Rules-based disorder: why prediction markets struggle if reality is contested
The platforms let their users bet on world events. But when outcomes are ambiguous, who gets to decide what happened?
By Avantika Chilkoti
Two months before Venezuela’s presidential election in 2024, a professional gambler known as Domer decided to take a punt on the incumbent, Nicolás Maduro.
There was no guarantee Maduro would win: Venezuela’s economy had been wrecked by years of mismanagement, and the opposition had united behind a credible opponent, Edmundo González.
But Domer knew Venezuela had a history of sham elections and suspected that Maduro wouldn’t relinquish power.
Domer logged on to Polymarket, a prediction market, and started to place a series of bets on Maduro winning the election.
Polymarket and its main rival Kalshi—from the Arabic for “everything”—allow users to wager on the outcomes of real-world events.
Will Binyamin Netanyahu leave office this year?
When will Taylor Swift announce she is pregnant?
Will America confirm that aliens exist?
Whereas casinos and bookmakers operate as “the house”, setting the odds and taking the opposite side of every bet, prediction markets let punters trade against one another.
The platforms are appealingly simple to use.
Traders log in, pick an event market (for example, whether Maduro will win the election), and choose between “yes” or “no” outcomes.
What they’re really doing is buying a contract that pays a dollar if their prediction is correct and nothing if it’s wrong.
Because of that, prices always stay between $0 and $1, which neatly reflects the probability the market puts on an event occurring.
For example, if a “yes” contract on Maduro winning is priced at 62.5 cents it implies traders collectively believe there’s a 62.5% chance that will happen.
As the election drew closer, Domer noticed footage of large opposition rallies spreading across social media and thought González sounded increasingly confident.
Yet Polymarket traders continued to give Maduro around an 80% chance of victory.
By this point Domer had made scores of trades, buying and selling both sides as the odds changed, even though a Maduro victory would still be his most profitable outcome.
But the night before the election, Domer unexpectedly changed his position.
He placed big bets on Maduro losing and on González winning.
“It’s smart to bet against what people collectively believe,” he told me when we spoke in May.
Domer stood to lose several thousand dollars if Maduro stayed in power, but earn tens of thousands if González won.
It wasn’t that Domer had changed his mind about the most likely outcome—“Everyone expected the election to be rigged,” he said.
But he hoped that a strong challenge to the official result would shift the odds enough for him to make a profit by selling his González victory contracts at a higher price.
The election unfolded as expected in late July, when Venezuela’s authorities declared Maduro the winner with 51% of the vote.
But even Domer hadn’t foreseen what came next.
The opposition released voting receipts printed from election machines across the country, which showed that González had won decisively.
Both candidates claimed victory.
China, Cuba and Iran congratulated Maduro, while most Western democracies rejected the official result.
In Caracas, Venezuela’s capital, angry crowds protested in the streets.
The chaos was mirrored on Polymarket, where the odds of a win for Maduro swung wildly as traders reacted to competing claims.
As millions of dollars became tied up in the outcome, the dispute escalated into what traders call a “rules fight”—a contest not about what was happening in the real world, but about how Polymarket should define what had happened.
Arguments broke out in the comments section of Polymarket, and on X and Discord, the social-media sites of choice for betting nerds.
Some wanted Polymarket to settle the bet as a win for Maduro; others wanted it to reflect the reality that González got more votes.
“Maduro has been announced as the official winner via official channels.
Anything other than resolving this market to yes soon would be a straight-up scam,” said one user.
Another wrote: “The bet is on the election’s winner, not on who holds power.”
Domer continued betting on González.
He was following updates obsessively and had stopped sleeping and eating properly.
When markets are contested, traders’ arguments can shape how Polymarket settles a bet.
“If there’s a rules fight you’re defending your idea and you have to be very persistent about it,” Domer explained.
“If you’re sleeping, that is time for other people to spread their propaganda.”
On August 4th he did something that traders who later lost money would never forgive him for.
In a 1,500-word post on X, he argued that Polymarket should resolve the market as a win for González.
