He Took Prediction Markets Mainstream

For decades, prediction markets were a backwater, a science experiment.

In 2024, Shayne Coplan, founder of Polymarket, turned them into a multibillion-dollar business and a popular barometer of the political winds, cited by everyone from Donald Trump to CNN.

In so doing, he demonstrated a real-world consumer use case for cryptocurrency – and, some argue, a new model for news media at a time when the public has lost trust in traditional sources of information.

“Most people I know were checking Polymarket for odds during the election,” said Meltem Demirors, a crypto O.G. and early investor in the company. “You’re creating so much signal that you’re getting people who don’t care about crypto, and would never care about crypto” to look at the site.

Like many crypto founders – and even some successful tech founders – the 26-year-old Coplan also took what looks like a calculated risk in pushing the regulatory envelope. In mid-November, the FBI raided his New York home and confiscated his devices, reportedly as part of a Department of Justice investigation into whether Polymarket was operating illegally in the U.S. Coplan has laid low since then, and would not comment for this article.

However that investigation shakes out, Coplan has brought unprecedented attention to an idea long advanced by academics: That the wisdom of the crowd, backed by skin in the game, can produce more accurate forecasts – or at least, more accurate gauges of sentiment – than traditional experts or polls.

“This man made prediction markets mainstream. Simple as that,” said Hart Lambur, co-founder of UMA, the decentralized oracle service that Polymarket uses to resolve contracts. “He’s just been the guy that’s grinded through the pain and been dedicated to the Polymarket concept for years.”

A stubborn wunderkind

Demirors recalls meeting Coplan in 2018, when the college dropout was about 18 years old, on the recommendation of a crypto colleague.

“Shayne came to my office, and we basically just argued with each other for two hours,” Demirors said. “I was like, ‘wow, this kid is sharp.'”

Pratik Chougule, executive director of the Coalition for Political Forecasting, got a similar impression interviewing Coplan for the Star Spangled Gamblers podcast early in Polymarket’s history.

“He’s a very unique figure in the sense that he’s this creative artist type, but he’s also delved deeply into academic literature, and he really understands technicalities of building something on the blockchain,” said Chougule.

Demirors said that in addition to investing in an early Polymarket round during the pandemic, she has been “a little bit of a big sis” to Coplan, acting as a sounding board as he built the business.

“He’s just an opinionated, stubborn little f*ck, and I love him,” she said, adding that Coplan’s headstrong personality served him well as a founder.

Early on, “people tried to pressure him to launch a token, and he was like, ‘we’re not doing that.’ People tried to pressure him to open up markets before the infrastructure was ready. He was like, ‘we’re not doing that.'”

Volume and vindication

Flip Pidot, a veteran prediction market trader and analyst, estimated that Polymarket racked up $3.6 billion in trading volume just from this year’s U.S. presidential election, giving it a dominant, 74% market share. In previous election cycles, the entire prediction market industry never cracked $1 billion, he said.

Many saw the election as a moment of vindication for Polymarket. In the weeks leading up to the event, Polymarket odds signaled a sizable lead for Trump while the polls showed a toss-up between the former president and his Democratic opponent, Vice President Kamala Harris. Trump won handily.

Read more: Polymarket ‘Manipulation’ Claims Miss the Mark

Yet a clearer validation of Polymarket’s informational value arguably came in July, when President Joe Biden dropped out of the race and endorsed Harris.

For months, cable news’ talking heads dismissed any talk of replacing Biden on the Democratic ticket, despite the 82-year-old’s frequent public stumbles.

Polymarket told a different story: Even after Biden won enough votes to clinch the Democratic nomination in mid-March, traders gave him only an 80% chance of being the nominee. A separate contract asking point blank if he would drop out gave low but nontrivial odds in the teens and 20s throughout the first half of the year.

“People were like, ‘Oh, these [traders] are right-wing crypto bros, they’re just conspiracy theorists. They don’t know what’s going on,'” said a Polymarket user who goes by the handle CSPTrading. “And they were completely vindicated.”

