Prediction markets: money changes everything

I was first introduced to the idea of prediction markets in 1969, in a book by humorist Sam Levenson, originally titled “Sex and the Single Child,” later renamed “A Time for Innocence.” (You see what they did there?) 

In it, a drawing by Whitney Darrow Jr. depicts three little boys in a classroom, one holding a mouse. A nervous girl watches from a safe distance (a familiar stereotype of the day). The caption:

“How can you tell whether it’s a boy mouse or a girl mouse?”

“I know!”

“How?”

“Let’s take a vote.”

OK, the pedantic may point out that the cartoon does not strictly depict a prediction market, although it does offer a prescient view of a country that will, in 50 years, allow people to adopt their own facts, and to act accordingly. Still, it’s kinda cute.

And easier to enjoy than, say, people making money off U.S. military actions.

It may be a long time before we understand why Donald Trump decided to wage war on Iran. Even the timing, while chosen for strategic reasons, was confusing on the surface: a recorded announcement released in the middle of the night on the East Coast — not what you’d expect from the ratings-conscious Trump.

Strategy aside, the timing was important for another reason: A lot of people had a lot of money riding on it. The so-called prediction markets have been doing a lively business around war, among disparate other events.

According to a Reuters review of Polymarket’s website, $529 million “was laid on a series of contracts tied to the timing of attacks, and $150 million was bet on contracts on the removal of Khamenei as Supreme Leader.”

According to NPR, “An account trading under the username “Magamyman” made more than $553,000 placing bets on the prediction market Polymarket that Iran’s Supreme Leader, Ayatollah Ali Khamenei, would be out of power just before an Israeli strike killed him on Saturday.”

An earlier military action also had a handsome payout: An unknown trader on Polymarket doubled down just five hours before U.S. forces swooped into Venezuela and extracted President Nicolas Maduro, raking in more than $400,000 — a 12-fold return on investment.

Kalshi, which was running a market on when Supreme Leader Ayatollah Ali Khamenei would be out of office, initially invoked its policy that death cancels the contract, lest markets incentivize assassination. Participants were irate, some have sued, and the company is trying to resolve the dispute with a significant payout.

In fact, U.S. law forbids trading on war information. Kalshi is based in the United States and subject to Commodity Futures Trading Commission regulations. Polymarket has a U.S. unit, but also operates internationally. Rules are different in overseas markets, and Americans can find ways to participate.

Betting — ahem, trading — on inside information also is illegal, although enforcement appears to have dropped sharply under Trump, and it’s hard to believe the Iran and Venezuela bets on Polymarket were just lucky. The CFTC last year rescinded rules promulgated during the Biden administration that would have prohibited political and sports-related event contracts, with announced plans for new regulations. Companies are allowed to self-certify that they are in compliance with CFTC rules.

The last-minute bets raised many eyebrows, but insider trading is just one of the concerns critics have with prediction markets. Many critics see it as gambling — illegal in some states and highly regulated in others. A number of states dispute the CFTC claim of sole jurisdiction, arguing that prediction markets are indeed gambling and should be regulated as such - by states. Those states have gone to court, and cases are pending.

Advocates insist the business is not gambling, that the difference in the business models is not trivial. They see these markets as a way to introduce broad, crowd-sourced intelligence to a wide variety of decision-making situations.

Wired quoted CFTC Chairman Michael Selig explaining the distinction on Bloomberg’s Odd Lots podcast: “These are not wagers — you’re not betting against the house,” he said. “We have significant overlay from a regulatory standpoint over these markets. And so we’re not gatekeeping particular categories of markets, elections, or sports by having different standards.”

According to Richard Warr, a professor of finance at North Carolina State University:

Prediction markets treat beliefs as tradable financial assets. Before prediction markets, betting on an event typically meant going to Las Vegas or placing a wager with a bookie. Prediction markets, by contrast, allow participants to trade positions as beliefs evolve and to profit from changes in perceived likelihoods. In this sense, they resemble markets for stocks or cryptocurrencies more than casinos.

This characterization matters, as financial theory has long emphasized that markets are powerful information-aggregation mechanisms because they reward accuracy and penalize error. As Friedrich Hayek famously argued, market prices reflect information that is dispersed across many individuals, information that no single person fully possesses.

Prediction markets are not just forecasting tools; they are decision-support tools. In principle, they can improve decisions wherever uncertainty matters. Governments could use them to assess policy outcomes. Firms could use them to forecast demand, project completion timelines, or regulatory risk. Organizations could use them internally to surface uncomfortable truths that hierarchical structures often suppress.

Warr concludes:

Prediction markets are perhaps best understood not as casinos, but as democratized information machines that convert many individual opinions into probabilities through economic incentives. Financial economics repeatedly shows that when people are willing to stake money on their beliefs, whether about the value of a company or the outcome of an election, these markets tend to produce forecasts that are both disciplined and remarkably accurate.