“The other road—rubber stamping a laughably stupid fraud, because an official state actor says so—seems like a far riskier road for prediction markets to go down,” he wrote, implying that the platforms shouldn’t be seen as condoning cheating.
Two days later, with more than $6m wagered on the election, Polymarket closed the market as a win for González.
Domer made more than $50,000—the biggest profit on the trade.
The outcome provoked fury among some users, who argued that, as a “whale” with a big online presence and more than $4m in cumulative winnings, Domer had improperly swayed the decision.
(When I put this to Polymarket, a spokesperson responded: “This is false.”)
Someone on X wrote: “Ironic that you respond to Maduro rigging the election by trying to rig the prediction market.”
Domer said people confront him about the trade to this day; last year an X user called him a “manipulative psychopath”.
It’s not just full-time punters like Domer piling into prediction markets: millions are trying their luck, from stay-at-home mothers to finance students to delivery drivers.
Nearly 40% of American men aged 18 to 34 have placed a bet on a prediction market, along with more than 20% of young women.
In June Kalshi processed $33bn in trades; Polymarket handled almost $14bn.
The platforms are backed by founders of finance firms, such as Charles Schwab and kkr, alongside tech companies such as Palantir and Airbnb.
Based on recent news reports, Kalshi and Polymarket are fundraising at valuations of around $40bn and $15bn, respectively.
The bosses of these platforms, all 30 or under, are among the world’s youngest self-made billionaires.
As prediction markets have grown, cases of insider trading and market manipulation have grabbed headlines.
In April a soldier involved in Donald Trump’s raid on Caracas to capture Maduro was accused of making around $400,000 on Polymarket by using top-secret information to make a bet.
Around the same time, French authorities opened an investigation after a weather sensor at Charles de Gaulle airport was allegedly manipulated with a hairdryer—a ploy that could have made thousands of dollars for Polymarket traders betting on Paris's peak temperature.
Last year the boss of Coinbase, a crypto firm, ended a call by reeling off buzzwords that punters had bet on him saying, including Ethereum, blockchain and Web3.
More recently, a little-known indie pop song released in 2024 unexpectedly reached number one on Spotify usa, just as suspicious bets were placed on Kalshi as to whether the singer would top the chart by the end of June.
Spotify later removed more than half a million streams after finding signs of bot activity.
“It is really the Wild West right now—anything goes,” said Andy Kim, a Democratic senator from New Jersey who has co-sponsored legislation to tighten rules around prediction markets.
To prevent manipulation and insider trading, the exchanges have increased surveillance and have suspended and fined offenders.
Both Kalshi and Polymarket told us they had processes in place to deal with suspicious activity and refer trades to law enforcement and regulators where necessary.
But, as the 2024 Venezuelan election exposed, prediction markets suffer from a more fundamental weakness: ambiguity in their rules.
In the race to attract users, the platforms have been launching new markets without giving enough thought to how their bets should be settled.
That matters, especially because prediction markets are expanding beyond the relatively straightforward two-way bets that gamblers have wagered on for centuries—such as sport results and election outcomes—into murkier territory.
Did Israel “invade” Lebanon when it began limited ground raids?
Was Ukrainian President Volodymyr Zelensky really wearing a “suit” when he ditched his military fatigues for a black jacket and trousers at a summit?
Did Cardi B “perform” at the Super Bowl when she briefly appeared in the background of another artist’s set?
In the most extreme cases, ambiguous rules have led to multimillion-dollar disputes, with traders clashing over how markets should be resolved and exchanges forced to make costly payouts.
The biggest losers, however, are often ordinary users.
Unlike deep-pocketed whales such as Domer, they may not understand how to interpret the nuances of the platforms’ rulebooks.
The credibility of prediction markets, meanwhile, is on the line.
They present their odds as reliable forecasts: both platforms have formed partnerships with media networks and their stats are cited alongside those from established pollsters.
Their goal is to compete with traditional financial markets, allowing asset managers and hedge funds to bet directly on the future rather than relying on complex portfolios of stocks and bonds as proxies.
