As the 1549 papal conclave dragged on into a second and third month, Venice’s ambassador to the Vatican, Matteo Dandolo, was outraged by insider trading. “It is more than clear that the merchants are very well informed about the state of the poll,” he said, “and that the cardinals’ attendants in Conclave go partners with them in wagers, which thus causes many tens of thousands of crowns to change hands.”
Although the conclave was supposedly secret, Roman traders ran a thriving market in the streets around the Vatican trying to predict the winner. Odds shifted rapidly, and bettors appeared suspiciously well informed and perceptive. Everyone knew the race was wide open, and the original favorite in the crowded field, at 5-to-1 against (or 17% odds), was Cardinal Giovanni Maria del Monte. Ten weeks later, he was elected Pope Julius III as a compromise candidate, but only after opposing cardinals had brought forward a parade of options. Bettors on the streets helped the population understand who was leading, and by spotting early on that Julius might be the inoffensive candidate who would provoke the least opposition, they even guided the cardinals as they tried to break the deadlock.
Prediction markets grew so widespread and so disliked that by the end of the 16th century, Pope Gregory XIV had outlawed them. The penalty for betting on the identity of the next pope was excommunication.
Five centuries later, the urge to bet on elections has not abated. Neither has the whiff of corruption surrounding them, nor the sense that bettors’ money can move not just markets but election results themselves. For centuries there have been cycles of political betting — in 19th century Pall Mall clubs, and in the early 20th century on the floors of the New York and London stock exchanges — punctuated by moves to ban them. As the 2024 presidential campaign drives towards a bitter conclusion, the cycle is repeating.
This time the ultimate authority is not the pope but the US Supreme Court. At present, the US’s biggest political betting markets are offshore. The only legal, regulated entities are nonprofits run as academic experiments with tight limits on how much can be wagered — the best known are the Iowa Electronic Markets and PredictIt.com, which operates in the US under the auspices of Victoria University in New Zealand. Now, the Commodities and Futures Trading Commission is fighting an attempt to set up a legal onshore political futures market, which would allow Americans to take part without limits on how much they can trade.
In the latest round, the for-profit Kalshi futures market won a stay against the CFTC’s ban that cleared the way for a huge experiment. The litigation could go on for years, but Kalshi can run a market on this year’s election, and its odds now appear in huge neon lights above Times Square. This has prompted established brokers to get in on the act; Interactive Brokers also launched political futures trading for the election.
The world’s richest man is paying attention. Earlier this month, Elon Musk urged his followers on the online platform X, which he owns, to follow the offshore venue Polymarket (launched by his former colleague Peter Thiel). “More accurate than polls, as actual money is on the line,” he wrote. At the time, Polymarket showed Donald Trump’s chances at 50.1%. Hours later, and with no relevant news, a still-anonymous French investor made the first of a series of giant trades now totaling some $45 million on a Trump win, pushing his odds way up. Ten days later, with Kamala Harris still ahead in polls of polls, Trump’s chances had surged to 62.2%.
PredictIt, populated mostly by academics limited to stakes of $850, also raised Trump’s chances, but at a much slower pace. The press was more interested in Polymarket. Since Musk’s tweet it has featured in 10 times more news stories than PredictIt. Polymarket’s odds helped change the narrative in markets where far more money is at stake. Treasury yields and share prices both rose, as traders started pricing in a Trump win and the corporate tax cuts that would come with it.
The problem of reflexivity
Like all markets, popular prediction markets are prone to what the hedge fund manager George Soros calls “reflexivity.” Rather than just reflecting or reacting to facts on the ground, moves in bonds, stocks or oil markets can actually change that reality. It is this fear that animates opposition to Kalshi. The pressure group Better Markets, an amicus in the CFTC’s suit, asserts that manipulating these markets could shift the facts. Cantrell Dumas, its director of derivatives policy, says that if the markets are taken seriously and show that one candidate is likely to win “people could stay home and not vote for their preferred candidate.”
On the other side, Kalshi’s supporters argue that manipulation is a problem for all markets, and the key is regulation. David Mason, general counsel of PredictIt and a former chairman of the Federal Election Commission argues, “these manipulation problems are dealt with in regulated markets every day. There are trading exclusions; if they have material non-public information then they are prohibited from trading. Kalshi has a long list of people who aren’t allowed to trade.”
But Dumas rejects the comparison. “In a traditional market, if a regulator sees manipulation and takes action, it has remedies. With election manipulation that changes the result, that’s irreparable harm. It’s a fundamental distinction,” he told me.
The excitement since Musk’s tweet looks very much like reflexivity in action. And social change has amplified the potential reflexive power of markets. We are now far more inclined to trust in the “wisdom of crowds.” Most importantly, the Efficient Market Hypothesis, developed in the 1950s and the basis for much modern finance, holds that markets continuously incorporate all information, and so they can’t be consistently beaten, except by insiders who know something others don’t. Central to the argument for investing passively in indexes and exchange-traded funds, this idea can transmogrify into a belief that markets can’t be wrong.
Historically, political betting is impressively accurate. The academics Justin Wolfers and Eric Zitzewitz demonstrated 20 years ago that over history, markets had come closer to getting presidential elections right (with an average error of 1.5%) than the Gallup poll, which was on average off by 2.1%. Market discipline should help experts thrash out a better consensus estimate, and many companies now run internal prediction markets to improve forecasting. Wikipedia and Google have convinced us that crowd-sourcing information can work, and popular culture is saturated with tales of brilliant statisticians, like Michael Lewis’ 2003 book Moneyball on how the Oakland A’s used statistics to spot good players others had missed. Nate Silver, an adviser to Polymarket, made his name as a baseball statistician.
