Polymarket is Making Wars More Dangerous

prediction markets and the new politics of conflict

When investigators recently questioned an Israeli Air Force major about placing bets on upcoming strikes, the officer was apparently unfazed. “The entire squadron is on Polymarket,” he told them. “The entire air force is betting.”

Prediction markets have now spread to modern warfare. So it’s worth asking what their rise means for global politics. What happens when wars and crises are subject to real-time price discovery by a network of anonymous actors? What does it mean for geopolitical uncertainty itself to become a financial instrument?

I see three ways prediction markets are reshaping global politics, and none of them are good. First, they create a new vector for treason and intelligence leaks; second, they produce new constituencies for conflict escalation; and third, they give new advantages to corrupt regimes who manipulate these markets for their benefit.

In theory, the big difference between prediction markets and gambling is informational. They convert private information into public knowledge in a more accurate way than polls or pundits. Polymarket did better than most pollsters in the 2024 election and the defenders will cite this fact for years.

But recent research suggests that’s not how they actually work. A study of every transaction on Polymarket found only three percent of accounts are responsible for almost all price discovery. The remaining suckers generate volume but no information. It works not because the crowd is wise, but because a small minority extracts profits from everyone else. This means superior information in political markets essentially requires insider information rather than a better model. (edit: Brad LeVeck points out that the paper distinguishes between skilled traders who drive price discovery and insider traders who profit from specific events but don't systematically move prices.)

In many markets, a trader can get an edge with skill or a better model. Winning in conflict markets, however, depends not on better models but insider knowledge. And in conflict markets, insider knowledge means state secrets.

Take the story we started with. An Israeli Air Force major attends a classified briefing, learns the date of the strike on Iran, and passes it along to a civilian accomplice who places a bet on Polymarket. They split $160,000 in winnings. They then bet on the end of the twelve-day war. Then on a strike in Yemen. Then they try again when fighting with Iran resumes, before getting spooked by social media posts about their account.

That’s the officer who said his entire squadron was on Polymarket. Another crew member admitted to making tens of thousands from similar wagers. He was placed under house arrest but not indicted.

Meanwhile last month federal prosecutors indicted a U.S. Army special forces soldier involved in the Maduro raid. He placed bets on Maduro’s removal just before Trump announced the capture, which paid out over $400,000. He was charged with fraud and unlawful use of classified information.

Trump, when asked about the story, compared it to “Pete Rose betting on his own team.” He added: “The whole world, unfortunately, has become somewhat of a casino.”

The link between state secrets and private gain predates Polymarket. Some economists tracked stock prices of Western companies in countries with CIA coups during the Cold War. They found that stocks of exposed firms (United Fruit in Guatemala, Anglo-Iranian Oil in Iran, etc.) began rising just before the coups, in response to top-secret authorizations. Someone with insider access was profiting on the information.

But the difference now is scale and access. Cold War coups generated insider trading profits for a tiny corporate elite. In 1954, you had to be a board member with a CIA contact; in 2026 you just need a phone and a friend in the squadron. As the man said, the entire air force is betting. Prediction markets have democratized the United Fruit model.

The other consequence of entire squadrons betting is prediction markets become an OSINT source. Foreign spy services don’t need to recruit agents if they can monitor unusual betting patterns before military operations. In a national security context, the very thing that’s supposed to make prediction markets valuable becomes another outlet for treason and espionage.

The Hair Dryer Problem

Insider information could in theory be regulated. A more intractable problem with geopolitical prediction markets is the entanglement between the bet and the outcome.

In March, military correspondent Emanuel Fabian reported that an Iranian missile struck an area near Beit Shemesh. No injuries, just a minor incident. But on Polymarket, more than $14 million had been wagered on whether Iran would strike Israel that day. The rules specified that intercepted missiles didn’t count, only impacts on Israeli soil. Fabian’s reporting would resolve the bet as “Yes.”

People who wagered “No” wanted him to change his report to say the missile was intercepted. It started as polite requests and then escalated into bribery. Then vague threats. “You have exactly half an hour to correct your attempt at influence,” said one message. When Fabian ignored the deadline, the threats escalated: “You will discover enemies who will be willing to pay anything to make your life miserable.” Someone posing as a lawyer called to tell him he was being investigated for “manipulation.” A journalist from another outlet contacted him at the request of a friend who was a bettor, offering a cut of the winnings if the article was rewritten.

Polymarket banned the accounts involved and condemned the behavior. But the fundamental problem remains: the market not only evaluates reality but generates a class of people with money riding on how reality is described. Every war correspondent, crime reporter, or journalist covering a trial or an election is now producing content that resolves bets. It’s not just governments and corporations bribing or pressuring reporters anymore, but thousands of anonymous bettors with money on the line.

And with enough incentives, people will try to change not just how reality is described but the nature of reality itself.

