Over the past year, John Aristotle Phillips and I have placed a lot of bets. Meeting regularly for sandwiches and beers at Soho House, a members’ club in Manhattan and his de facto office, we debated and wagered on the outcome of many contests — football games, macroeconomic data, Kentucky Derby thoroughbreds, election results and more.
If Phillips, 68, is well known, it’s for designing a nuclear weapon while he was an undergraduate at Princeton in the 1970s. The FBI confiscated his term paper and his mock-up of the bomb, and the press dubbed him the A-Bomb Kid. As a young man in the early 1980s, he ran twice, unsuccessfully, for Congress in Connecticut. In 1983, he founded Aristotle Inc, a technology consultancy that sold voter data to political campaigns. About a decade ago, he started PredictIt, which is currently the only legal popular venue for political-prediction trading in the US. A non-profit collaboration with Victoria University of Wellington, in New Zealand, the company offers markets on who will win gubernatorial elections, Senate races and the presidency.
Our meetings were prediction-market making in miniature. We discussed the events of the day and their particulars, we bid prices and reconsidered our own positions. Then we put money on it. For Phillips, prediction markets are a truth generator, powered by the invisible hand. Therein lies their public good. If you trade based on fake news or half-baked punditry, you’re going to lose your money. The problem, he says, is not that there is too much predictive betting but that there is too little. “It’s better to know, in life,” he told me. “It’s better to know, as a general principle.”
In August, six United States senators wrote a letter to Rostin Behnam, chairman of the Commodity Futures Trading Commission (CFTC), decrying betting on the outcomes of American elections. The senators were weighing a proposal to sanction another prediction market. It’s a supremely dangerous business, they argued — “a clear threat to our democracy” — prone to exploitation by political insiders and the super-rich, and corrosive to trust in elections. It went on to enumerate the deleterious effects of citizens believing “that the democratic process is being influenced by those with financial stakes”.
Of course, elections always have inherent financial incentives — they determine trillions of dollars of government spending, for example. And Americans are already well aware that their democratic process is influenced by those with the means to donate to and lobby elected officials. Detractors argue prediction markets are just more of a bad thing — the cash that could break the system.
Proponents say markets encourage engagement with the democratic process. They incentivise deep understandings of rules and structure, the internal clockwork of government. They encourage a more informed electorate. While so much popular political coverage is unaccountable punditry, full of its own empty predictions, betting markets motivate getting to the truth of the matter.
At the same time that American politics has devolved into debates over the nature of facts and truth, legal betting has swept the US. Sports betting, for example, increased from $470mn in 2018 to a projected $7.6bn this year, and advertisements for it permeate primetime television. Meme stocks and crypto became mainstream during the pandemic. Yet political betting remains a taboo, stuck in legal limbo. And with a crucial general election less than a year away, the parties fighting over its legality are waging their own shadow campaigns.
Americans have been wagering on presidential elections since George Washington and doing so in organised markets since at least Abraham Lincoln. More than a century ago, in the hotels and billiard halls of old New York and on the streets outside its stock exchanges, citizens swarmed to place bets on their next political leaders. Glorified bookies offered prices shifting with demand and kept healthy cuts for themselves.
In an era before scientific polling, these markets served a unique purpose — forecasting. Major newspapers would report their prices daily during the campaigns, as a barometer of the electorate. Modern scholarship finds that these early markets were remarkably accurate, with more predictive power than any other readily available source.
At times, the trading volume on presidential candidates surpassed that of securities. In the 1916 election, between Woodrow Wilson and Charles Evans Hughes, the size of prediction markets reached the equivalent of $280mn in today’s dollars, according to a paper by the economists Paul Rhode and Koleman Strumpf. “Betting in the run-up to elections was a cherished ritual,” the economists write. “A widely held value was that one should be prepared to ‘back one’s beliefs’ either with money or more creative dares.” These included the loser literally eating crow, or pushing the winner around in a wheelbarrow.
Rhode and Strumpf argue that these markets largely disappeared because of the advent of scientific polling and the availability of other forms of gambling, such as literal horse races. The senators’ concerns — moral hazard, election tampering, strategic manipulation — accompanied these historical markets too, without much bite. Rhode and Strumpf conclude that “many current concerns about the appropriateness of prediction markets are not well founded in the historical record”.