Following Biden’s disastrous, doddering performance in the June 27 debate with Trump, the narrative quickly changed, with Democratic leaders and donors calling for the incumbent to step aside, as he did a month later.

More so than with the election, the pundits (who had nothing to lose from being wrong) got it wrong by claiming epistemic certainty. Polymarket’s traders (who had money on the line) got it right by telegraphing a modicum of doubt.

Spectrum of decentralization

In prediction markets, traders bet on verifiable outcomes of events in specified timeframes. (Which movie will gross the biggest box office of 2024? Will this be the hottest year on record?) Questions are usually framed as yes-or-no propositions, for which traders can purchase “yes” or “no” shares. Each share pays $1 (or, in Polymarket’s case, the equivalent in crypto) if the prediction comes true, bupkis if not.

Bettors can buy and sell shares any time, and prices fluctuate like on stock markets. Expressed as cents on the dollar, these prices signal the market’s assessment of an outcome’s probability. On Dec. 4, for example, “yes” shares for the Detroit Lions winning the next Super Bowl traded at 18 cents on Polymarket, meaning bettors gave the team an 18% chance of victory. The corresponding “no” shares were priced at 82 cents.

Prediction markets date back to the late 19th Century, when Wall Street traders would bet millions (tens of millions in today’s dollars) on city, state and national elections. “There was more money bet in presidential betting markets than in the stock markets at the time,” said Robin Hanson, an economist at George Mason University.

Since the late-1980s, Hanson has championed prediction markets as a way to aggregate information and thereby improve decision making by corporations and even governments.

“One of the obstacles, of course, was that betting markets had many legal barriers, and cultural barriers [because] many people disapproved of them and thought they had little social value,” Hanson told CoinDesk.

This is one reason why blockchains, decentralized financial systems with no central authority that a government can shut down, have long been seen as a natural home for prediction markets. They are one of the use cases Ethereum architect Vitalik Buterin described in his 2014 white paper for what would become the second-largest blockchain. (As a teenager, Coplan bought into the Ethereum crowdsale; a decade later, Buterin invested in Polymarket.)

The modern-day prediction markets Hanson inspired can be viewed on a spectrum. On one end there’s the model used by Augur, one of the first projects built on Ethereum.

“One of the advantages is that it’s 100% decentralized,” said Joey Krug, who co-founded Augur in 2015. “If you’re building it, you’re effectively writing code. It’s effectively free speech, assuming you’re not taking a fee for yourself, and it’s also pretty flexible in the sense that anyone can kind of create a market on anything.”

But as crypto veterans know all too well, decentralization requires trade-offs.

Best of both worlds?

“It’s really hard to market if you’re building something decentralized,” said Krug, who is now a partner at Peter Thiel’s Founders Fund and led its investment in Polymarket’s $45 million Series B round.

(For whatever it’s worth: Thiel was an early investor in Bullish, two years before that company acquired CoinDesk. Bullish has not disclosed a cap table since 2021, and CoinDesk journalists do not know the current roster of investors in its parent.)

“The whole point is that you don’t want to take on the regulatory version of being this central operator that does everything,” Krug said. “And so you don’t really market it. … You don’t do all this stuff that you need to do to actually get usage.”

Consequently, Augur had very little. (In fairness, Polymarket benefits from Ethereum infrastructure that wasn’t around when Augur debuted).

On the “very centralized” end of the continuum, there’s Kalshi. Founded in 2018, the startup boasts about its status as the first (and, until recently, only) regulated prediction market platform in the U.S.

This route has its own disadvantages. In 2023, the Commodity Futures Trading Commission denied Kalshi’s application to list election-related contracts, and the company spent most of this year fighting the regulator in court for the right to do so – while watching Polymarket enjoy the volume and publicity from political betting fever. Only after an appeals court upheld a ruling in its favor in early October, a month before the election, was Kalshi cleared to list political contracts.

