Six weeks before US and Israeli aircraft killed Iran’s supreme leader on February 28, a cluster of new, previously dormant crypto wallets began buying “Khamenei out by March” contracts on Polymarket. In the hours before the strike itself, six accounts turned a combined $1.2 million profit on bets timed almost exactly to the raid. By the time it was over, $529 million had been wagered on the strike’s timing and $150 million more on Khamenei’s removal — most of it settled in traders’ favor before the Pentagon had said a word. No diplomat, briefing, or wire report beat that market to the news. This week, as Kalshi and Polymarket announced a combined $50 billion-plus month, a fresh round of commentary asked whether prediction markets have become a faster, cleaner signal than official diplomatic statements. That is the right question. Most of the people asking it are about to get the answer wrong.
Kalshi and Polymarket cleared $50.6 billion in trading volume in July, a record — but the headline number is misleading almost by design. Roughly $20 billion of it traced back to World Cup contracts alone, and sports and crypto-price bets, not geopolitics, are what pushed both platforms past prior highs. The genuinely political and geopolitical share is smaller, faster-growing, and far more consequential: Polymarket alone carried over $2 billion in war-related wagers during the worst weeks of the Iran war in early March, and both platforms now run live, continuously priced contracts on Ukraine’s on-again, off-again ceasefire process — including one, as of this week, on whether Russia will even respond to a Ukrainian proposal to halt strikes on Black Sea civilian targets. This is not a new debate. In 2003, the Pentagon’s own research arm tried to build a government-run version of exactly this — the Policy Analysis Market — and Congress killed it within a week of its existence, after senators branded it a “terrorism betting parlor.” Twenty-three years later, the public is running that experiment anyway, just without anyone in government able to switch it off.
Start with the case for the defense, because it is stronger than most skeptics admit. The Central Intelligence Agency’s own in-house journal argued in 2006 that prediction markets could outperform conventional analysis — citing the Iowa Electronic Markets’ record calling presidential elections, orange juice futures beating official National Weather Service forecasts, and an internal HP Labs experiment in which employee prediction markets beat the company’s official sales forecasts roughly three times out of four. The mechanism is simple: markets aggregate dispersed information that no single analyst, however senior, holds all of. The pre-strike Iran betting looks, on its face, like that mechanism working at wartime speed — money moving on knowledge before it became news.
But look more closely at what kind of knowledge that was, and the “markets beat the CIA” framing starts to fall apart. Blockchain analytics firms traced the pre-raid profits to a small number of wallets with no trading history before mid-January, clustered almost entirely around contracts tied to the specific date and target of the strike. That is not the textbook picture of dispersed public information converging on a price — it is the picture of a leak, or a very good guess by someone close to the planning, being laundered into an anonymous, liquid, and legal instrument. A member of Congress has since demanded answers from Polymarket over a single $500,000 Iran-strike bet. The market did not out-analyze American intelligence agencies; at best, it gave someone with access to inside information — American, Israeli, or Iranian — a faster and more deniable way to monetize it than existed before. That is a genuinely important development for open-source intelligence. It is not the same claim as “the market knows more than the analysts,” and conflating the two is exactly the mistake now spreading through IR commentary.
Stay ahead of the geopolitical week.
MD Briefing delivers expert analysis across five global fronts — the Indo-Pacific, energy, geoeconomics, European security, and the Middle East — every Monday morning. Free.
The second problem is more mundane and, for analysts, more dangerous: contract language. A widely cited Polymarket contract asking whether Russia and Ukraine would reach a ceasefire before the end of 2026 resolved “yes” months ago — settled by a 32-hour Orthodox Easter truce in April that both sides violated within hours and that collapsed into renewed full-scale fighting by the time the exchange paid out. Anyone who read that resolved contract as evidence the war had ended, or was ending, would have been badly and specifically wrong: Kyiv and Moscow have cycled through at least two further truces and collapses since, and this week Ukraine is waiting on an answer to a narrower proposal covering only Black Sea civilian targets. Iran markets have the mirror problem in the other direction — a $345 million Polymarket contract on a “permanent peace deal” has stalled in trader disputes over whether a temporary US-Iran arrangement even counts, because the contract’s language was never pinned to a verifiable, specific event. A price is not a probability of the thing you think it is unless the contract defines that thing precisely. Most don’t.
Put those two problems together and the honest position is neither “ignore these markets” nor “trust the price.” It is that prediction markets are best read as a real-time leak-and-sentiment detector, not a forecasting oracle — genuinely useful for spotting when someone with access is moving early, genuinely useless, and often actively misleading, as a stand-alone probability estimate of what will actually happen. That distinction is exactly the one the political system has never let mature. The 2003 Policy Analysis Market was killed not because its statistical premise was wrong — the CIA’s own analysts still cite the underlying research approvingly — but because formalized betting on state violence is politically toxic. The same instinct is visible today in the lawmaker letters over Iran bets and the SEC’s delayed rollout of prediction-market ETFs. The result is that a genuinely useful analytic instrument has only ever existed in an unregulated, gambling-adjacent form that mixes World Cup bettors, political whales, and possible leakers in the same order book, with no institutional mechanism to separate the signal from the noise.
Three ways the next year plays out for this practice.
Base case (roughly 55% probability): the current pattern continues without correction. IR analysts, journalists, and even some government officials increasingly cite Polymarket and Kalshi prices in briefings and notes as a supplementary indicator, because a price feels more concrete than a hedged diplomatic cable. Occasional embarrassing misreads — like the Ukraine “ceasefire resolved yes” episode — surface, get mocked on social media, and are quietly walked back without changing the underlying practice. Volume keeps growing, sports keeps dominating the headline totals, and geopolitical contracts remain a noisy minority that gets outsized attention relative to its share of the money.
Downside case: a specific market-priced event feeds into an actual policy or military decision — a strike, a negotiating position, a sanctions call — that is later shown to have been driven by a small, possibly coordinated cluster of informed or manipulative traders rather than genuine aggregated wisdom. The resulting scandal, amplified by a Congress already primed by the Iran-bet controversy, produces exactly the kind of political backlash that killed the Policy Analysis Market in 2003 — restrictions or an outright ban on geopolitical and war-related contracts specifically, distinct from the sports-focused lobbying campaign already underway. The useful signal (informed-money detection) gets regulated away along with the noise, because no one built the tools to separate them before the backlash arrived.
Upside case: a specific, credible actor — a quantitative research shop, an open-source-intelligence outfit, or a university lab building on the forensic wallet-tracing work firms like Bubblemaps are already doing informally — formalizes the leak-detection method into a repeatable product: flagging anomalous, newly active wallets clustering on tightly defined, verifiable contracts, separately from ordinary political and sports flow. That would turn today’s ad hoc journalism into the analytic tool the CIA’s 2006 paper imagined, without needing a government agency to sponsor it or a political fight to kill it.
The interesting story was never “prediction markets versus intelligence agencies” — that framing flatters both sides and tests neither. The real finding is narrower and more useful: these markets occasionally move ahead of public disclosure because someone with real information is trading on it, wrapped inside a much larger volume of sports bets and loosely worded contracts that will mislead anyone who reads the price as a probability. Learning to tell those apart, not learning to trust the number, is the actual skill.
Watch what happens to the price of Ukraine’s Black Sea ceasefire proposal over the coming days: Kyiv made the offer on August 13 and Moscow has not yet responded. If the market moves hard before any Russian statement does, that is as clean a live test as this methodology will get of whether these venues are catching a real signal — or just repricing a rumor.
