Anonymous traders profited from Iran's bombing while Congress debated rules for a market the Pentagon killed decades ago
The prediction market industry promised the wisdom of crowds. What it delivered was the world's most accessible intelligence leak, and the scramble to contain it may already be too late.
Seventy-one minutes. That was the gap between the moment a Polymarket user called 'Magamyman' placed a $32,000 bet that the United States would strike Iran on 28 February 2026, and the moment the first explosions were reported across Tehran. The platform's own odds had the probability of a strike that day at 17 per cent. Magamyman cleared more than $553,000.
He was not alone. Six other accounts, all freshly created that month, all funded through suspiciously similar cryptocurrency channels, had been buying strike contracts at roughly ten cents a share in the hours before the bombs fell. Blockchain analytics firm Bubblemaps traced the wallets, mapped their funding paths, and published the results within hours: collectively, the six had pocketed around $1.2 million. None had ever placed a bet on anything other than Iran.
This was the second time in two months that anonymous traders appeared to profit from foreknowledge of an American military operation. In January, a newly created account wagered more than $30,000 that Venezuelan President Nicolás Maduro would be removed from power. Hours later, the Trump administration launched its raid. The account collected over $400,000. Maybe these traders were spectacularly lucky. Maybe they read geopolitical signals better than anyone in professional intelligence. Or maybe someone with access to classified war plans decided to cash in. Polymarket cannot say. The platform processes bets through cryptocurrency wallets on the Polygon blockchain, and it neither knows nor seeks to know who its users are.
Its chief executive, the 27-year-old Shayne Coplan, has described this anonymity not as a vulnerability but as the product's central appeal. The financial incentive for people to divulge information to the market, he told an interviewer last year, is what makes prediction markets valuable. What Coplan frames as efficient price discovery, the national security establishment increasingly views as something else entirely: an uncontrolled leak of classified intelligence into a public signal that any government on earth can read for the cost of a browser tab.
The idea the Pentagon killed, and the industry that resurrected it
The American military tried to build this tool once before. In 2001, the Defence Advanced Research Projects Agency began funding the Policy Analysis Market, a proposal to let anonymous traders bet on political developments across the Middle East. The University of Iowa's political futures market had consistently outperformed polls in predicting election outcomes, and the logic seemed transferable: if markets could aggregate dispersed knowledge about elections, perhaps they could do the same for security threats.
The programme never went live. In July 2003, two senators discovered that sample screens on the project's website included scenarios involving the assassination of Yasser Arafat and a North Korean missile strike. Ron Wyden called the concept grotesque. Byron Dorgan called it useless, offensive, and unbelievably stupid. Within a day, the Pentagon cancelled the programme. Within a week, John Poindexter, the DARPA official responsible, had resigned. Careers ended. Bipartisan outrage was absolute.
The proposed market would have served a thousand traders placing bets of a few tens of dollars.
Twenty-three years later, Polymarket processed $529 million in Iran-related wagers in a single weekend, operating entirely beyond the reach of American regulators, with no identity verification, no reporting obligations, and no mechanism for detecting or preventing insider trading. The Trump administration's response has not been to regulate this market but to cultivate it. Donald Trump Jr. serves as a Polymarket adviser. His venture capital firm, 1789 Capital, has invested millions in the company. Two federal investigations into the platform, opened under President Biden, have been dropped. The company is valued at $9 billion.
What destroyed careers in 2003 now attracts presidential family investment. The shift is worth sitting with.
Where the theory breaks
Robin Hanson, the George Mason University economist whose research inspired the original DARPA programme, has long argued that insider trading in prediction markets should be welcomed. The logic is Hayekian: prices aggregate scattered information faster than any bureaucracy, and insiders carry the most valuable information of all. Let them trade, and their knowledge flows into prices that everyone can read. In commercial contexts, the argument has force. A prediction market that rapidly incorporates non-public information about a company's earnings is, by some lights, performing a useful service.
The argument fails when the domain shifts to warfare. The intelligence being traded is not a quarterly earnings figure. It is the timing of an air strike whose success depends on the adversary not knowing when it is coming. The entire value of the information lies in its secrecy. Surfacing it in a public price signal does not make society better informed. It makes the operation more dangerous for the people carrying it out and more survivable for the people being targeted.