A share on PredictIt settles at $1 if the listed event comes to pass and $0 if it doesn’t, and its price fluctuates somewhere in the middle meanwhile. Shares can be bought and sold at any time, as a candidate’s fortunes wax and wane before election night, say. The price, therefore, can be read as a probability. A share of Joe Biden winning re-election, for example, is trading at 43 cents, implying a 43 per cent chance of him winning a second term.
Operating such a market in the US requires the blessing of the CFTC, which regulates the country’s derivatives markets, including futures. In general, following a 2012 ruling, the markets it regulates cannot offer political-event contracts. The CFTC cited the Commodity Exchange Act and “determined that the contracts involve gaming and are contrary to the public interest”. In their letter to the CFTC, the senators, too, wrote derisively that certain political markets “could effectively allow legal gambling”.
There is a tricky definitional issue at the heart of the argument over political prediction markets. In a very real sense, everything is gambling. We take our biggest gamble the day we’re born and smaller ones every day after that. And, in a very real sense, everything is gaming, including voting itself. “All voting is a sort of gaming, like checkers or backgammon, with a slight moral tinge to it, a playing with right and wrong, with moral questions; and betting naturally accompanies it,” wrote Henry David Thoreau in 1849, in Civil Disobedience.
There was, however, a narrow avenue of exception to the CFTC’s anti-politics position, which PredictIt sought to navigate. In 2014, the CFTC’s Division of Market Oversight granted PredictIt a “no-action letter”, giving it limited permission to operate. That permission was predicated upon important conditions: markets were strictly limited in size (5,000 traders) as were traders’ positions ($850). It also required that PredictIt be operated by Victoria University and that the data the marketplace generated be provided to academics.
The idea was that the markets would be small, limiting their ostensible threat to democracy, and that the research it spurred would be in the public interest. The market was a sort of academic laboratory. The arrangement had precedent in the Iowa Electronic Markets, at the University of Iowa, which got its no-action letter in 1992. In the years that followed, some 120,000 PredictIt traders took positions in some 8,000 markets. According to one scholar’s analysis, their collective forecasts were more accurate than even the most careful aggregations of polls.
(By way of disclosure, I traded small positions in Supreme Court markets on PredictIt years ago; I currently have 7 cents on deposit on the site.)
But in August 2022, the CFTC abruptly revoked its no-action letter, essentially shutting PredictIt down, writing that Victoria had “not operated its market in compliance with the terms”. It provided no further detail, justification or guidance.
PredictIt sued, arguing that the revocation was done “arbitrarily, capriciously and without legally required process”, and that there was no consideration given to “the chaotic, disruptive and economically damaging wind-down of the market [the CFTC’s] decision forces”.
The suit was filed in federal court in Texas, though the parties wrangled over the venue. The case wound up before the Court of Appeals for the Fifth Circuit, in New Orleans, one of the second-highest courts in the land.
The PredictIt legal case is known as Clarke vs CFTC. Clarke is Kevin Clarke, 31, a PredictIt trader and former competitive debater and debate coach. He lives in Texas and has a gig travelling the country mining gems and crystals, which he sells online to collectors. His educational background is in the philosophy of political systems.
On PredictIt, Clarke carved out a specialty in pardon markets, studying the minutiae of pardon and clemency procedures, a strategy which proved lucrative during the Trump administration. Following the lawsuit closely, as the lead plaintiff, has spurred Clarke to consider law school; his LSAT prep book sat nearby as we spoke. “PredictIt cultivates civic literacy,” Clarke told me. “It provides checks on how to interpret media, how to not just go by a soundbite, how to not allow a headline to take on a life of its own.”
PredictIt is an object lesson in epistemology, he said: learning the dance of political communication, its tactics and strategy, and how we, the voters, know what we know. “PredictIt answers a lot of questions about what you do to get beyond the cynicism of institutional modes of knowledge, the cynicism of reporters and journalists,” he said. “It provides an organic and anonymised way of understanding data.”
And Clarke objected to the characterisation of this sort of trading as mere gambling, something driven merely by chance or chaos. It is driven, he says, by empirics and careful study.
Trevor Boeckmann, 34, another PredictIt plaintiff, is a public defence attorney in New York City who works to reunite separated families. His trading specialty is cabinet appointments. “I’ve learnt more about Senate procedure than I ever knew I could,” he told me.
He started betting small amounts on the site in the early days of the 2016 presidential campaign. Over the years, he learnt to spot small edges in markets that were yet to close, but whose events were essentially decided. For example, in 2020, he bet thousands of dollars on Biden in states that Donald Trump claimed had been stolen from him, and profited handsomely.