But Wall Street institutions will shy away from betting contracts with fuzzy wording.
And if the rules governing these platforms can’t be trusted, neither can their odds.
The central challenge of economics, Friedrich Hayek argued in the mid-20th century, is the “knowledge problem”: how governments and businesses can harness information scattered across millions of people.
By the 1980s economists were growing excited about a potential solution: prediction markets.
The hope was that allowing participants to trade contracts on future outcomes would create prices reflecting the “wisdom of crowds”.
When people put money behind their forecasts they are more likely to spend time doing research and to challenge their own biases.
Prediction markets could therefore be a more accurate and responsive guide to the future than expert forecasts or opinion polls.
As Justin Wolfers of the University of Michigan put it, “the average estimate of any ten dumb people is better than that of one dumb person.”
In 1988 the University of Iowa launched the first serious prediction market.
Studies found that the Iowa Electronic Markets (iem), which forecast American elections, consistently outperformed opinion polls such as Gallup.
In 2014 came PredictIt, a non-profit political betting exchange that was initially run in partnership with a New Zealand university.
It quickly became popular with political junkies, proving that as well as money to be made, there was a lot of fun to be had.
“People want to prove themselves, especially young men,” said Robin Hanson, an economist who led early research into prediction markets.
For decades, regulation held the platforms back.
In America they fall under the Commodity Futures Trading Commission (cftc), which oversees derivatives markets.
The cftc has, in the past, taken a dim view of prediction markets, worrying about the potential for misuse.
Bets on sports and elections also blur into gambling, which is regulated separately by state law.
There were workarounds.
The iem and PredictIt were granted special exemption by the cftc, with conditions placed on the size and scope of bets they offer.
But there were also casualties: Intrade, an Irish exchange, collapsed after it was sued by the regulator for allowing unauthorised trading by Americans.
For a long time, it looked impossible to run a large, for-profit prediction market in America—until Kalshi came along.
When I met Luana Lopes Lara, the platform’s 30-year-old co-founder, in May she described herself variously as “resilient”, “stubborn” and “ok with suffering for a long period of time”.
Growing up in Brazil, she trained as a ballerina at the Bolshoi Theatre School, an offshoot of the Russian institution, dancing for hours each day and tracking her calorie intake to a fraction of a strawberry.
She somehow found time to win medals in international maths and science competitions before going on to study computer science and maths at the Massachusetts Institute of Technology (mit).
Also on her course was Tarek Mansour.
Like Lopes Lara he had grown up outside America—in Lebanon—and was intensely driven: he had reportedly taught himself English and was a competitive skier.
In summer 2018 the friends interned at Five Rings Capital, a quantitative trading firm in New York.
They noticed that investors often used complex bundles of stocks, bonds and currencies to bet indirectly on events like elections or economic data releases.
It made them wonder why investors couldn’t bet on the event directly.
And why shouldn't these bets be opened up to the public?
There was one big hurdle to overcome.
Mansour and Lopes Lara were determined to build a cftc-regulated exchange.
“From the start we had this deal that we’re never going to do anything that breaks the law,” Lopes Lara said.
The cftc made them answer long lists of questions about market manipulation and insider trading.
Despite the arduous back-and-forth, the pair were convinced their startup was well within the law.
Timing was on their side: the rise of cryptocurrency had shown regulators how quickly financial activity could grow outside traditional oversight, making bans feel pointless.
In 2020 Kalshi was approved as the first regulated American exchange dedicated to trading contracts tied to events.
It listed markets on everything from the Oscars to the fate of legislation going through Congress.
Mansour and Lopes Lara managed to secure investment from venture-capital firms.
But Kalshi remained niche, used by a few traders making modest bets.
The platform took off in the run-up to the 2024 American presidential election.
The cftc had tried to block the exchange from listing election markets, arguing that political betting amounted to “gaming” and could threaten the integrity of the electoral process.
But Kalshi challenged the decision in court, claiming that the regulator was overstepping its authority.
Weeks before the election, a federal judge sided with Kalshi.