Last month, Polymarket was settling daily temperature bets for Paris using an unguarded sensor near the Charles de Gaulle airport. On April 6, someone bought the long-shot contract for abnormally high temperatures. Later that evening, the sensor saw a four-degree spike before dropping back to normal. No neighboring station registered anything similar. The contract resolved in the bettor’s favor. He collected $14,000, and then again on April 15 for another $20,000. The suspected method was a portable hair dryer held near the thermometer.

France’s meteorological agency filed a criminal complaint and Polymarket quickly switched its data source, but did not cancel the contracts or refund the bets. The resolved payouts stand.

It’s really kind of beautiful. The entire prediction market edifice, with its blockchain infrastructure, smart contracts, and information-aggregation theory was defeated by a man with a hair dryer. But then, Hayek could have told you the same thing. When you build a financial instrument that pays based on a number produced by a heat sensor in a plastic box, someone will try to point a heat source at that box.

And if someone will point a hair dryer at a thermometer for $14,000, imagine the incentives around a missile strike. Conflict prediction markets create entire new constituencies for violence escalation. If you’re long on “Iran strikes Israel” then you want that strike to happen. Aggregated across thousands of bettors, this creates a constituency for escalation. This kind of racket has always existed among arms manufacturers and war profiteers, but prediction markets democratize it to more people.

Rigging the Truth Machine

When odds become authoritative they also become worth manipulating, and not just by people with hair dryers. If prediction markets are seen as credible expressions of public feeling, moving the market can be an effective form of propaganda.

A well-funded actor like, say, a political party trying to demoralize an opponent, can buy prediction contracts to manufacture a self-fulfilling prophecy. During the 2024 election, for example, there were widespread allegations that a single large Polymarket trader was placing massive bets on Trump to create a narrative of inevitability. Whether this was a deliberate propaganda operation or just a wealthy person’s conviction was never established, but the problem has not gone away.

For all their praise of dispersed knowledge, prediction markets will not decentralize power where it’s already concentrated. Instead they will become another tool of regime entrenchment. Autocrats already have institutional and informational advantages through their control of security services, courts, and companies that collect information. They are perfectly placed to extract rents from informational markets, and to use informational market manipulation to signal strength or inevitability.

Remember, prediction market accuracy relies not on the crowd but on a small minority of informed traders. If only three percent of accounts drive price discovery, you don’t need to outbid millions of independent signals, only to co-opt or suppress some fraction of those three percent.

Defenders will point to arbitrage: if a state actor artificially inflates the odds of an unlikely event, rational bettors will correct the price and take the manipulator’s money. That’s another form of the efficient markets defense, and assumes something that doesn’t exist for geopolitical contracts: a knowable correct price. If someone moves a contract on NATO intervention from 15 to 35 percent, the arbitrageur would have to be certain it’s manipulation and not insider information, against an actor who may in fact know something. That’s very different than correcting a mispriced equity with anchor fundamentals. Not to mention it assumes the manipulator even cares about losing the money. States spend billions on information warfare each year. A few million dollars lost to arbitrageurs is a rounding error in the propaganda budget.

Besides, even if the correction eventually comes, it doesn’t matter. The propaganda value is extracted immediately and the headline (“Polymarket odds of NATO intervention surge to 35%”) is published the moment the price spikes. Market manipulation in this context doesn’t have to be sustainable to be newsworthy.

The Hayekian defense of prediction markets assumes a liberal institutional substrate that allows dispersed private knowledge, a free press, and an independent judiciary. But these things just don’t exist in much of the world. Or, increasingly, in the US. Donald Trump Jr. is a strategic adviser to Kalshi and an investor in Polymarket. A Polymarket trader made roughly $300,000 correctly betting on Biden’s last-minute pardons. Another made nearly a million dollars betting on unannounced U.S. and Israeli strikes against Iran, with a 93 percent success rate across three separate operations. The only person indicted so far is a sergeant who used his personal email.

When institutions are opaque, prediction markets will not aggregate dispersed knowledge. They will consolidate regime power and create new unregulated venues for corruption and propaganda.

The Price of Everything

National security aside, the most annoying thing about prediction markets has been the way they’ve elevated gambling into something socially desirable and even prestigious by associating it with “data”. Everything worth knowing is now worth pricing. If you’re successful on Polymarket you get invited for an interview on 60 Minutes. The CEOs of both Polymarket and Kalshi have called their products “truth machines.” John Arnold, an energy trader, claims there’s a distinction between geopolitical prediction markets, which he says can provide useful signals, and sports betting engineered purely for addiction.

But this distinction is less clean than he thinks. Geopolitical prediction markets have their own pathologies, which have direct implications for national security, democracy, and state conflict. They incentivize treason and fraud, create constituencies that benefit from escalation, and give structural advantages to unconstrained actors.

Conflict prediction markets are the logical endpoint of a culture that has already financialized everything else. But when you try to financialize reality itself don’t be surprised when the market starts making the news instead of predicting it.