In 2021, Boeckmann was on a hiking trip sitting around a fire with a group of people he’d just met; one of them was a lawyer for PredictIt. This chance meeting would later connect him to the suit. “We spent the rest of the night talking about all the characters on PredictIt and everything going on there,” Boeckmann said.
“What I support, big picture, is very different than what the CFTC is talking about,” he continued. It was less neon-clad casino and more considered roundtable. “We have this 24-hour news system of talking heads and people who are never held accountable for any of the predictions that they make. The power in PredictIt is that there are actual outcomes, dollar values. You can’t just say whatever you want to say if you put money on something — it really changes the calculus.”
The attention to detail yielded some interesting insights. In 2021, for example, PredictIt traders discovered that test ballots had been accidentally included in the official counts of a New York City mayoral primary, and brought the issue to the attention of the Board of Elections. They were ignored, though the error was later fixed.
Boeckmann did welcome a sensible dose of caution. “There needs to be guardrails, there needs to be some sort of regulation,” he said. But the CFTC delivered “a very opaque and drastic decision, given what is some real value that PredictIt is able to add to the political discourse”.
The Fifth Circuit court heard oral arguments earlier this year. A panel of three judges — two Republican appointees and one Democratic — heard the case, each sounding sympathetic to PredictIt’s position. The CFTC’s argument hinged on a complicated and at times paradoxical argument about “permission”, whether or not it had actually been granted, and its right to revoke it.
“This sounds like a licence to bully,” Judge James Ho told a CFTC attorney. “You can issue these threatening letters — there’s no judicial review.”
Around that time, the CFTC would not comment for this article beyond an email that read: “Due to the ongoing litigation, we are unable to arrange an interview with anyone on staff to talk about this case or discuss Victoria University’s non-compliance with the law.” Another more recent request for an interview went unanswered.
While he was in New Orleans for oral arguments, Phillips’s PredictIt hat caught the attention of a woman who told him that she paid her freshman room and board with the money she’d made off the Supreme Court nomination of Brett Kavanaugh. “That was the best day of my life,” Phillips told me.
One of PredictIt’s attorneys in its legal battle has been Michael Edney, a partner at Hunton Andrews Kurth in Washington, DC. Phillips called him “the Darth Vader of the Fifth Circuit”. “These markets are a crucial piece of information for folks who are trying to make a projection, or understand what is likely to happen in an election,” Edney said. “Talk is very cheap — it often has ulterior motives, seeking to influence the outcome rather than predict it.”
Scientific polling, Edney said, does not preclude the role of prediction markets. Respondents are sometimes loath to reveal their true beliefs, and polls are snapshots in time rather than direct predictions of a future election night. (Boeckmann also noted an infamous conservative pollster who “put out all kinds of bullshit polling” ahead of last year’s midterms, which didn’t budge the markets.) A prediction market, Edney said, can “strip away some of the pathologies that human beings have”.
In March, the CFTC replaced its initial terse withdrawal with a fresh one, laying out its position in more detail. It accused the project of being run not by Victoria University as agreed but by Phillips’s company Aristotle, of making too much money and of offering markets outside the scope of the agreement. (PredictIt takes a cut of profitable trades and withdrawals.) “It was pretty clear the guillotine was coming, and they were scrambling to justify what they were doing,” Phillips said.
Victoria responded in April with a lengthy letter, claiming that it had been open about its operating arrangements and had rarely made any money and defended the markets in question. It also said that, all along, “PredictIt was likely to evolve.”
The most interesting part of this back-and-forth concerned what constitutes a meaningful political question. The CFTC, for example, objected to the following markets, among others: will Pope Francis vacate the papacy? Will Caitlyn Jenner address the Republican National Convention? Who will win the Nobel Peace Prize? Will North Korea test a hydrogen bomb? “You can say the papacy is not a political issue,” Phillips said. “I’m sorry, but it is.” The pope is a head of state, after all. At the very least, these are good questions for a truth machine.
In early May, PredictIt scored an interim victory. In a short order, the Fifth Circuit ruled that the CFTC was temporarily “enjoined from closing the PredictIt market or otherwise prohibiting or deterring the trading of market contracts”. Phillips sent me an email with the subject line, “PredictIt Wins!” and offered to take me to his club to celebrate.
