Pre-election polls are often referenced in the build-up to major elections, offering a flavor of the results that might be on the way. However, every political commentator will be quick to say they are not predictions – and indeed they are often proved wrong when the results actually roll in.

In both the 2016 and 2020 US presidential elections, polls underestimated the support for Donald Trump in key battleground states. While there was some improvement in accuracy in the 2024 election, with most major polls estimating that it would be a close run between Kamala Harris and Trump, the majority were leaning more towards Harris than Trump.

Enter: prediction markets

By comparison, prediction markets that were offering political contracts in the run-up to the election were offering more accurate odds. For example, Polymarket offered 60/40 odds on the Trump/Harris voting split, which, while overstating Trump’s Popular Vote margin, was impressively close when it comes to the Electoral College result.

Trump’s odds are up 2.5% after Kamala Harris’ Fox News interview.

He now has a 24% lead. pic.twitter.com/7YFOFzwrfu

— Polymarket (@Polymarket) October 17, 2024

Could prediction markets therefore provide a more accurate snapshot of public mood and intentions? Prediction markets like Polymarket certainly do need to base their data and odds on a wealth of information to ensure accuracy and tempting bets – but that doesn’t mean they’re infallible.

“If anything, we have reason to believe that participants in prediction markets are somewhat atypical as, for example, they all would seem to enjoy at least this one form of gambling.” Michael Montgomery, political scientist and former US diplomat

“The short answer is maybe yes, maybe no,” Michael Montgomery, political scientist and former US diplomat, told Readwrite.

“Participants in prediction markets are entirely self-selected. As a result, we have no reason to believe they constitute an adequate sample of any larger group – such as voters – for which we might want to make nuanced predictions of future behavior.”

Fans of prediction markets often argue that incentives, liquidity, and the “wisdom of crowds” help them outperform traditional polls. When people have money at stake, they tend to pay closer attention, react quickly to new information, and correct bad assumptions.

In active markets, all those individual judgments get distilled into a single price reflecting the crowd’s collective expectation. It’s not foolproof, but it helps explain why prediction markets sometimes edge out polls.

For opinion polls, on the other hand, pollsters for presidential elections, as one example, take deliberate care to poll a range of adults within voting age, spanning different regions of the country. The goal is to source data from a varied cross-section of the country. The same cannot truly be said for prediction markets.

“If anything, we have reason to believe that participants in prediction markets are somewhat atypical as, for example, they all would seem to enjoy at least this one form of gambling,” noted Montgomery. “At least from a classic social science perspective,  prediction markets don’t seem to have much to recommend them.”

Research comparing prediction markets and political polls shows that both tend to move in the same direction, but markets typically react faster to new information. The study from a student at the University of Arizona found that prediction markets adjusted immediately to debates and primary results, while polls, which are updated more slowly and shaped by question wording and sampling decisions, took longer to reflect those shifts.

Both methods ultimately identified the winner of the 2020 Democratic nomination, though prediction markets did so earlier and by larger margins. The study also notes that markets carry their own quirks, such as traders clustering around certain “reference prices,” which can distort accuracy. In practice, polls offered structured snapshots of voter sentiment, while prediction markets acted as real-time aggregators of scattered insight, giving each a different role in political forecasting.

Potential for abuse

A recent example of a playful comment from Coinbase CEO Brian Armstrong exposed another issue: the potential to abuse prediction markets by the players the contracts center on. Armstrong made a quip about prediction markets taking bets on what terms he would say during the company’s quarterly earnings call, before quickly rattling them off.

While the comment was clearly made in fun, it did show how easy it would be to manipulate prediction markets results. If a CEO could do it, couldn’t a major political player too?

The Crown Prince of Saudi Arabia just referenced the polymarket about him wearing a suit & tie.

Simulation confirmed?pic.twitter.com/jPHJpDkzXK

— Polymarket (@Polymarket) November 19, 2025

Low-liquidity prediction markets are also much easier to sway. If it takes only a small amount of money to move prices, one well-timed trade or even a pointed public comment from a political figure can nudge the odds in a certain direction. In those moments, the market isn’t really showing what people believe will happen. It’s showing what a few motivated players want it to look like.

You would hope that the stakes of a US presidential election would be enough to not encourage candidates to manipulate the results just to win a prediction market contract, but it’s feasible that smaller elections could tempt people to pull out of races or otherwise manipulate results for financial gain.

Where prediction markets fit alongside traditional political polls

That’s not to say that there’s no space for prediction markets to play a role moving forward, however. Media outlets increasingly cite market odds to show public opinion, and the accuracy of Polymarket’s odds ahead of the 2025 US presidential election suggests that there are some ways that prediction markets can succeed where opinion polls do not.

“What we may be seeing is prediction markets functioning in the manner of an informal ‘big data’ application that aggregates the predictions of tens of thousands of observers with different perspectives and methods and then spits out a single prediction,” explained Montgomery.

Put simply, political commentators and onlookers cannot use prediction markets to definitively predict election results – but the same can be said for opinion polls. Instead, perhaps there’s a future where both tools can be used for insightful analysis in the run-up to major elections.

Featured image: Pexels

The post Prediction markets challenge political polls in elections appeared first on Our Story Insight.