How polling failures, gambling legalization and political gridlock paved the way for the explosive rise of prediction markets

Though prediction markets have been legal in the U.S. for less than 18 months, they can’t stop making news and making money^[1].

On prediction markets such as Kalshi and Polymarket, users can stake real money on just about anything, from the winner of the 2028 U.S. presidential election^[2] to when Taylor Swift will get married^[3].

But this isn’t simple entertainment: In theory, these wagers serve as a means of collecting the public’s insights into the future.

That’s why you may have seen CNN’s pundits^[4] casually mention Kalshi’s election odds for the 2026 primaries, or watched CBS offer real-time Polymarket projections of which actors would win awards during the Golden Globes^[5].

Existing research on the principles and history of prediction markets suggests they can be a valuable way of pooling collective knowledge about the future.

But as researchers like me^[6], journalists^[7] and legislators^[8] race to understand the impact these markets are having on society and politics, several questions have emerged about the regulation of these platforms and their forecasting abilities.

The what and why of prediction markets

In practice, prediction markets are quite simple.

Each market offers what are known as “event contracts” on whether some future outcome will occur. Each contract costs between 1 and 99 cents, paying out US$1 if the event occurs or nothing if it does not.

Similar to sports betting, purchasing a contract represents a wager. There are higher returns for positions on outcomes deemed less likely. Like in the stock market, a trader can buy and sell contracts over time, as odds – and thus prices – fluctuate.

At the time of writing, Kalshi traders put the odds of the passage of the SAVE Act^[9], legislation centered on requiring proof of U.S. citizenship to register to vote, at about 10%^[10]. So each contract for this outcome costs 10 cents. If I think the act is more likely to pass than that, I could purchase some “shares” and sell them at a higher price if the odds go up in the future. If I hold them and the bill ultimately becomes law, I would receive a return that’s 10 times what I originally paid.

Two theories support the idea that prediction markets should excel at forecasting: the wisdom of crowds and the efficient market hypothesis.

First described over a century ago^[12], the wisdom of crowds^[13] refers to the idea that the median judgment of a large, diverse group of people operating independently is often more accurate than that of a single expert.

A related argument appears in the efficient market hypothesis, which emerged in the mid-20th century among economists who championed free markets^[14]. It holds that prices encode all available information, reflecting the collective judgments of profit-seeking sellers and deal-seeking buyers.

At their best, then, prediction markets aggregate collective intelligence to weigh the likelihood of future events.

Polling’s credibility crisis creates an opening

Gambling on the outcomes of the day’s events has a long history. In 16th-century Italy^[15], gamblers could wager on the election of civic magistrates and the outcome of papal conclaves. And from the 1880s to 1930s^[16], New York City was the hub of political wagering, which sometimes exceeded the stock market in daily volume.

Reporting on bets ahead of the 1924 presidential election, The New York Times observed^[17], “It is an old axiom in the financial district that Wall Street betting odds are ‘never wrong.’”

However, the rise of scientific polling^[18] and legal crackdowns on political wagering^[19] forced prediction markets to fade to the background.

That changed in 2024.

One month before the U.S. elections, a federal court granted the prediction market startup Kalshi permission to legally operate prediction markets^[20] concerning U.S. election results.

Around the same time, Elon Musk posted^[21] on X about Donald Trump leading Kamala Harris in prediction market odds. Trump followed suit^[22]. Kalshi put up billboards with live election odds in Times Square. Users and dollars flowed in. By election day, a volume of over $500 million in presidential election bets^[23] had been traded on Kalshi alone. Polymarket^[24] featured over $3.6 billion more in volume.

Political polling, meanwhile, was facing a crisis of confidence. Response rates had been declining^[25] for decades, and Trump voters had been undercounted in 2016 and 2020.

The polls forecast^[26] the presidential election as a coin toss. The prediction market, meanwhile, favored Trump at roughly 60% odds to win.

After Trump won at the ballot box, prediction markets declared victory over polling^[27] as the new, trustworthy forecasters of public opinion.

The utility of the markets

Over the past 50 years, journalists have increasingly^[28] incorporated quantitative data in their reporting, and audiences have come to expect political forecasting^[29] as part of their news diet.

With polling experiencing a crisis of confidence^[30], prediction markets have become an increasingly attractive way for journalists to offer a data-backed snapshot of public beliefs.

Prediction markets have other advantages over polls for journalists. They respond to events in real time, and they’re free to access. Polls, meanwhile, take time and money to administer. They provide forecasts for political outcomes that go beyond elections – such as Cabinet nominations^[31] and Supreme Court decisions^[32] – which are usually outside the purview of polling.

In recent months, Kalshi and Polymarket have inked several^[33] partnership^[34] deals^[35] with news outlets. There’s a symbiotic relationship at play: Prediction markets provide journalists with data to report and discuss. Journalists, in turn, legitimize prediction markets^[36] by citing them as a trusted source.