Every election season, two very different tools are competing to tell Americans who will win: opinion polls and prediction markets. Each has a long track record, and each has stumbled. "We don't set the odds; the people do," David Piegaro of Kalshi, one of the largest U.S. prediction markets, told Just the News. 

"Our markets are super dynamic as new information comes out, whether that be a poll, whether that be an event in the real world." Each has a long track record, and each has stumbled.

How each tool works and how accurate they are

Polls ask a sample of voters about their preferences at a given moment. By contrast, prediction markets let traders buy contracts that pay if an event happens, so a $1 dollar contract trading at 60 cents implies a 60% chance. 

Polls survey a representative sample of voters and ask about their personal preferences or intentions at a specific point in time. On the other hand, prediction markets work like financial exchanges where participants buy and sell contracts based on the likely outcome of an event. Contract prices fluctuate based on demand, reflecting the collective probability estimated by traders. 

In short, polls measure opinion, while markets put money on outcome probability.

The best-known long-term test comes from the University of Iowa. Researchers compared Iowa Electronic Markets Forecasts with 964 national polls from the 1988 through 2004 presidential elections and found the market was closer to the final result 74% of the time. Its edge was largest more than 100 days before Election Day.

The 2024 presidential race gave markets a high-profile win. On election morning, Kalshi priced a Donald Trump victory at 57 cents and Polymarket at 62 cents, while forecasts built on polling averages showed a near tie.

The record is not one-sided, however. A Vanderbilt University study of political markets during the final five weeks of the 2024 race found that similar contracts on different platforms were often priced differently and moved independently, suggesting traders sometimes reacted to one another rather than to news.

A long history for both

Betting on elections is surprisingly older than modern scientific polling. In 2004, economists Paul Rhode and Koleman Strumpf documented organized presidential betting markets centered on Wall Street from 1868 to 1940. In the 15 elections from 1884 to 1940, the mid-October betting favorite won 11 times, and the underdog won only once in 1916. Those markets faded as scientific polls spread and other forms of gambling became legal.

Polling has roots nearly as deep. The Harrisburg Pennsylvanian ran an informal straw poll in the 1824 presidential race. Straw polls peaked with the Literary Digest, which in 1936 mailed more than 10 million postcards, got about 2.4 million back and wrongly forecast that Alf Landon would beat Franklin D. Roosevelt. Pollsters using more scientific sampling, including George Gallup, called the race correctly and set the template for modern surveys.

Markets returned in 1988, when the University of Iowa launched the Iowa Electronic Markets as a small, real-money research project. In October 2024, a federal appeals court cleared Kalshi to resume offering contracts on which party would control Congress, opening the door to regulated election trading in the U.S.

Weak spots and a shared role

Each method has known flaws. Pollsters struggle to reach people: the response rate for Pew Research Center telephone polls fell from 36% in 1997 to 6% in 2018, according to NBC News. Pew says that low response raises the risk of error and drives up costs, though its own studies found little link between response rates and accuracy.

Unlike polls, markets can be moved by a few large traders, called "whales" in Wall Street parlance, because their sheer size can change the ocean's waveform. In 2024, a French trader known as Theo placed about $42 million in bets on a Trump win across four Polymarket accounts and at one point held more than 20% of the "yes" shares. Markets also answer a much narrower question, pricing who will win but saying little about why voters decide as they do.

That is why many analysts treat the two as complements rather than rivals. Polls supply detail about what voters think and why, and traders fold that information into a single price. Both face their next public test in the midterm elections on November 3.