NOTUS Says Three Upsets Prove Prediction Markets Are Broken. The Math Says the Opposite.
Political operatives told NOTUS this week to stay away from Kalshi and Polymarket when reading about the 2026 midterms, and they have three real examples to back it up.
Angie Nixon won Florida’s Democratic Senate primary after Kalshi gave her single-digit odds. David Crowley upset Francesca Hong in Wisconsin’s governor primary the same way. Abdul El-Sayed won Michigan’s Senate primary by less than a point after Polymarket had him above 98% to win by double digits. Every one of those results checks out.
What doesn’t check out is the conclusion operatives are drawing from them, because a growing body of research on how these markets actually price political risk says the opposite of what “useless” implies. The upsets aren’t proof the markets are broken. They’re closer to proof the markets are working exactly as a probability is supposed to work.
The Upsets Are Real, But That’s Not the Same as Proof
Let’s start with what a single-digit chance actually means. If Kalshi is right that a candidate has a genuine 5% shot, that candidate should win one time in twenty, not zero times in twenty. Three longshots landing in one primary season sounds damning until you remember how many primary races these platforms priced this cycle, most of which resolved exactly the way the market said they would.
Cherry-picking the outliers and calling them representative is the same statistical mistake pollsters get accused of every time a 90-10 lead evaporates on election night.
None of that means campaigns should treat Kalshi like gospel. It means the “useless” framing conflates two different critiques: markets being poorly calibrated, and markets being correctly calibrated but misread by people expecting certainty out of a probability. Those are not the same problem, and only one of them is actually about prediction markets.
The Math Says Political Markets Run Underconfident, Not Broken
There’s a clear academic answer to which of those two problems is happening, and it challenges the skeptics. A recent analysis of 353 million trades across 429,000 binary contracts on Kalshi and Polymarket, using Bayesian hierarchical modeling to isolate calibration effects, found that political markets are systematically underconfident, not overconfident. Prices sit chronically compressed toward 50 cents relative to true outcomes.
A political contract trading at 70 cents a week before resolution corresponds to roughly an 83% true probability, not 70%, according to the model.
That finding flips the entire premise of the NOTUS piece on its head. If political markets already understate the odds of the favorite, then a “single-digit chance” longshot may have had an even smaller true probability than the market displayed, and still won anyway, because unlikely things happen. That’s not a flaw in the pricing. It’s what correctly calibrated uncertainty looks like from the outside, and it’s a genuinely harder thing to sell to a campaign staffer who wants a number that reads like a guarantee.
Where the Skeptics Are Actually Right
Not every part of the critique falls apart under scrutiny. Charles Franklin’s “wildly unrepresentative sample” line holds up better than the upset anecdotes do. Navigator Research surveyed 1,000 registered voters in early April and found only 14% report betting on prediction markets “sometimes” or “frequently,” and the demographic gap inside that number is sharp: 39% of men age 18 to 34 use these markets versus 23% of women the same age, and 29% of Gen Z versus just 2% of Baby Boomers.
A skewed poll is a real reason to treat market pricing as a different kind of signal than a demographically weighted poll, and that skepticism is grounded in data, not three cherry-picked election nights. Campaigns also have a track record of treating these markets as more than a spectator sport: Gaming America reported on staffers admitting to trading on unreleased internal poll data, which is a genuinely different and more serious problem than miscalibration. That distinction matters more now that Kalshi has struck data partnerships with both CNN and Fox News, putting these prices in front of a mainstream audience that has no reason to know the difference between a representative poll and a thin, young, male trading pool pricing the same race.
Gaming America raised this exact risk when prediction markets first started courting news outlets for legitimacy: a chyron showing “62% chance” reads to a viewer like a poll number, even when the underlying sample looks nothing like one.
South Carolina Shows the Opposite of What the Critics Claim
The clearest counterexample to the “stay away” advice happened in South Carolina, and it’s the one detail the skeptical framing glosses over. Kalshi’s runoff market repriced live as votes came in on election night and correctly signaled that Darline Graham’s first-place finish would fall short of the 50% threshold, requiring a runoff, before that fact was officially confirmed.
The market beat the Associated Press’s own call by roughly 20 minutes. That’s not a case of a probability tool getting outrun by reality. That’s a case of it reading a fast-moving, structurally complicated result faster than the newsroom desks that operatives trust by default.
The honest verdict sits between the two extremes. Prediction markets are not a polling substitute, and the unrepresentative-sample critique is real and worth repeating every time a campaign staffer starts treating Kalshi’s number as a forecast rather than a probability.
But “useless” doesn’t survive contact with the calibration research, and neither does citing three correctly-priced longshots as if they were three failures. The markets did what they were built to do. The people reading them just wanted something they weren’t built to give.
Players trust our reporting due to our commitment to unbiased and professional evaluations of the iGaming sector. We track hundreds of platforms and industry updates daily to ensure our news feed and leaderboards reflect the most recent market shifts. With nearly two decades of experience within iGaming, our team provides a wealth of expert knowledge. This long-standing expertise enables us to deliver thorough, reliable news and guidance to our readers.
Latest News