Rajiv Sethi, an economist at Barnard College and Columbia University who studies prediction market microstructure, has observed that spotting suspicious trades has become increasingly easy. Newly created accounts placing large wagers against the prevailing consensus stand out. A growing industry of monitoring tools now flags such activity automatically, in real time, for a monthly subscription fee. The implication should unsettle anyone involved in operational planning: if a private analytics startup can identify probable insider trading within hours, so can the intelligence services of Iran, Russia, and China.
Alex Goldenberg, a fellow at the Rutgers Miller Center who studies war markets, was blunter in an interview with the Atlantic: these markets, as currently designed, are built to leak. The American military's own intelligence bulletin published an article last year urging the armed forces to integrate Polymarket data into their threat assessments. The tool the Pentagon killed in 2003 has come back, uninvited, and the Pentagon is now reading it like a morning briefing.
The attack surface nobody saw coming
The most dangerous scenarios run in the opposite direction from insider trading. If governments monitor Polymarket as an intelligence signal, then Polymarket becomes a channel for strategic deception.
Consider: the Islamic Revolutionary Guard Corps subscribes to one of the commercial monitoring services that flag unusual Polymarket activity. The service detects a surge of large, well-timed bets on an imminent American strike, several hours before bombs fall. The Ayatollah cancels the in-person meeting with his top advisers. Iran disperses its leadership. Pre-emptive strikes hit American bases across the Gulf. Six American service members had already died from Iranian drone attacks in the region before the February operation. The margin between operational success and catastrophic escalation hinged, in part, on Iran's ignorance of the precise timing. Anonymous bettors narrowed that margin for personal profit.
Now reverse it. A state actor spends a modest sum pushing Polymarket odds on a fictitious imminent attack, triggering panic, military mobilisation, or diplomatic breakdown in a target country. The cost of moving prices on a low-liquidity geopolitical contract is trivial compared to mounting a conventional disinformation campaign, and the signal carries a strange authority: markets, after all, are supposed to reflect real information. When the information is fake, the authority becomes a weapon.
The cypherpunk blueprint
In April 1995, a crypto-anarchist and MIT-trained chemist named Jim Bell began posting an essay called 'Assassination Politics' to the cypherpunks mailing list. Over ten instalments, Bell described a system of anonymous prediction markets where participants would wager on the date of death of public officials. The person who guessed correctly, presumably because they had arranged the killing, would collect the accumulated bounty. Bell framed this as a libertarian accountability mechanism: leaders who taxed, regulated, or waged war against their citizens would face market-priced consequences. The FBI took a different view. Bell was eventually imprisoned, though on charges of tax evasion and harassment of federal agents rather than the essay itself.
For three decades, the proposal was treated as a thought experiment. Academics debated its theoretical implications. Libertarians argued over its ethics. Nobody built it, because the technology to run anonymous, untraceable betting markets at scale did not exist.
Now it does. Polymarket operates on cryptocurrency. It conceals user identities through the blockchain. It offers contracts that resolve based on whether specific individuals remain in power. Rajiv Sethi and the researchers Sean Guillory and Dan Zimmermann have pointed out that markets resolving on whether someone stays in power, speaks publicly, or holds a particular office carry implicit incentives connected to their continued existence. A contract on Khamenei remaining Supreme Leader is, in its incentive structure, indistinguishable from Bell's assassination bounty. The phrasing differs. The financial geometry does not.
Polymarket briefly listed a contract on the probability of nuclear detonation by the end of 2026. It attracted $850,000 in bets before the platform took it down. The Iowa election market that once inspired DARPA's cautious experiment involved a thousand participants and stakes of a few dollars. Thirty years later, anonymous strangers were wagering on the apocalypse.
The legislative scramble
Congress has responded with unusual speed. Senator Jeff Merkley and Senator Amy Klobuchar have introduced the End Prediction Market Corruption Act, banning the President, Vice President, and members of Congress from trading on prediction markets. Representative Ritchie Torres has introduced parallel legislation in the House. Senator Chris Murphy wants to go further, potentially prohibiting prediction markets altogether.