Kalshi CEO Tarek Mansour was born in the United States, grew up in Lebanon, and returned to this country for college. He founded Kalshi with Luana Lopes Lara, whom he met while they both were students at MIT. After finishing school, Mansour went into finance, he told Steven Levy at Wired:

A lot of people were going into finance, which is a very good application of math, so in 2016 I went to work at Goldman. That was my first time coming in touch with the idea of prediction markets. The vast majority of the demand that summer was, will Brexit happen or not? Then it was like, will Trump win the election or not? When Brexit came, it was a massive shock, it completely destabilized markets. The Trump one was even more striking. So I asked, “What if people could price these simple questions about the future?” Markets are an effective, efficient way to average out opinions about something.

Mansour and Lara decided to launch Kalshi as a purveyor of regulated markets under the jurisdiction of the CFTC:

We, as 22-year-old cofounders, said we want to do things right. We want to bring this innovation to America. We believed in this market so much, we were willing to spend four years — or however long it was going to take — to make it happen. We wanted to make it safe and do it responsibly, because we wanted to do it for the long term. The coverage of Kalshi misses that, because all of our competitors didn’t do any of that.

Mansour’s view is that the knowledge generated by these markets benefits democracy. The industry and its advocates like to lead with the math: Prediction markets have brought a new level of reliable knowledge about upcoming events, most conspicuously in elections. They’re more accurate than traditional polling.

That may be their best argument, but for me, it’s a red flag.

Polling is unquestionably a valuable tool for political campaigns, providing data to guide the deployment of limited resources. Polls are so ubiquitous in news organizations’ coverage of politics that we take them — and their value — for granted.

We shouldn’t.

Polling typically asks randomly selected respondents a series of questions of which in many cases they have limited knowledge — and about which they may have rarely thought. Those answers are then assembled into “public opinion,” and that public opinion is packaged by news organizations as content.

Polling now serves as the foundation for much of campaign coverage, but it is journalism devoid of meaningful content. It offers no value to democratic decision-making, but rather is trotted out to draw an audience to a fluctuating but easily accessible story arc.

In her book, “These Truths: A History of the United States,” historian Jill Lepore argues, “The leftward drift of American politics in the 1930s was kept in check by the new businesses of political consulting and public opinion polling, the single most important forces in American democracy since the rise of the party system.”

The 1940s proved to be a bumpy ride for the nascent business. Polls in advance of the 1944 elections had underestimated Democratic strength in two-thirds of the states. Polling for the 1948 presidential election was famously wrong.

According to Lepore, University of Chicago sociologist Herbert Blumer argued the entire premise of the polls was wrong:

“Blumer argued that public opinion does not exist, absent its measurement; pollsters created it: “public opinion consists of what public opinion polls poll.” The very idea that a quantifiable public opinion exists, Blumer argued, rests on a series of false propositions. The opinions held by any given population are not formed as an aggregation of individual opinions, each given equal weight, as pollsters suppose; they are formed, instead, “as a function of a society in operation”; we come to hold and express the opinions that we hold and express in conversation and especially in debate with other people and groups, over time, and different people and groups influence us, and we them, in different degrees.”

So prediction markets are a more accurate forecasting tool because they reflect the views of motivated participants. But that’s not participating in democracy. It’s monetizing arbitrarily chosen events. It’s betting on it.

In rejecting the characterization of prediction markets as gambling, Monsour points to other financial mechanisms that were at one time or another viewed askance. He tells Steven Levy:

“Yes, you are risking money to make money on something you don’t control. That is the definition of a bet. Could you use the word bet to bet on the stock market?”

“Probably,” Levy says.

“When people came out with life insurance, the initial social response was like, ‘You are betting on people’s death?’ Obviously we would agree that life insurance is a good thing in society. Any type of trading activity that involves speculation will have some similarity to markets,” Mansour says.

So it seems like a good time to take a look at these “democratized information machines,” as Warr calls them. Kalshi offers lots of markets. Trending when I looked:

• the government shutdown (when ending).
• the 2028 Republican presidential nominee.
• Department of Homeland Security funding (when)

While Warr argues prediction markets can serve as decision-support tools, nothing in these trending markets on Kalshi helps you understand the government shutdown or the dispute over DHS funding. Kalshi clients may improve the accuracy of predictions, but they don’t contribute to the underlying deliberations. Speculating over the GOP nominee is fanciful … fun, if you like.

And maybe that’s OK. If you iron out the wrinkles in prediction markets, are they really harmful? Aren’t they just entertainment?

Well, as Cyndi Lauper sang, “Money changes everything.” (Written by Tom Gray.)

Whenever you monetize a piece of an information chain, you change the purpose of that link. Wherever money shows up, money has an opinion.

In sports, widespread betting has fragmented the fan experience: Betting on specific, narrow outcomes within the game ultimately changes how we see the game.

If we can bet on anything in our lives, if we can wager on pieces of our public deliberations, how does that change our deliberations? When does the size of the wager change the content of our decisions? If we weigh going to war, how much do we consider the payout by prediction markets in choosing today vs. tomorrow?

The country has a long history of monetizing pieces of the democratic process, pieces that might have been considered a public good. Prediction markets may just be a better version of a bad idea.