It’s been a huge month for all the degenerate gamblers hoping to cash in on the United States’ elections. In early September, the New York–based prediction market Kalshi notched a temporary victory in its ongoing lawsuit against the Commodity Futures Trading Commission, which, one year ago, had forbidden the platform from offering American users the opportunity to bet on which political party would win majority control of Congress.
U.S. District Judge for the District of Columbia Jia Cobb sided with Kalshi in its case against the regulatory agency, sparking excitement among fellow traders that political betting could soon be a reality across the country. But the CFTC quickly filed an emergency motion seeking to block those markets while it appealed the decision, a situation that Kalshi claimed would be “devastating” for the company’s business, in part by driving would-be customers to rival apps like the Nate Silver–approved crypto platform Polymarket. (It’s worth noting that Kalshi offers plenty of other opportunities to bet on newsy events, like whether the Federal Reserve will cut rates and whether a particular movie will earn a positive Rotten Tomatoes score.)
Nevertheless, the U.S. Court of Appeals for the D.C. Circuit ruled last Thursday that Kalshi had to pause its election-betting operations as the appeal goes to trial—just hours after Kalshi had set up two separate markets for gamblers to bet, respectively, on House and Senate control. Interactive Brokers, the largest digital trading platform in the country, also agreed to set aside its plans to offer bets on the presidential elections and on certain swing-state races, for now.
Just like that, the decision to allow Americans to turn the electoral process into a gambling den was granted and then retracted, with the prospect of far more litigation to come. Still, even if the CFTC hadn’t appealed and won the stay, it’s not exactly the case that election betting would have been fully permitted everywhere. The initial ruling in Kalshiex LLC v. CFTC, for one, was more a procedural decision than a sweeping legal shift.
The decision is less a judgment on the operations of Kalshi and similar prediction markets and more a gauge of the CFTC’s regulatory prowess (which has already been weakened by recent Supreme Court decisions that gutted federal agencies’ oversight powers) when it comes to said platforms. “This case is not about whether the Court likes Kalshi’s product or thinks trading it is a good idea,” Cobb wrote. “The Court’s only task is to determine what Congress did, not what it could do or should do. And Congress did not authorize the CFTC to conduct the public interest review it conducted here.”
There is also the matter of local governance. At least 23 states have explicit laws against and punishments for elections betting, with only limited exceptions for CFTC-approved platforms. Then, there’s the matter of whether betting markets do actually serve the public interest—and whether there’s any appetite for them among the electorate.
For-profit prediction markets are driven by people attempting to impose simple resolutions to complex real-world events, on behalf of people who are trying to extract as much money from the system as possible. To fully open up elections betting by supposing that the wisdom of the crowds will keep things thoughtful and steady is a dangerous bet all its own—and we already have examples that demonstrate this.
Political-gambling operations once ran rampant across the U.S., from the post–Civil War period up through the Great Depression. While these markets often held great predictive value, they were soon usurped by the rise of scientific public polling (and also discouraged by the moral standards of the era).
They found new life by the late 1980s, thanks to shifting political winds and the establishment of a platform that still runs to this day: the Iowa Electronic Markets, a digital betting project from the University of Iowa’s Tippie College of Business. IEM has run trades on everything from where an IPO price will settle to when a hurricane will first hit land—but its ability to run contracts for political party nominations and the ideological makeup of Congress is unique, having carried CFTC approval since 1988. It’s under exacting conditions, however: IEM does not take revenue from users, its primary application is for academic research, it doesn’t advertise its services, and it can only allow up to 2,000 people to bet on a select few markets at any time.
It’s a stark contrast with how private markets like Kalshi and Polymarket function, as they seek to maximize the number of users and dollar amounts in their systems. But that’s what gives IEM its utility, program director and Tippie professor Thomas S. Gruca told me. “We’re interested in high levels of accuracy. We’re not interested in having the most traders,” Gruca said. “All the money that goes in comes out to the traders. We know that if you put any type of transaction fee for revenue, the markets don’t really work that well.”
Gruca speaks from the experience of having worked with IEM since the early 1990s, shortly after it was established to fill what appeared to be holes in electoral and event forecasting. “The person who actually spurred this was Jesse Jackson. In 1988, he did super well in the Michigan primary and no polls were even asking about him,” Gruca explained. Yet despite being the impetus, politics remain a limited part of IEM—with the project offering contracts with who’s going to control the House, the Senate, the party nominees, and “combinations thereof.”
“It becomes more complex when you try to figure out when someone is elected president,” Gruca said. “Do you call that on Election Day? No, because we have to count all the votes, then elect the electors, then send all the electors’ information to Washington, then have that approved, and then inauguration. If these are people in their 80s, there’s some probability that they’re not literally going to live until January. That’s morbid, but if you put in actual names, then how do you pay those off?” For that reason, IEM only runs markets on which presidential candidate will win the popular vote, specifically.