The Merkley-Klobuchar bill targets a real problem, but a narrow one. The Secretary of Defence is not an elected official. Neither are the thousands of military personnel, intelligence analysts, and defence contractors who hold fragments of operationally sensitive information. A soldier who overhears deployment timing in a briefing room faces no technical barrier to placing a bet from a phone. Israel has already charged two people for using classified information to trade on Polymarket during strikes on Iran last June. Both were relatively low-ranking.
Eric Zitzewitz, a Dartmouth economist who has studied prediction market manipulation, notes that the damage extends beyond the insider's profit. The victims are ordinary users who provide liquidity by posting limit orders, and the threat of insider trading erodes that liquidity over time. In extreme cases, the markets stop functioning as forecasting tools altogether, becoming pure extraction mechanisms where the informed take money from the uninformed.
The regulated competitor, Kalshi, has tried a different tack. After the Iran strikes, Kalshi refused to pay out on contracts whose resolution depended on Khamenei's death, invoking a 'death carveout' policy that prevents the platform from enabling profit from assassination, war, or terrorism. Traders were furious. Some called it a scam. But Kalshi's stance reflects a recognition that federal regulations already prohibit futures contracts tied to political violence, a constraint that Polymarket, operating offshore and beyond regulatory reach, simply ignores. Kalshi has investigated more than two hundred suspected cases of insider trading. Polymarket has investigated none.
What a wager does to the person who places it
During his months at Camp East Montana, the nation's largest ICE detention facility, a legal permanent resident named Owen Ramsingh told the Associated Press that he overheard security guards discussing a betting pool on which detainee would next die by suicide. One guard reportedly said he had paid $500 into the pot. Ramsingh found the conversation particularly disturbing because he had contemplated taking his own life. The Department of Homeland Security denied the account without explaining how it had verified the denial.
A few guards wagering pocket money on human suffering inside a detention camp is not a prediction market. But it shares a psychological mechanism that the language of information aggregation tends to sanitise. When you place a financial stake on an outcome, you change your relationship to it. You acquire an interest. The interest may not make you act, but it reshapes how you feel about what happens next.
Polymarket posted a statement during the Iran crisis claiming that its markets provided accurate, unbiased forecasts as a public service to people directly affected by the attacks. A platform that cannot identify its users, cannot police insider trading, and processes bets through untraceable cryptocurrency was, in its own telling, performing an act of civic generosity by letting anonymous strangers wager on which cities would be hit by missiles that day.
The bind that no bill resolves
The structural problem with war prediction markets resists every proposed fix. Ban officials from trading, and the information leaks through contractors, relatives, and proxies. Require identity verification, and you kill the anonymity that is both the source of the intelligence risk and the basis of the market's claimed forecasting value. Prohibit contracts tied to war, and you face an impossible definitional boundary: a bet on whether the Strait of Hormuz will close is simultaneously a war market and an oil market. A bet on whether a head of state will remain in power is simultaneously a political forecast and, as the cypherpunks recognised thirty years ago, an assassination bounty.
The industry's intellectual justification rests on an assumption that holds in some domains and collapses in others. Elections are not altered by the bets placed on them. Earnings announcements do not change because a prediction market exists. Wars are different. In warfare, the prediction and the event are entangled. The bet can change the battlefield. The price signal can alert an adversary. The financial incentive can corrupt a commander. No regulatory framework yet proposed addresses this entanglement, because the entanglement is not a flaw in the system. It is what the system does.
Polymarket keeps launching new war markets. Will the United States strike Iraq? Will Israel strike Beirut? Will Iran strike Cyprus? On the day of the initial Iran attack, the platform processed $478 million in bets. Substack has since partnered with Polymarket to embed its forecasts in newsletters. The company is valued at $9 billion, backed by Intercontinental Exchange, advised by the president's son, and shielded from every regulator that governs its domestic rival. Anonymous wallets are being funded. Operational plans are being drawn. The distance between those two activities is shrinking, and seventy-one minutes may already be too close.