compared the process to what happens in sports betting, when “smart people create a line" and "sometimes it moves, sometimes it doesn’t.”
One key distinction of prediction markets is their anonymity. Prediction markets are subject to Commodity Futures Trading Commission regulations, and information on how much money is in a market and the value of individual contracts is public. But the identities of those buying contracts are not always public, and buyers don’t need to provide reasoning for their transactions.
Since July 2024, sports has made up 80 percent of total trading on Kalshi and 39 percent on Polymarket, according to a Pew Research study. Politics made up 4 percent of Kalshi’s volume and 32 percent of Polymarket’s volume.
Unlike sports gambling, prediction markets are not as tightly regulated. One major question regarding prediction markets is whether they fall under laws banning bets on elections. Thirty-two states ban betting on elections in one form or another.
“I think all of these statutes were written well before there were prediction markets, and so the question is how do you apply an old statute to a new situation,” Rick Hasen, a UCLA law professor, said on the July press call.
In Indiana, North Dakota, Pennsylvania, South Dakota, and West Virginia, people cannot serve as poll workers if they bet on the election they want to work on. In April, Delaware County, Pennsylvania, barred poll workers from using prediction markets to bet on election outcomes.
“If you found out at the end of the NCAA that the basketball referees were in prediction markets picking Michigan, you would lose a lot of faith in the fact that Michigan actually won that game legitimately,” James Allen, the election director for Delaware County, said during the press call. “Likewise, the same thing could happen with elections.”
Kalshi accused Wisconsin of “voter suppression” after the Wisconsin Elections Commission issued an advisory warning voters they cannot cast ballots in races they bet on. Wisconsin’s law barring people who have “become interested, directly or indirectly, in any bet or wager depending upon” election results dates back to 1849.
“The logic behind that is simply that people should not be choosing a candidate based on who’s going to get them the most money on a bet,” WEC Commissioner Ann Jacobs told National Journal. “They should be choosing a candidate because that candidate is the best person for the job.”
Kalshi has suspended a host of candidates who have bet on their own races, and Polymarket announced late last month it would investigate misconduct in bets related to the midterm elections.
The legality of election-related prediction markets is still largely undecided. In July, Maryland Elections Administrator Jared DeMarinis asked the state prosecutor to investigate their legality. While he has yet to hear back from the office, he told National Journal he plans to approach the state’s General Assembly if current law does not apply to the use of prediction markets in elections.
“It goes to the integrity and public trust of elections,” DeMarinis said.
In Minnesota, a federal judge temporarily blocked the state’s first-in-the-nation law banning prediction markets. Late last month, the U.S. Court of Appeals for the Ninth Circuit directed a lower court to decide whether election-related contracts are illegal under Nevada law. Last week, New Jersey asked the Supreme Court to review its challenge to prediction markets.
“Prediction markets are moving very quickly, but law moves very slowly,” Hasen said. “And so there are going to be a lot of questions that courts can’t answer in time for an election.”
