The billion-dollar prediction markets industry began like a joke: Three economists walked into a bar.
This was in Iowa City in 1988, long before anyone could bet on elections or Super Bowl halftime shows with their phone. The professors were trying to solve the sort of problem social scientists tackle over lunch. Why did polls get elections so wrong — and what could be done about it?
Civil rights leader Jesse Jackson had just trounced Michael Dukakis in the Michigan presidential primary — despite polls predicting Dukakis would win easily. Professor Robert Forsythe and two of his University of Iowa colleagues, George Neumann and Forrest Nelson, wondered how economics might do a better job than traditional polls.
It doesn’t take a Ph.D. to guess the punchline here. “Well, we would make a market off of it,” Forsythe recalled them concluding.
By the time they finished their meal, the trio had settled on an idea: a marketplace for betting on the outcome of events that could serve as a tool for forecasters. “It’s amazing after a three-beer lunch what you come up with,” he said.
The economists built a prediction market at the University of Iowa as an academic experiment to test whether their idea that betting markets could provide useful information about future events would even work. It did. The Iowa Electronic Markets helped plant the seed for the modern-day prediction market companies that have ballooned into a multibillion-dollar industry, offering contracts on everything from the Super Bowl to the number of times Elon Musk will post on X in early February.
Forsythe sees Kalshi and Polymarket as direct descendants of his academic experiment, and he has been gratified to see the ideas that he and his colleagues spent decades working on now operating at such a massive scale: Kalshi saw more than $1 billion in trading on Super Bowl Sunday alone. “It’s good to see markets having the kind of success they are — we really launched something,” Forsythe said, adding that the success of companies running with their idea has also stirred “a bit of jealousy.”
The original market in November 1988 was a modest enterprise. They stuck the server in an empty faculty office that was so small it barely had room for two chairs.
They called it the “Iowa Political Stock Market” and recruited university staff members and students to buy and sell contracts on how likely Dukakis was to prevail against Republican George H.W. Bush. At the time, most students didn’t even have personal computers to make trades, so they went to student computer labs — destinations that became outright crowded on Election Day.
Forsythe remembers waiting anxiously with his collaborators for the returns to come in, watching two separate television sets at Forrest Nelson’s house. “At one point, we were pretty much dead on,” he said, with one of them exclaiming: “Oh God — it better not change very much.”
They didn’t have to worry: Even with a tiny pool of less than 200 traders, the market was shockingly accurate. Its prices correctly indicated that Bush would get exactly 53.2% of the popular vote. The market’s forecast for Dukakis’ vote share was 45.2% — just 0.2 percentage points below the actual total of 45.4%.
While many of the early traders were political junkies who followed every campaign development, others were students who got more interested once they had “a few dollars on the line,” giving them a financial incentive to seek out new, relevant information that could affect the future outcome, Forsythe said.
The biggest profit in that first market? $13.54, on a $250 investment.
By the next presidential election, the operation had gone national as the “Iowa Electronic Markets” with the blessing of the federal Commodity Futures Trading Commission — but only with the understanding that it would remain an operation that ran for “solely academic and experimental purposes,” with no outside advertising, and individual traders couldn’t invest more than $500. The Iowa prediction markets continued to outperform many leading polls, especially during the run-up to an election.
The experiment caused a stir among economists and political scientists and people in finance and business circles. While betting on elections had been around for centuries — including in the U.S. during the late 19th and early 20th centuries — the idea of turning such a market into a modern financial exchange opened up a universe of possibilities.
The Iowa experiment inspired a host of early efforts to create electronic prediction markets in the private sector, and even the Defense Department, that went well beyond elections. But most of the new entrants, such as Intrade and Tradesports, have floundered after running afoul of federal regulators and federal laws prohibiting internet and sports gambling.
“There were a lot of people who knew immediately what it was all about and were excited to have the opportunity to trade on the markets,” said Eric Zitzewitz, an economics professor at Dartmouth who has studied prediction markets for decades. “Part of the reason it was at universities first is because it was illegal everywhere else.”
Forsythe said companies wanting to take their prediction market overseas approached him and colleagues, given the restrictions in the U.S. “In hindsight it probably would have been a smart thing to do,” he said. George Neumann died in 2015 and Forrest Nelson in 2023.
After Polymarket was founded in 2020 and Kalshi the following year, both companies faced fierce pushback from federal regulators during the Biden administration. But a federal court in September 2024 ruled in the industry’s favor.
President Donald Trump’s return to office in January 2025 ushered in a new era of regulators who have opened their arms to commercial prediction markets. It didn’t hurt when the president’s son Donald Trump Jr. became an investor in Polymarket and a paid adviser to Kalshi.
Both companies have grown exponentially since then, offering the opportunity to bet on events far beyond the original scope of the academic experiment. This month, for example, Kalshi traders were able to bet not only on the outcome of the Super Bowl, but also which song Bad Bunny would sing first during the halftime show and whether the game’s announcers would mention the word “Gatorad.”
“While the IEM was quite different from Kalshi in liquidity and scale, it did provide something critical to the history of prediction markets: it helped prove that they are accurate,” Jack Such, a Kalshi spokesman, said in a statement.
Commercial prediction markets have resisted efforts by states to classify and regulate them as gambling companies, insisting that they — like their precursor in Iowa — should be overseen by federal regulators. But there are no federal rules specifically designed for prediction markets to prevent market manipulation or protect consumers.
That has raised some concerns among longtime prediction market experts and researchers.
“It’s very easy to get into these markets and spend a lot of money and lose a lot of money, and so it is worrisome,” said Thomas Gruca, a professor of marketing at the University of Iowa and the current director of the Iowa Electronic Markets, which has been in continuous operation since its founding. “There are only so many people who actually have good information. Then you have a lot of people with opinions and money.”
Forsythe agrees that some guardrails are necessary as for-profit prediction markets have taken off, particularly when it comes to issues like market manipulation. The bigger and more influential a market becomes, the more tempting it could be for certain actors to try to manipulate the prices to inflate the chances of a political candidate, for instance. “You need some kind of oversight in terms of fair reporting, market manipulation and so forth,” he said. (Kalshi recently detailed its support for federal regulation to protect prediction market integrity and trust.)
And though Kalshi and Polymarket made headlines for correctly predicting that Trump would comfortably win the 2024 election, some research studies have also shown that prediction markets tend to be less accurate on speculative questions for which there is little public information to help traders make forecasts.
Still, Forsythe believes that prediction markets, both old and new, can still offer real value in predicting events with major, real-world consequences. And he always expected the industry to take off once it had access to a broad commercial market. “The only thing I’m surprised about is it took so long to get there,” he said.
Leave a Reply