358,236 Markets, One Verdict: The Price Is Usually Right
Prediction markets finally have a report card big enough to trust. PredictDesks graded 358,236 resolved markets, 333,464 on Polymarket and 24,772 on Kalshi, against what actually happened. A day before resolution, the favourite won 76.9% of the time, and in standard markets the average gap between price and reality was just 0.9 points on Polymarket and 1.7 on Kalshi.
The feed did not publish this story's article address.
Full attribution under the Digital Millennium Copyright Act (DMCA), 17 U.S.C. § 512: this article is an original report by Infiniti Intelligence that summarizes the reporting of the publishers named above, links to their original work, and reproduces no protected expression beyond brief factual reference and in-line attribution (“according to PredictDesks”). All rights in the underlying reporting remain with the publishers listed; nothing here implies endorsement by or affiliation with them. Rights holders may request correction, update or removal through the contact address published at infiniti.w3ai.io.
Strip out the easy questions and the picture sharpens. Two-thirds of Polymarket's sampled markets resolved No, most of them long shots parked at a cent or two. Limit the test to markets priced between 10% and 90% a day out, the genuinely uncertain ones, and the favourite's win rate falls to 65.9%. The headline accuracy is real, but it is flattered by questions nobody was ever unsure about.
What is being forecast matters more than where. Tech and science markets scored a Brier 0.040 a day out on Polymarket; sports, the biggest category on both platforms, scored 0.196, the hardest call on the board. Weighted to the same mix of topics, the two exchanges finished in a dead heat: 0.150 for Polymarket, 0.147 for Kalshi. Across 24 Federal Reserve meetings priced on both, their odds differed by just 0.8 points a day before the decision.
The warning is aimed at the industry's fastest-growing product: templated ladder markets, temperature brackets, spreads, totals and exact scores, now 38% of the sample and 47% of Polymarket's 2026 markets. There, outcomes priced under 10 cents won 1.6 times as often as their prices implied, and favourites priced at 80 to 90 cents won only 76% of the time. In standard markets, long shots were the leak: every $100 spent on contracts below 5 cents returned about $81, before fees.
The rarest line in the data may be the most honest: in just 1.1% of markets, the winner had been priced at 10% or less a day earlier, a list that includes the election of Pope Leo XIV. The crowd is usually right. The study's gift is showing exactly where it is not.