Polymarket is in the middle of the spectrum. In some ways, it’s decentralized. It uses smart contracts on a blockchain (Polygon, a layer-two, or auxiliary network, to Ethereum) and doesn’t custody users’ funds. Bets are denominated in USDC, a stablecoin that trades 1:1 for dollars. Early on, an internal market integrity committee resolved Polymarket’s contracts, before Coplan’s team delegated this job to the decentralized UMA protocol.

“If you are sufficiently sophisticated, you can interact entirely with Polymarket without ever touching the website,” said Haseeb Qureshi, a managing partner at Dragonfly, another VC investor in Polymarket. “The trades settle all on-chain. You can interact with everything through APIs.”

But you don’t have to. Unlike Augur (which co-founder Krug admitted “kind of sucks to use”) or for that matter many crypto exchanges (decentralized or otherwise), traders have found Polymarket easy to use and reliable.

“The platform’s really smooth, it runs really well,” said CSPTrading. “On election night, it was basically up the entire time, which is crazy because… all the other sites were crashing.”

‘Decentralized enough’

One way Polymarket is centralized is that it curates markets. Community members can suggest ideas in the Discord server, but the team decides which ones get posted. With little fanfare, the platform recently debuted a “creators” page where big names like polling analyst Nate Silver (a Polymarket advisor) and the financial blogger Zerohedge have their own branded markets.

“I think Polymarket is moving its way towards more decentralization,” said Qureshi. “They’re also right to be doing this in a gradual, thoughtful way, rather than just turning everything on and saying, ‘let the dogs of hell run loose.'”

In Demirors’ view, Polymarket is “decentralized enough.” The key to winning this game, she said, is amassing “a large enough global pool of market participants,” because traders want to be where the liquidity is. By building on crypto rails at the right time, that’s what Polymarket has become.

“That’s the beauty of crypto. It’s global. Anyone with a wallet address can join,” Demirors said.

However, Polymarket wasn’t decentralized enough for U.S. regulators to consider it untouchable. In January 2022, the company paid a $1.4 million civil penalty and entered into a settlement with the CFTC, which said the company had been operating an unlicensed derivatives exchange because its services were available to U.S. citizens and residents.

Since then, the company has blocked U.S. IP addresses, but wily Americans have been using virtual private networks, or VPNs, to get around the geofencing. Apparently, the government thinks the company should have done more to keep Americans out, perhaps by requiring customer identification. (which Polymarket has requested only from a subset of users).

“Polymarket is required to adhere to the terms of the settlement they reached with the CFTC. Full stop,” a CFTC spokesperson told CoinDesk in late October, two weeks before law enforcement officials raided Coplan’s home. “That means they cannot accept any business from people living in the United States.”

In a post on X (formerly Twitter), Coplan called the raid a “last-ditch effort” by the lame-duck Biden administration “to go after companies they deem to be associated with political opponents,” though he reiterated that Polymarket is nonpartisan.

Challenges ahead

Polymarket’s investors and supporters are hopeful the incoming Trump administration will end the probe as part of a broad pro-crypto agenda.

Even if Polymarket receives clemency, Coplan faces other challenges, not least of all maintaining volumes without a galvanizing tent-pole event like a presidential election.

The company, which currently doesn’t charge trading fees, also must figure out a long-term revenue model. And a handful of outcome disputes, including for a market on whether Trump’s son Barron was “involved” in a memecoin, suggest Polymarket needs to improve its resolution criteria.

Yet, by at least one measure, Coplan has already succeeded.

“Shayne’s vision has always been that this is a product that can disrupt traditional media and political discourse … and he achieved that” said Chougule, at the Coalition for Political Forecasting. “This was always the dream, that you would have major talk shows, cable news, places like Politico and Bloomberg citing prediction markets as a source of information, as something that can enlighten even people who know nothing or don’t care about prediction markets.”