The platform began taking bets almost immediately. In the week of the presidential vote, the exchange processed around $750m in trades—more than it had handled in its entire history.
While political pundits and pollsters agonised over the outcome, traders on Kalshi were leaning towards Donald Trump, proving the wisdom of crowds once again.
By choosing to play by the rules, Kalshi has taken a difficult path.
It must verify the identity of its traders, take steps to prevent market manipulation and submit its rules to the cftc.
Its chief rival, Polymarket, founded by a cherub-faced New York University dropout called Shayne Coplan, took the opposite approach.
Even though its main office is in New York, the company claims to be based offshore, beyond the reach of American law.
Its international exchange also operates in cryptocurrencies rather than dollars, allowing users to trade anonymously.
“You have two horses,” Lopes Lara said of the two firms’ rivalry, “but one has a carriage to pull and the other is free to run.”
Not that Polymarket has had entirely free rein.
In 2022 the cftc accused the firm of running an unlicensed exchange.
In a settlement, Polymarket paid a $1.4m fine and pledged to stop taking bets from Americans.
It has since launched a regulated American platform, Polymarket us, which takes dollar-denominated trades and operates under the same rules as Kalshi.
But many users still access the original platform using tools that mask their location.
In June trading volume on the regulated exchange was about 30% of that on the unregulated international site.
Since Trump’s re-election, both firms have had an easier ride.
The president has turned the cftc from watchdog to cheerleader, purging officials who took a tough line on prediction markets and backing the exchanges against state governments trying to restrict them under local gambling laws.
His son, Donald Trump junior, is an adviser to both Kalshi and Polymarket.
Even if a more cautious Democratic administration takes over, it seems unlikely that the industry will be banned, given how big it has become.
One serious threat to prediction markets could lie within.
Lopes Lara told me she sees Kalshi’s ability to define and resolve bets clearly as “existential”.
“At the end of the day we’re going to grow a lot faster if the users trust that we are going to settle things correctly,” she said.
If Lopes Lara and Mansour are king and queen of Kalshi, then Xavier Sottile and Nicole Kagan are judge, jury and executioner.
As leaders of the markets team, they run the listing and settlement of bets.
The pair were star debaters at Yale and Harvard respectively, which stands them in good stead, Kagan told me.
Debaters are “very granular, extremely meticulous people who get caught up in words,” she said, adding they are used to having their opinions challenged.
“We can volley ideas back and forth with no ego.”
To come up with bets, the markets team scans the news and takes suggestions both from Wall Street and from traders on Discord and X.
On any given day, Kalshi has more than 10,000 live markets.
“We have a bias to list,” said Kagan, which makes sense as the firm makes money from transaction fees.
Then comes the tricky part: writing the rules.
Many markets follow existing templates, but fresh guidelines are sometimes required for a new market.
Either Sottile or Kagan takes the lead, pinning down every detail—from when a market should settle to the source Kalshi will use to decide the outcome.
For interest-rate decisions it might be a central bank, and for something like a celebrity-wedding location, a reputable newspaper.
Drafting the rules can take weeks.
Noah Zingler-Sternig, who until recently ran operations at Kalshi, described a “two-sided mandate”: rules must be detailed enough to eliminate ambiguity, yet simple enough for traders to understand.
“If an exchange is optimising to write perfect rules, they’re never going to list a single market,” he told me.
When the team is satisfied with the rules, they run them through an ai tool for a final check before sending them to the cftc.
Some bets are easier to define than others.
In sport, which still accounts for the vast majority of trading on prediction markets, outcomes are usually clear-cut: there are only so many ways a game can go, and the league or association governing the sport is an obvious source.
Centuries of betting have also created conventions for handling unexpected outcomes, such as rain delays or disputed calls.
In areas like geopolitics or entertainment, outcomes can be more ambiguous, making rules trickier to write.
The Super Bowl provided the backdrop for two of Kalshi’s most popular markets.
One was on which firms would advertise during the broadcast.
Kalshi listed dozens of brands, and users took yes/no positions on whether each would appear.
