Academic Study Finds Unrepresentative Bettors Do Not Compromise Election Prediction Market Accuracy
Crane and Strumpf's paper shows election bettors are consistently unrepresentative across eras, with Kalshi 2024 users 86% male and over half under 30. Yet heavy traders driving 93% of volume bet purely for profit, enabling Kalshi to beat major pollsters in the presidential race and Senate contests. House markets with $236k volume underperformed.
SCCG Take — Prediction markets deliver actionable probability signals when liquidity concentrates among profit-focused traders, but congressional races need far higher volume to match that reliability for operators and investors ahead of 2026.
A new academic paper finds that the unrepresentative nature of participants in political prediction markets does not compromise their accuracy. Harry Crane, a researcher at Rutgers University, and Koleman Strumpf of Wake Forest University reviewed more than 100 years of election betting data. Their analysis concludes that financial incentives encourage bettors to prioritize profits over political preferences.
The study addresses ongoing skepticism about prediction markets ahead of the upcoming midterms. As detailed in reporting by InGame, no group of bettors has ever mirrored the voting public, yet this demographic mismatch has not thrown off market prices. On Kalshi in 2024, 86% of election traders were male and more than half were under 30. On PredictIt in 2016, men comprised 92% of presidential traders. Records for the 1900 election in New York showed every bettor was male, 44% were bankers or brokers, and participants were older than the city average.
Incentives Drive Market Efficiency Over Bias
Detailed examination of 2024 Kalshi data reveals distinct patterns. ZIP codes weighted by trading volume supported Donald Trump only 28% of the time, yet 60% of volume favored Republican positions. In contrast, one-time traders placed 73% of their money on Kamala Harris.
Among participants making more than 100 trades — accounting for 93% of total volume in the final month — political leanings showed no influence on betting direction. Strumpf noted, “Their demographics don’t seem to impact how they trade. And I should say this is disproportionately true for heavy traders.” Strumpf added that “dumb” money provides essential opportunities for sophisticated traders. Kalshi's odds surpassed those from Silver Bulletin, FiveThirtyEight, and The Economist in the presidential contest. The platform also outperformed poll-based models across battleground states and 19 Senate races.
Liquidity Challenges in Congressional Contests
Results diverged for 2024 House races. Kalshi launched 24 district-specific markets just six days before the election, attracting only $236,000 in combined volume. FiveThirtyEight produced superior forecasts in 20 of those races. The researchers cautioned that the limited sample prevents firm conclusions about whether the shortfall stemmed from insufficient trading activity. Strumpf stated, “I guess I won’t go to the mat to say markets are going to work great. But I guess I’m very curious to see.”
@RealPData posted on X: "Personally, I think Pandora’s box has already been opened with political betting and you will never be able to close it fully. But I do think there needs to be serious discussions on restrictions and regulations to reduce the likelihood of shenanigans from happening. 1. Any…"
Sufficient liquidity appears central to whether these markets deliver reliable signals in lower-profile races. The 2026 midterms will test how volume scales in congressional contests and whether the efficiency observed at the top of the ticket extends further down the ballot.
Gaming, betting and prediction markets — the desk’s read, every weekday.
Subscribe →