Study finds big prediction-market bettors lag smaller traders

A new study tied to UC Berkeley raises a direct challenge to one of prediction markets’ biggest selling points: the idea that prices can turn crowd opinion into useful forecasts. For everyday investors watching platforms like Kalshi and Polymarket, the finding is a warning that bigger bets do not necessarily mean better information.

The paper, identified as Daleep et al. (2026), examined the trading structure of Kalshi and Polymarket across 5,456 markets, according to the researchers. Those markets covered three areas: short-term crypto contracts, so-called Mention Markets tied to what companies might say on earnings calls, and forecasts for NBA games.

Prediction markets let users trade contracts linked to real-world outcomes. In practice, the market’s headline forecast gives more influence to participants who put more money behind a view, because larger wagers have a bigger effect on the aggregate price. The study asks whether those larger traders are actually more accurate.

The researchers’ answer: often, they were not. The paper found that traders making larger wagers, often called whales, did not show a better edge than smaller participants. “Edge” means the average profitability of their trades, a way to measure whether a group is making better-than-market predictions or losing money against other traders.

According to the study, the pattern varied by market type, but the core issue was consistent. In 15-minute crypto markets, the smallest traders had the strongest edge, while the largest traders had the weakest. In Mention Markets, the biggest traders showed a negative edge, meaning their trades lost money on average. NBA-related markets showed a similar result, with larger traders losing on average to smaller participants.

That matters for how readers should interpret the probability shown by a prediction market. If bigger traders have at least the same forecasting skill as smaller traders, then giving them more weight can make sense. If they have less skill, the market’s money-weighted forecast can be worse than a plain average of many views, according to the argument described in the study.

The researchers also found systematic bias in prediction markets, though the direction was not the same everywhere. Some markets showed optimistic bias, while others showed pessimistic bias, according to the paper. That means the problem was not just that markets leaned one way across the board, but that their errors depended on the type of contract being traded.

The study’s broader conclusion is that conviction and information are not the same thing. Larger traders may be more willing to bet heavily, but the researchers found no statistically significant link between the intensity of sentiment and informational edge.

In the abstract, the authors wrote: “We find no statistically significant correlation between sentiment intensity and informational edge, indicating that the most prominent voices in these ecosystems function primarily as sources of communicative noise.”

For investors, the takeaway is less about any single Kalshi or Polymarket contract and more about how to read these markets. A visible price can be useful as a snapshot of where money is placed, but the study suggests that the loudest capital in the room may not be the most accurate signal.

This story draws on original reporting from Klement on Investing.