(A 30-second advert mid-match counted; brief logo flashes on stadium billboards didn’t.)
Sources were listed in order of precedence: first the National Football League, then broadcasters such as cbs Sports, then other media outlets.
The other was on who would perform during the halftime show.
Kalshi defined a performer by their legal name, stage name or widely recognised professional identity.
What counts as a performance, though, has since had to be tightened.
After Cardi B’s brief appearance this year, the rules were updated to clarify that “singing” could either mean being the “audible lead or [providing] backing vocals”.
“We’re always learning,” Kagan said, “and it’s pretty often public.”
For typos and omissions in its rules, Kalshi runs a bounty scheme that pays users rewards ranging from $25 to more than $1,000.
One bounty hunter, who goes by ProTrader on Discord, told me he has carved out a niche in misspelt names.
His first prize was $25 for spotting the New York City mayor’s first name written as “Zorhan”.
He later built a tool that scanned the rules for a market on candidates for the Texan Senate and made $350 by flagging around ten errors at once.
Users’ main gripe with Kalshi is its insistence on adhering strictly to its rules, even when that produces perverse outcomes.
The exchange settled a market on how many people had watched the 2025 Oscars by using as its source a New York Times article, which cited an estimate of 18m by Nielsen, a research firm.
When Nielsen later revised the audience figures to 19.7m, Kalshi refused to overturn its decision, arguing that its rules forced it to settle based on the first reported figure.
The bet on whether Anthropic would advertise during this year’s Super Bowl—which took in $33m from punters—closed as a “no”, despite the ai company running an ad for its chatbot, Claude.
Kalshi argued that the advert didn’t mention Anthropic itself, as per its rules.
“You guys are scammers,” one user commented on Kalshi’s website. “We all have been cheated,” another complained.
To Scottilicious, a trader who has made over $1.5m from forecasting, Kalshi’s insistence on following the rules to the letter misses the point of prediction markets.
“When you’re placing a trade it is to predict what’s going to happen in the real world,” he told me, “not fit some very definite paradigm.”
That rigidity reflects the fact that Kalshi answers to American regulators.
Polymarket’s main exchange, by contrast, prefers to settle disputes on what Scottilicious calls “vibes”.
Earlier this year, both platforms were running markets on when Ali Khamenei would be “out” as Iran’s supreme leader.
Punters poured in money as tensions between Iran and America ramped up.
On February 28th America and Israel launched strikes on Tehran, prompting a furious burst of activity on Kalshi, even though it was still the middle of the night for its mainly American users.
As rumours of Khamenei’s possible demise circulated, the odds of him being “out” jumped sharply.
Within hours, Western media outlets were reporting that Khamenei was presumed dead, even though Iranian officials insisted he was safe.
Kalshi’s market had ostensibly been on Khamenei’s peaceful departure from office—by federal law, the cftc is allowed to ban markets on war or assassination.
Kalshi’s rulebook included a so-called death carve-out, specifying that if Khamenei died the market could settle not as a “yes” or a “no” but at the last traded price “prior to the death”.
But even as reports of strikes on Tehran rolled in, Kalshi kept taking bets and promoted the market on social media, to the consternation of some users.
On Discord, Domer warned that the Khamenei markets risked turning into “a macabre circus”.
Others argued the rules were technically clear but asked how the time of death would be determined.
One user added they expected the market to settle at the pre-strike price, “instead of incentivising murder”, and hoped the platform would “do the right thing”.
By mid-morning, Sottile and Kagan realised they had to step in.
Kalshi has a process for clarifying ambiguities quickly without involving too many people (this is partly to avoid insider trading).
Once the team becomes aware of a dispute, it can pause trading for a few minutes, clear the order book (so bets that have yet to go through are cancelled) and issue a clarification.
All traders receive a notification and an update appears on the website in a bright green box.
But the clarification Kalshi posted that morning only added to the confusion.
It said the market would settle at the last traded price before “confirmed reporting” of death.
Traders said this differed from the wording in the rules, which referred to the death itself.
