On Election Night, millions of Americans will watch anxiously as the ballot counts stream in. Most will be worried about the political future of their country. Some will also have money on the line.

Over the past several months, election betting has gone mainstream. On Polymarket, perhaps the most popular political-betting site, people have wagered more than $200 million on the outcome of the U.S. presidential election. The election forecaster Nate Silver recently joined the company as an adviser, and its election odds have been cited by media outlets including CNN, Bloomberg, and The New York Times. Polymarket is officially off-limits to U.S. users, but the website is still accessible using technical work-arounds. Americans can directly place bets on other platforms such as PredictIt and Kalshi, the latter of which was recently approved to offer legal election betting. Just this week, the investing app Robinhood launched its own presidential-election market.

In a sense, election betting is like sports betting: Think Donald Trump will win next week? Put money down on it, and profit if you’re right. But these sites present themselves as more than just a way to make a quick buck. They assert that how people bet, whether on the benign (who will be the next James Bond?) or the consequential (will Israel and Hamas reach a cease-fire before the end of the year?), can help forecast the future. Because there’s money involved, the thinking goes, these prediction markets leverage the collective wisdom of what people actually think will occur, not what they hope will. For example, this summer, prediction markets accurately forecast President Joe Biden’s withdrawal from the race. If they are right about the election, Donald Trump has the edge: On Polymarket, for instance, Trump currently has roughly a 65 percent chance of winning the election.

But what will happen if the outcome is contested? Many Trump loyalists are already preparing for the next “Stop the Steal” campaign rooted in unfounded claims of a rigged election. A disputed election could plunge these betting sites into chaos. Prediction markets sometimes describe themselves as “truth machines.” But that’s a challenging role to assume when Americans can’t agree on what the basic truth even is.

Prediction markets have become popular among Trump supporters—no doubt because they show that Trump is favored to win even as the polls remain deadlocked. If Trump loses, election denialists may look to the betting markets as part of their evidence that the race was stolen. The groundwork is already being laid. “More accurate than polls,” Elon Musk recently tweeted to his more than 200 million followers on X, alongside an image displaying Trump’s favorable Polymarket odds. “You shouldn’t believe the polls,” J. D. Vance has agreed. “I think that chart’s about right,” he said in reference to Kalshi’s presidential odds. Even Trump himself has talked up his betting odds, both online and in real life. “I don’t know what the hell it means, but it means that we’re doing pretty well,” he recently said of Polymarket, during a speech in Michigan. Indeed, if you follow only betting markets, a Trump loss might even be surprising, potentially fueling claims of foul play.

Prediction markets have already received significant attention in the lead-up to the election, but this might be only the start. Strange activity could occur on these betting sites after the polls close. That’s because most of these markets will remain open for bets for weeks and months after the election, in some cases as late as Inauguration Day. A significant amount of money will likely be wagered after votes have been cast, and the market odds could diverge from election results.

That’s what happened during the previous presidential election. In 2020, even after an audit had confirmed Biden’s win in Georgia and his victory was certified, PredictIt still gave Trump a nontrivial chance of winning the state, at one point reaching as high as 17 percent. Putting money on a Trump win after he officially lost wouldn’t make much sense—unless, that is, you genuinely believed that the election was stolen or that Trump would be successful in an extralegal attempt to overturn results. This time around, with more money on the line and election denialism already in the air, a contested election could result in even more anomalous election odds after the polls close. In other words, betting markets can’t be disentangled from a reality in which a segment of the country does not believe the election results.

Especially on Polymarket, such a scenario could get weird fast. Polymarket runs on the blockchain—bets are made with cryptocurrency, and official decisions about who wins are made by the holders of a crypto token called UMA. If there is a disagreement over what occurred, UMA token-holders can vote to determine the official outcome. These are not lawyers scrupulously analyzing predefined rules, but people considering evidence posted to a Discord server. Although token-holders have strong incentives to vote honestly, the system is still vulnerable to manipulation. And in a highly contentious election, things could get messy.

Consider how the Venezuelan presidential election this summer played out on Polymarket. According to Polymarket’s rules, the winner was to be determined based primarily on “official information from Venezuela.” Given that the authoritarian incumbent Nicolás Maduro controlled the election, bettors initially favored him by a sizable margin—in part, because it seemed likely that he would stay in power, regardless of how Venezuelans voted. That’s what happened. Although the opposition candidate, Edmundo González, got more votes, Maduro stole the election. But the UMA arbiters declared González the winner, overriding Polymarket’s original rules. Some bettors defended the decision: Rubber-stamping Maduro’s fraudulent win, they argued, would be “very bad, even dystopian.” Others felt they had been scammed. “What happens next, if Trump doesnt recognize the election results,” wrote one user in the Polymarket comments section.

Venezuela is a unique case. Trump cannot steal the election like Maduro did—he’s not even currently in office. Still, UMA decision makers could go against official sources if the results are disputed. Even in the case of a contested election, such an outcome would be unlikely because it would be a massive blow to Polymarket’s credibility, Frank Muci, a policy fellow at the London School of Economics, told me. However, he added, “if there are Supreme Court rulings and dissenting opinions and Trump is saying that the election was really stolen, [then] politics may override the narrow bottom line.” Polymarket, which did not respond to multiple requests for comment, could always intervene and overrule UMA’s results. It didn’t do so after the Venezuela debacle, but earlier this year Polymarket refunded some users after UMA got a resolution wrong.

Other election-betting sites have more precautions in the case of a contested election. Both Kalshi and PredictIt determine market outcomes in-house. Xavier Sottile, the head of markets at Kalshi, said in an email that if Kalshi’s users have a credible reason to dispute who is declared the winner on the platform, the company has “an independent market outcome review committee” that includes “election-focused academics” to verify the resolution. But if people disagree on who won the election, some percentage of bettors are destined to be deeply unhappy, no matter how fairly these markets are resolved.

After the election, betting sites may look less like oracles than mirrors, reflecting the nation’s disunity back at us. In 2020, Trump’s outsize odds on prediction markets following Biden’s win led Nate Silver to write that that markets were “detached from reality.” So too is our country. Many Republicans falsely believe that Trump won the last election, a lie that Vance has repeated of late. In a way, prediction markets act as a microcosm of America’s political psyche, distilling the confusion of our political moment into tidy charts.