In July, the Fifth Circuit ruled that the CFTC’s rescission of the no-action letter was “likely arbitrary and capricious” and remanded the case to the district court to enter a preliminary injunction. In September, the Western District of Texas followed along, enjoining the CFTC — “pending consideration of the merits of the claims in this case” — from taking any action against PredictIt. The CFTC is once again arguing to change the venue to Washington, DC, where perhaps its regulatory arguments would fall on more sympathetic ears.
Similar markets elsewhere remain shut down. Congressional markets on Kalshi, the commercial prediction market that was the focus of the senators’ letter, were thrown out by the CFTC in September. These would’ve been much larger than PredictIt’s, and raise different questions. Kalshi had proposed a $250,000 cap on individuals’ positions and would’ve allowed certain firms to wager $100mn on which party would control Congress.
Tarek Mansour, Kalshi’s CEO, argues that many now-integral financial products were met initially with scepticism and opposition. “Insurance and grain futures were once distrusted concepts that evolved into the important risk-management tools we know today. In fact, in 1905, the US Supreme Court considered the case of whether future contracts were a form of gambling,” he told me in a written message. “The implication from our critics is that when Main Street gains access to the tools of Wall Street, it is gambling, failing to recognise that these contracts will democratise access to an important risk management practice.
“We’ve seen how political polls often lead to misinformation and distrust in the electoral process and we are aiming to fix this,” he continued. “Political prediction markets will represent a source of truth and credibility in the democratic process.”
“There’s no greater risk that Americans face nowadays than election risk,” Luana Lopes Lara, Kalshi’s CTO, said in an interview with Bloomberg in October. While large investors and institutions can access financial products with exposure to this risk, she said, it’s “a little crazy” to think that everyday Americans shouldn’t also be able to. Lopes Lara praised the CFTC as a regulator, but added, “Sometimes the government gets things wrong.”
In early November, Kalshi, too, sued the CFTC, in federal court in Washington, DC. It described the CFTC’s actions as “arbitrary, capricious and otherwise contrary to law”.
I spoke recently with Senator Jeff Merkley, of Oregon, one of the signatories of the letter to the CFTC. He emphasised his own work to secure the integrity of elections, and then considered the possibility of allowing big-money political markets. “We’re going to incentivise very rich individuals or groups of people to bet millions on an election outcome and incentivise, therefore, massive efforts to put their thumb on the scale and affect the race — it is hard to imagine anything more corrupting,” Merkley told me.
Merkley was pleased with the CFTC’s decision against Kalshi. “It should have never been in question,” he said. “The CFTC rules prohibit contracts that involve or relate to gaming. Betting on sports teams is gaming, betting on elections is gaming.”
But he wishes that PredictIt had come to him and his legislative colleagues for authorisation. “They should have come to Congress and said, ‘Well, we’d like to allow very limited gaming for research purposes.’” Under strict conditions — low ceilings on position sizes, transparency and checks for any corrupting influence — “I could see an argument for it,” Merkley said.
Precisely why the CFTC tried to rescind PredictIt’s no-action letter when it did is an open question. Phillips and Edney believe it has something to do with Sam Bankman-Fried, the convicted crypto fraudster. He had repeatedly urged Behnam, the CFTC chair, to allow his exchange FTX to offer political event contracts. FTX had advertised a “Trump 2020 futures contract”, along with similar ones for Biden, Bernie Sanders and Elizabeth Warren. Last year, the bankrupt exchange still held “TRUMPLOSE” coins on its balance sheet. Perhaps it was easier for the CFTC to pull the plug on everybody rather than having to deal with a wave of new requests all pointing to PredictIt.
In any case, some electoral traders are trading again and, as I write, Joe Biden is the 2024 presidential favourite. Donald Trump trails at 37 cents, and anyone else is a long shot.
Both Phillips and Edney believe, due to the “arbitrary behaviour” of the CFTC, that this case has implications for regulation far beyond prediction markets. “We believe this is going to change the way that the administrative state does business in this country,” Edney said.
He thinks that the courts are getting “fed up” with people far removed from anyone elected or accountable to the people, deep within administrative bodies, making these large decisions. This case, he said, will bring that frustration to a boil. “If they want to take it to the Supreme Court, we’ll see them there,” Edney said. “We’ll be there with bells on.”
Oliver Roeder is the FT’s US senior data journalist and author of “Seven Games: A Human History”
Follow @FTMag to find out about our latest stories first wherever you listen
Read the full article here