Given how sharply prices had moved that day, the distinction could make a big difference to payouts.
Also, wondered traders, what counted as confirmation?
On Discord users began tagging Kalshi staff.
“This is a huge deal,” one user commented.
Around midday Kalshi clarified its clarification.
If Khamenei died, the market would settle “upon the confirmed reporting of the death” at a price “based on the last traded price prior to death”.
If that price was unclear, the exchange would convene a three-person committee to determine a fair outcome.
To Flip Pidot, who does a similar job to Sottile and Kagan at PredictIt, this was a case of “sloppy rules, followed by sloppy adjudication”.
Around 10pm that night Kalshi emailed traders to explain that bets made before the strikes would be settled at the price just before they happened, while bets placed afterwards would be cancelled and stakes and fees refunded.
It was a rare case where Kalshi paid users out of pocket.
The bill reportedly came to around $2.2m.
“I’m sorry for the disappointment,” Mansour tweeted.
“We’ll improve, thank you for bearing with us.”
(Kalshi has since changed its rules to say the firm can suspend trading if it “reasonably believes" that a death “has occurred, is imminent, or that circumstances giving rise to the death may be occurring”.)
Still, many punters who had bet that Khamenei would be out of office were irate.
They had expected to be paid out in full and believed Kalshi was depriving them of their winnings.
They had become, in traders’ jargon, “rules cucks”, cheated by Kalshi’s adherence to the rules.
They argued it was disingenuous to list a market on Khamenei leaving office if assassination wouldn’t count.
Users also claimed that Kalshi hadn’t applied its rules consistently.
When Jimmy Carter, the former American president, died before Trump's inauguration, Kalshi quietly settled a market on whether he would attend the ceremony as a “no”.
In a class-action lawsuit over the Khamenei market, brought on behalf of more than 100 traders, Kalshi is accused of “unfair competition, deceptive corporate behaviour and consumer fraud”.
Kalshi denies wrongdoing.
The firm calls episodes such as this “edge cases”: rare outcomes nobody had foreseen.
Yet as prediction markets expand into ever more areas of life, unexpected events will happen more often.
The challenge will lie in tightening the rules while keeping traders on side.
Kalshi’s insistence on doing things by the book can frustrate its users.
But Polymarket’s “vibes”-based culture leads to even bigger problems.
The exchange’s rules around trades are short and loosely worded, with outcomes relying on a “consensus of credible reporting”, a standard that Polymarket describes as “flexible”.
When markets are contested, the platform outsources resolution to holders of a cryptocurrency called uma.
As Coplan, the founder of Polymarket, put it during a talk at Harvard University: “That’s where shit gets messy.”
Until recently, anyone has been able to suggest a market should close in a certain direction.
In the event of a dispute, uma tokenholders can vote for one of four options: yes, no, 50-50 (if the outcome is unclear), or too early to call.
The process is far from democratic: the more uma tokens someone has, the more weight their vote carries.
In an analysis for The Economist, trm Labs, a blockchain investigations firm, looked at more than 3,250 uma disputes over the past three years.
It found the five biggest wallets together controlled on average 40% of the vote and the ten biggest almost 58%.
In recent disputes, the largest wallet, known as BornTooLate.eth, held almost 8% of the vote.
There is also nothing to stop a trader from voting on the outcome of a bet they have made themselves.
I asked Dune, a blockchain data provider, to find out if this was happening.
Analysts at the firm linked eight of the 200 most profitable wallets on Polymarket to uma voters by tracing funds that were moving between them.
In one case, a uma tokenholder linked to one of the biggest Polymarket trading accounts voted in 41 markets which that account had bet on.
In more than half the instances, the tokenholder was on the winning side, resulting in $7.4m of profits (including $1.8m on whether America would ban TikTok before May 2025, $1.3m on whether Israel and Hizbullah would agree a ceasefire in 2024 and $1m on whether Israel would strike an Iranian nuclear facility that same year).
When contacted for comment, Polymarket said that 99.986% of markets resolve without objection and pointed out that the exchange doesn’t stand to gain financially if a market closes one way or the other.
