Prediction markets face a major test in the US midterms, with more than US$1.48 billion wagered and prices implying Democrats are favoured to win both chambers.

Kalshi and Polymarket have grown fast, but past misses and claims of market manipulation raise doubts about how reliable their odds are in close races.

Experts say the markets are useful and often well calibrated, but they are not perfect, and a bad midterm result could damage the industry’s reputation.

AI generated WASHINGTON – With more than US$1.48 billion (S$1.9 billion) wagered on the Nov 3 US midterm elections, prediction markets such as Kalshi and Polymarket are facing their biggest test yet of how reliably they can forecast political races amid growing scrutiny of their businesses.

The platforms rose to prominence in 2024 when they bucked traditional opinion polls in signalling that Donald Trump would win the presidency, and have since amassed a global following among media outlets and investors.

On the afternoon of Nov 3, prediction market prices aggregated by analytics firm BetInt were mostly unchanged from noon, implying Democrats had a 94% chance of taking the House of Representatives and a 74% chance of taking the Senate, while the odds on some competitive Senate races had moved.

Though the markets tout high accuracy rates in providing betting odds, some high-profile misses during the US primaries and in Brazil’s presidential election have shown they are fallible.

If the midterm results diverge sharply from prediction-market odds, the industry’s reputation could suffer, analysts say, especially given the record sums being wagered.

Kalshi and Polymarket, the biggest platforms, are also under pressure to guard against misconduct after several candidates bet on their own races, and research from Washington nonprofit the Anti-Corruption Data Collective (ACDC) found political markets can be shifted with relatively small bets.

“Tuesday does, in fact, represent a real test for the new wave of prediction markets,” said Yesha Yadav, associate dean at Vanderbilt University Law School, adding that 2026’s many congressional races could help contribute to the industry’s overall scorecard.

Polymarket and Kalshi say they have ramped up controls and are well equipped to guard against misconduct. Kalshi blocks politicians and campaign workers from political wagers. The companies also say their markets quickly correct when large bets move prices.

Sean Patrick Maloney, chief executive of the Coalition for Prediction Markets, said the markets do not pretend to be perfect but that in “a world full of chaos and biased news sources,” they provide an unbiased way to forecast events.

Studies say markets are well-calibrated at scale Kalshi on Nov 2 reported US$1.33 billion in trading volume across all midterm bets, while about US$145 million has been wagered on Polymarket’s congressional markets, according to ACDC.

More than 28,000 midterm markets have been opened so far, with a fraction of individual contests, including the Senate races in Texas, Michigan, Maine and South Carolina, accounting for a large chunk of the trading volume, the nonprofit said.

The Texas Senate race, for example, had drawn nearly US$67 million in wagers as of Oct 29, according to ACDC.

On the afternoon of Nov 3, prediction markets implied a 74% chance that Texas Democratic candidate James Talarico will defeat the state’s Republican Attorney-General Ken Paxton for that seat, up three percentage points on midday, according to BetInt.

In the Michigan Senate race, which has drawn more than US$26 million in wagers, the markets implied a 76% chance that Democrat Abdul El-Sayed will overcome Republican Mike Rogers, down two percentage points on midday, according to BetInt.

Academic studies, including a 2025 Vanderbilt University paper which analysed nearly 2,500 markets on the 2024 election, have generally found prediction markets to be well calibrated overall. While calibration measures whether odds were reliable, accuracy measures whether the market picked the winner.

Because most races are not competitive, broad measures of performance do not necessarily show how well the markets perform in close contests, said Joshua Clinton, the Vanderbilt political science professor who led the 2025 research.

A small proportion of sophisticated profit-seeking traders appear to account for a large chunk of prediction market capital, according to Harry Crane, professor of statistics at Rutgers University. Such traders, using polling and other data, help to correct prices, said Crane.

BetInt on Nov 3, for example, flagged two trades of US$500,000 each and another US$100,000 trade, all placed on Republicans on Kalshi last week, likely by hedge funds. Still, even the most expert traders can get it wrong, Clinton said.

Prices on Kalshi and Polymarket implied just a 20% chance, for example, of Brazilian President Luiz Inacio Lula da Silva defeating his right-wing rival Flavio Bolsonaro, which he did in October.

That may have been because sophisticated traders over-corrected for the risk that polls often underestimate the pull of right-wing candidates, said Clinton.

Company spokespeople said that events with a 20% chance of occurring happen all the time and political upsets are no different.

Carlos Hardenberg, portfolio manager at MCP Emerging Markets, said the lesson for investors was that the markets are a useful gauge. “But they should not be confused with certainty.”REUTERS