Forecasting the Future of Election Prediction Markets

I’ve used prediction markets for years, never for trading but rather as a source of information, an interesting adjunct to polls, economic and political models, and traditional reporting, especially when elections grow near.

But the U.S. prediction markets that allow people to place legal bets on American elections have run into regulatory problems.

PredictIt, the larger of the two prediction markets operating in the United States, has emerged over the last several years as a go-to source for journalists and academics seeking to harness the “wisdom of crowds” for a sense of where the elections are heading.

At the moment, the bets on its site amount to a forecast of Republican control of both the House and the Senate. PredictIt’s older and more purely academic counterpart, the Iowa Electronic Markets, is showing the same essential picture.

Kalshi, a third and overtly commercial derivatives market, had hoped to start trading contracts on the outcome of the midterm elections by now, but its application has stalled at the Commodities Futures Trading Commission.

The various markets — PredictIt, the Iowa market, overseas markets like Betfair in Britain and predecessors like the Irish sites Intrade and Tradesports — have had plenty of glitches through the years. They aren’t always accurate, and their results, if not interpreted carefully, look deceptively extreme. A thin edge in election polling is often magnified in these markets as a definitive advantage, but they are often fundamentally correct. Academic studies have found that they stack up quite well against opinion polls and standard political forecasts.

Such markets have also provided compelling results in experiments aimed at estimating Hollywood movie box office results, improving weather forecasts and providing corporations and the Defense Department with information on security, health care and product quality control issues.

They all work on the basic notion that when markets are broad and efficient enough and money is at stake, the collective information embodied in their prices is closer to the truth than the conclusions that most individuals can come up with on their own.

This may well be true over the long run, but one flaw in this approach is obvious if you have been following the stock, bond, foreign exchange or commodity markets this year: Over short periods, markets are as fickle as a cat. I see no reason to assume that election markets are inherently wiser or steadier than the stock market, which I don’t trust as a guide to much of anything over short periods.

Still, prediction markets are fascinating, and provide a worthwhile source of data on a vast array of subjects. They also have tremendous financial possibilities.

“Prediction markets are going to be a very big business,” said William R. Hambrecht, an early investor in Silicon Valley companies like Apple and Adobe and a major investor in Aristotle, the company that runs PredictIt. “I don’t think you can overstate the potential, once the regulatory issues are resolved.”

An Old American Pastime

Betting on elections is, technically, illegal in the United States. But the reality is that it has been commonplace since the early days of the republic, Koleman Strumpf, a Wake Forest University economist, said in an interview.

By the late 19th century, New York City had become a national center for betting on both finance and politics. Early in the 20th century, whenever elections rolled around, traders outside the New York Stock Exchange placed high-volume wagers on the Curb Market, a rollicking, over-the-counter outdoor marketplace for stocks, bonds and politics.

In the current era, the Commodities Futures Trading Commission has permitted the two low-stakes markets — the Iowa Electronic Markets and PredictIt — to operate under academic exemptions.

But in August, the commission ordered PredictIt to cease operations by Feb. 15. The agency hasn’t publicly explained its reasons for the cancellation, which PredictIt is fighting in the courts. One issue may be PredictIt’s popularity — the volume of betting on the site has sometimes exceeded the limits set when PredictIt began operating as an educational venture in 2014.

A week ago, the commission staff recommended against Kalshi’s bid to start a higher-stakes trade in futures contracts on the control of Congress resulting from the midterm elections. The commission tends to side with its staff’s recommendations, but has not indicated how it will rule on the Kalshi case.

In an email, Steven Adamske, a commission spokesman, said only: “Kalshi’s application is still pending, and I don’t have a timeline for when it will be announced.”

In an interview, Tarek Mansour, a founder of Kalshi, pointed out that his exchange already offered robust trading on important questions like the future of inflation in the United States and the path of Federal Reserve interest rate increases, and that it was running an unofficial contest asking people to forecast the elections.

“That’s an effort at educating people about our site,” he said. “We agree that we need regulation, and we are waiting patiently for guidance from the C.F.T.C.”

The Two Election Markets

It is possible that once the midterms are over, only the Iowa market will remain as a legal outlet for election betting. It has functioned since 1988 as a modest, money-losing “Internet-based teaching and research tool” that allows up to 2,000 people — students of the University of Iowa and anyone else with the money and interest — to place bets of $5 to $500 on the outcome of events, including U.S. elections.

As long as it sticks to these unequivocally educational goals and people at the university are willing to take the time and effort to keep it running, its future seems reasonably secure.

“It’s a labor of love,” said Thomas S. Gruca, the director of the prediction market and a professor at the university’s Tippie College of Business. “It takes a lot of time from a lot of volunteers to keep it going. But a lot of people are learning a lot because of it.”

PredictIt is something of a hybrid. It is a joint venture. One partner is Victoria University of Wellington, a New Zealand institution. The other is Aristotle, a for-profit American political consulting, compliance, data and software company, whose founder, John Aristotle Phillips, first gained national attention in 1976 as “the A-bomb kid” — a Princeton undergraduate who successfully designed an atomic bomb in a physics class project.

“It was about arms control,” he told me in an interview. “I showed that a bomb could be built, and that we needed more controls.”

Aristotle does the day-to-day work running PredictIt, and the university has been playing a passive role. While the data from the prediction market at the University of Iowa is regularly used in classrooms and in research there, that is not the case for Victoria University.

“We are not aware of any scholars at Victoria University of Wellington using the data, but, as they don’t need to come through us to access it, that would be a question better directed to PredictIt,” Katherine Edmond, the university’s director of communications, said in an email.

PredictIt is used extensively by scholars around the world, but mainly by American scholars, who are listed on its site and have relied on it for years, as have journalists like me. In addition to filing suit, Aristotle has applied for permission to become a commercial exchange, like Kalshi, a move that would end the restrictions on its size and scope.

A Modest Prediction

From the standpoint of many economists, the prediction markets, for all their flaws, have been spectacularly successful.

“There’s tremendous social utility to having these markets operate, and having this information available,” said Eric Zitzewitz, a Dartmouth professor who has studied prediction markets extensively. “There’s a lot of demand for them — people enjoy participating in them and consuming the information they provide.”

Betting on elections won’t go away, no matter what the regulators decide. Such betting could migrate to overseas markets or to unregulated markets in cyberspace that are outside U.S. regulatory control.

Far better, I think, would be to allow them to operate within U.S. borders as transparent, robust — and carefully regulated — operations. Until now, both the Iowa market and PredictIt have been small enough to be fairly innocuous in terms of their effects on elections themselves. When big money flows into prediction markets — as I suspect it one day will — the potential for real trouble will be far greater. Markets can be manipulated and corruption flourish, so smart and active regulators are needed to keep markets honest.

That’s true for the stock market. Regulating U.S. prediction markets will become far more critical when large sums are focused directly on the outcome of elections, which are, after all, the foundation of our democracy. Prediction markets are important enough to be preserved. But elections are important enough for regulators to move slowly and carefully.