“We are the most transparent marketplace on the planet, and view democratisation and scrutiny as a feature,” a spokesperson said.
One of the most contentious markets ever listed on Polymarket was on whether America and Ukraine would reach a deal on critical minerals by April 2025.
In late February the odds surged after officials said an agreement had been reached.
But following a heated meeting between Trump and Zelensky days later, talks collapsed before any paperwork was signed.
The odds of a deal crashed from almost 100% to around 40%.
A few weeks later, with no updates from either side, a trader called 50-Pence proposed closing the bet.
There is a 48-hour voting period for uma holders.
During that time, many tokenholders state which way they are leaning on Discord (though there is no way to tell how they voted as the ballot is secret).
Meanwhile, trading continues on Polymarket, so prices can move even as the outcome is being decided.
Some voters said that the February announcement meant a deal had already been reached, even without signed paperwork.
Others pointed out that another market on whether a deal would be reached on the day of the White House meeting resolved as a “no” (meaning that a deal hadn’t been reached), and that it was too early to settle this one.
Prices swung with each new post on Discord.
Minutes before voting ended, Polymarket shared a notice on its website: “The us and Ukraine have not mutually agreed to a qualifying deal.
It is too early to resolve this market.”
The odds of a deal dropped.
Many traders understood that uma tokenholders rarely go against Polymarket’s guidance, but some did not realise it was too late to change votes that had already been cast.
When the ballots were counted, “yes” won, with roughly 55% of the vote (BornTooLate.eth controlled 6.87% of the total, according to Dune data).
uma tokenholders had decided there had been a deal, even though to all intents and purposes there hadn’t.
Polymarket has on occasion overturned uma decisions, including on a bet about whether Barron Trump, the president’s son, was involved in the creation of a cryptocurrency associated with the family.
In the case of the Ukraine minerals deal, the firm apologised to traders but stopped short of issuing refunds.
“This is an unprecedented situation,” a Polymarket official wrote on Discord.
“This is not a part of the future we want to build.”
Ironically, disputes often fuel trading.
Take a Polymarket bet on whether Portugal captain Cristiano Ronaldo would cry at the fifa World Cup.
Shortly after Portugal was knocked out by Spain, about $2m had been staked.
Then the debate took off.
Had Ronaldo actually cried, or was it sweat?
Punters dissected close-ups of his face, some doctored to add tears.
By the time uma voted that Ronaldo had cried, almost $90m was on the line.
Since the minerals deal market controversy, the rules have been changed so that only a select “whitelist” can propose closing a market, including longtime users, Polymarket staff and the team behind uma.
(Others can tag them on Discord and ask them to intervene.)
Before, a simple majority was all that was needed to resolve a market in one direction; now you need 65% of the vote.
Hart Lambur, co-founder of uma, denied that some tokenholders had outsize influence.
“A big holder who votes for a falsehood only profits if everyone else also lies, and with no way to co-ordinate that in advance, large stakes have the same incentive to vote honestly as small ones,” he told me.
uma voters are unapologetic about the power they wield.
A group called uma.rocks pools votes from roughly 55 tokenholders and casts them through a three-person committee, effectively controlling around 8.5% of ballots (the size of the syndicate fluctuates).
One committee member, whose username is JessicaOnlyChild (a reference to the Korean film, “Parasite”), posts detailed analysis on Discord ahead of votes, a time-consuming task given that dozens of markets can be disputed on any given day.
Some traders suspect the person behind the account is making big money on Polymarket.
I tried to contact the user on Discord but received no reply.
Lancelot Chardonnet, uma.rocks's founder, told me that JessicaOnlyChild runs a data-centre business, doesn’t trade on Polymarket and takes part in uma debates purely for fun.
“It’s a strange hobby,” he said. “But then, some people collect stamps.”
Because uma holders lose tokens for voting against the majority, many watch uma.rocks’s comments on Discord closely.
Polymarket traders have told me they switch their bets if uma.rocks and JessicaOnlyChild appear to be voting the other way.