Then there’s what Wall Streeters call the Elaine Garzarelli Effect: We tend to place excessive trust in whichever guru made a good call last time. As a Lehman Brothers investment strategist, Garzarelli brilliantly urged clients to get out of the stock market the week before the Black Monday crash in 1987. That made her one of the most widely quoted strategists on the Street; her words would move markets, and she compiled a good track record. But she was fallible — in December 2007, on the eve of the global financial crisis, she proclaimed that stocks were undervalued by 25% — and people lost interest.
In politics, the offshore Intrade prediction market developed a Garzarelli effect after calling 49 states correctly in 2008, and then all 50 — amid much attention — in 2012. Then regulators shut it down, but other venues rose to replace it. The market odds on a Donald Trump victory were about 35% on election night in 2016.
Betting markets may also have grown more reflexive because the legalization of sports betting has made gambling socially acceptable, while investing looks ever more like gambling. Baseball banned Shoeless Joe Jackson and Pete Rose for life over wagering on the sport. Now, Major League Baseball has its own formal spread-betting partner. Investment apps now resemble video games: Robinhood.com drove the meme stock craze of 2021 as people poured stimulus checks, and a lot of political anger, into bets on companies like GameStop Corp. While papal envoys disdained people betting in the street, our current society has little problem with people trading elections on their phones.
Reflexivity is amplified by a constant of human nature; our psychological need for an anchor. We want a number, and once we have one we cling to it, because it gives us an illusion of control. The idea of anchoring has been a staple of behavioral finance since it was first applied to economics by the psychologists Daniel Kahneman and Amos Tversky 50 years ago. Their experiments showed that people’s answers to numeric questions can be skewed just by showing them a number at random before they answer. In the unmoored climate of a polarized election that appears to be in a dead heat, our desire for an anchor will naturally grow.
In the UK’s Brexit referendum, betting markets put the chances that Britons would vote to leave the EU at one in five. Statisticians David Rothschild and Andrew Gelman argued afterwards that there was “a feedback mechanism whereby the betting-market odds reify themselves.” People distrusted polls, they said, but “we do watch the prediction markets, which all sorts of experts have assured us capture the wisdom of crowds.” Pollsters and pundits — and even the arch-Brexiteer Nigel Farage — were “to some extent anchoring themselves off the prediction odds,” and dismissed polls that showed Brexit in the lead. The clue, Rothschild and Gelman argued, was that a “mature prediction market should show measured, but real, swings in prices in response to news,” but the Brexit odds stayed weirdly stable.
Polls closed with the market odds of Brexit at 31%. When they were proved wrong, the pound dropped by more than 10% in a matter of minutes — more than treble its previous worst fall. Brexit looks very much like a classic example of Soros’ reflexivity at work — and on the face of it, suggests that the CFTC is amply justified in trying to keep political betting out of mainstream investment.
Self-fulfilling prophecies
The CFTC’s problem is that the other markets it regulates exist to help manage risks, from fluctuating commodity prices or interest rates to the possibility of a catastrophic hurricane. Investors wish to insure against political surprises — with the fallout from the Brexit vote showing why they might be sensible. So how to allow that, while limiting the chance of either deliberate manipulation or innocent irrational exuberance which, with reflexivity, could turn markets into a tool for changing election results? There are two plausible solutions.
The first, advocated by Kalshi and Polymarket, is to make markets liquid by attracting more money. With billions at stake, no individual can move the market. But for now, while they remain on a smaller scale, whales can make a big splash.
The alternative is to limit liquidity. PredictIt’s $850 limit on stakes thwarts whales. A rich person’s views don’t count for more just because they’re rich. But it’s harder to move the price when genuine news happens, and it reduces the market’s usefulness as a place to manage risk. Limited markets might produce the public good of more robust predictions, but a multi-trillion dollar manager like BlackRock cannot hedge a given election result on PredictIt.
Recent experiences with opening new markets have shown how difficult this can be. In the decade before the global financial crisis, quants spotted that emerging markets and commodity futures tended to gain over time, with a low correlation to stocks and bonds. That sounded like the holy grail for asset allocators, and big institutions piled in. Once they did so, however, they changed the way these markets behaved and interacted. If money moved into copper as a result of a deliberate decision to move out of bonds, for example, it was inevitable that bonds and copper would grow more correlated, which they did. In 2008, stocks, commodities and emerging markets all crashed together.
Similarly, once big financial hedgers and/or wealthy individuals arrive in betting markets, there’s no reason to assume those markets will keep behaving the same way. Treasury bonds, oil or tech stocks are deep markets combining people with many different motivations, and the messages they send should be taken very seriously; Kalshi, Polymarket and others are far from that scale.
There’s also the eternal issue of Garbage In, Garbage Out. Markets synthesize information and give participants an incentive to sniff out fake news. But they aren’t a protection against bad information — if the polls are wrong, markets will be wrong.
If well regulated, political prediction markets might yet provide probabilities that help guide us in times of uncertainty. But like all markets they will be prone to get things wrong, and while they remain largely unregulated, it behooves us to be careful. The threat of excommunication doesn’t deter miscreants the way it once did.