Even with all these precautions, Gruca noted, there’s never any guarantee that irrational actors won’t come in and skew things wildly, reducing the predictive accuracy. He pointed to past examples of traders staking tons of bets in the belief that Ross Perot would win the presidency in 1996, or that Ron DeSantis would trounce Donald Trump in 2024 Republican primary—both of which, obviously, did not occur, making it so the proceeds all went to the handful of people who’d correctly wagered on Bill Clinton and Trump, respectively.
And this cautious approach isn’t reflected in other betting markets. Polymarket, which can’t legally operate stateside (but has plenty of American users via VPNs), is most famous for its presidential election contract, on which nearly $950 million is at stake. While its founders tout it as “the future of news,” with Polymarket charts having been integrated into reports from Bloomberg Terminal and CNN, there are traders raising doubts about how predictive it really can be, especially for a high-stakes electoral showdown.
One crypto tracker on X who goes by @Dumpster_DAO pointed out on Sept. 6 that a bunch of accounts on Polymarket banded together to manipulate the odds presented in the real-time president-election contract … because they had set up another market where they bet on who was leading the Polymarket forecast by a particular day. A month before that, British economist Frank Muci wrote a Substack post on how ill-equipped Polymarket’s system was to handle the highly contested Venezuelan election, when incumbent dictator Nicolas Maduro disputed public voting tallies in order to claim victory for himself.
The crypto-based resolution program known as UMA, which employs specific stakeholders to help settle and decide disputed outcomes, followed the Venezuelan government line to reward bettors who’d predicted Maduro would “win” reelection, but was overwhelmed and persuaded by other traders (many of them Venezuelan) to have it declare his opponent the winner, even though he was blocked from taking power. No matter which way it would have spun, a lot of people would have lost money and decried the unfair nature of UMA’s decisionmaking here. (Lest you believe it’s just an issue with foreign elections: PredictIt infamously kept pricing in chances of Trump winning the 2020 election long after Biden’s victory was confirmed.)
A gambler and Polymarket trader who is based in Kronstadt, Russia, and prefers to stay anonymous reached out to me to share some odd occurrences on the Polymarket contracts for smaller U.S. political disputes. One particularly wild moment: a furious debate over whether Kamala Harris said the word joy in her Democratic nomination acceptance speech, a prospect on which Polymarketers had bet nearly $3 million. While the vice president said the word joyful, the UMA governors claimed that that is not a compound word with joy, so everyone who’d bet she’d say joy lost their money. The rules and terms for this particular ticket adjust quite a bit as the outcome was disputed twice, spurring one commenter to lament that “the additional context was provided after the speech” and was “probably just entered by a single low wage person without any professional expertise in english.”
The trader who’d flagged this incident for me had come to Polymarket by way of sports betting—as an MMA fan and bettor who “sought more liquidity” and enjoyed Polymarket’s easy crypto-based exchanges. But incidents like that, coupled with the large and rapid swings that can suddenly shift the “probabilities” calculated by once bearish markets (e.g., whether Robert F. Kennedy Jr. would endorse Trump in August) made him much more nervous.
“The people deciding it are basically on Discord. There’s no legalese in the wording,” he said incredulously. “They’re using ChatGPT to decide whether words are compound. They’re not proper adjudicators.”
Polymarket’s boosters claim that its high-volume trades and globe-spanning customers could make it even more accurate than the media and better at predicting outcomes than the polls—qualities that they’d say would also apply to Kalshi if it got the chance to legally administer elections-betting contracts within the U.S. But Gruco has his doubts.
“Having more people isn’t good universally. We don’t want a whole bunch of people that don’t know what they’re doing,” he said. “We want people that are informed, because informed people make for better forecasts. If you’re smart and you can understand things like arbitrage opportunities, you can make money when people make pricing mistakes. That’s a lot more sophisticated than saying, ‘Gimme $20 on Kamala Harris.’ ”
There’s plenty of recent history to show how both prediction market creators and government regulators get cold feet when these platforms are given free rein and try to operate at scale. The Irish website InTrade gained prominence for correctly forecasting Barack Obama’s reelection victory in 2012, but soon shut down thanks to scrutiny that brought to light plenty of price-manipulating activity on the platform. In April 2020, the sports-gambling app FanDuel opened a presidential election market in West Virginia, only to have to quash it after 24 hours when state officials began to disapprove this “terrible idea” that had “no place in our American democracy.”
In a time when people have less trust in traditional electoral institutions than ever, when multiple presidential elections have taken days to count and certify (to the point where a loser attempted to carry out a coup), and when democratic processes hinge on reasoning that doesn’t neatly default to a “yes/no” hypothetical scenario, Kalshi may want to exercise more caution in what it’s trying to bring to America in its legal fight (which will continue for a while, because inevitably, the loser will try to escalate it to the Supreme Court). Because where is the actual public utility in letting even more money, and the petty disputes that arise from it, supposedly inform and “predict” our already fraught elections?
“You’ll look at a market price and go, ‘Oh my gosh, it’s up, it’s down’—it’s meaningless. It’s just noise, it’s a bunch of overconfident people trading on pure supposition,” said Gruca. “They sometimes do a better job than existing methods. Then sometimes they don’t. But nobody wants to hear that. They want to hear that there’s a magic bullet.”