One Polymarket trader, ContraQ, looked at more than 300 disputes in the first quarter of this year and found around 100 cases where uma.rocks and JessicaOnlyChild both signalled they planned to vote either yes or no.
In all those cases, the bet closed the way they were leaning.
“It’s extremely profitable following their opinion,” he said.
The result is that Polymarket traders are no longer basing their predictions on what they think will happen in the world, but on how they think the most powerful uma tokenholders will vote.
Coplan has hinted that uma could soon be replaced, though it is not clear with what.
Polymarket’s whales believe the system is broken.
50-Pence compared the exchange to a struggling casino group with expansion plans: “They want to build a swimming pool.
Meanwhile, there are people fighting the card dealers.”
Domer was even more direct: “uma is far more vulnerable than it was a year ago,” he posted on X in April, “and the inmates are starting to take the asylum.”
At the entrance to Kalshi’s office in Manhattan’s Chelsea neighbourhood—past a reception desk scattered with poker chips—hangs a still from a cnn broadcast showing Kalshi’s odds on a third term for President Trump.
The firm moved to the office last September and already needs more space; in that time, its workforce has more than doubled to 170 employees.
“It’s crazy,” said Lopes Lara.
Another sign of how far prediction markets have come is their growing presence on the conference circuit.
There used to be only one event worth going to: Manifest.
Held in Berkeley, California, it was described by the New York Times as “equal parts Math Olympiad and Burning Man”.
A few years ago, in a clear case of insider trading, a bet on whether there would be an orgy there closed as a “yes”.
In March Kalshi held its first research conference in New York.
A few weeks later I attended a prediction-market event in Las Vegas—an awkward setting for an industry trying to distance itself from traditional gambling.
Over cups of tepid coffee and toothpicks stacked with tiny mozzarella balls, I met college dropouts building startups that aggregate data across exchanges, entrepreneurs pitching prediction-market kiosks for petrol stations, and young women aspiring to be prediction-market influencers.
Punters greeted each other by their Discord handles and dissected old trades. “Nicole” and “Xavier” were spoken of as if they were old friends.
Domer turned up looking nondescript in a blue polo shirt and thick plastic glasses.
But people in the crowd spotted him immediately.
Domer, whose real name is Ken, knows Las Vegas well.
He moved there in his 20s to become a professional poker player.
One night, scrolling on his phone between hands, he discovered InTrade, the Irish prediction market that was a forerunner to Polymarket and Kalshi.
His first big win on InTrade was in 2008.
John McCain, the Republican presidential candidate, was due to announce his running mate at Wright State University in Ohio.
Few expected him to pick Sarah Palin, the inexperienced governor of Alaska.
But Domer, who was tracking arrivals at airports near the university, spotted a plane from Anchorage.
Twelve hours before McCain’s speech, he staked $1,000 on Palin.
The payout was $25,000.
He quit poker soon after and turned to “future-gazing” full time, adopting as a pseudonym the nickname given to fans of his beloved University of Notre Dame football team.
His profile pictures cycled through Ryan Gosling stills; when Gosling played Ken in Barbie, the image stuck.
Over the years, Domer has become a prediction-market celebrity.
Traders copy his bets and exchanges ask his advice.
At the conference’s welcome drinks, the organiser even called for a round of applause.
“This is the direct opposite of my personality,” Domer told me afterwards.
Jeff Yass, co-founder of Susquehanna, an investment firm, also used to be a professional poker player in Las Vegas.
Whether a bet is made in a casino, on Wall Street, or on prediction markets, Yass sees them as fundamentally similar: “They’re decisions made under uncertainty.”
As more money flows through prediction markets, Susquehanna’s interest in the industry is growing.
The firm has around 75 staff trading in prediction markets; Yass hopes this figure will soon rise.
He is mindful that the platforms still have some growing-up to do, but told me Susquehanna is closely watching top traders.
“We’re losing a lot of money to people that are smarter than us”, he said, “and we’d like to hire them.”
Avantika Chilkoti is global business correspondent at The Economist
0 comments:
Publicar un comentario