In January 2026, US Special Forces raided a compound in Venezuela and captured the country's leader, Nicolás Maduro. Within days, an American soldier named Gannon Ken Van Dyke had cashed out more than $400,000 on a website called Polymarket — a platform where users bet real money on whether future events will happen. The bets he had placed, prosecutors allege, were not lucky guesses. They were, according to charges filed this month, the work of an insider who knew the raid was coming.
Van Dyke's case is the first US prosecution for insider trading on a prediction market, and it has thrust a once-niche corner of the internet into a national-security debate. Polymarket and rivals like Kalshi let users buy contracts tied to outcomes — 'Will China invade Taiwan by 2026?', 'Will the Iranian regime fall by April 30?' — with prices that function as probabilities. Defenders argue these markets harness collective wisdom to forecast reality. Critics now ask whether the wisdom on display is collective at all.
A new report from the Anti-Corruption Data Collective, a non-profit research group, examined more than 400,000 Polymarket markets settled between January 2021 and March 2026. The findings are striking. Long-shot wagers — defined as bets of $2,500 or more placed at odds of 35-to-1 or worse — succeeded 52% of the time in markets concerning military and defence actions. Across all politics-focused markets the figure was 25%. Across the platform overall it was just 14%. In other words, the people willing to risk serious money on improbable military events were winning at a rate that, statistically, looks less like prediction and more like prior knowledge.
Lawmakers have noticed. Yassamin Ansari, a Democratic representative from Arizona, has called wagers on military actions 'a disturbing national security risk.' Her colleague Ritchie Torres of New York warned that such markets create 'a perverse incentive' for government insiders to personally push policies that line their pockets. Their concern is structural: when classified information has a price tag, the people who hold it have a motive to leak — or to act.
The platforms themselves are responding in different ways. Kalshi, Polymarket's main rival, has heavily promoted its compliance efforts, partnering with surveillance firm Solidus Labs and presenting itself as the largest regulated prediction market in the United States. Kalshi bans what it calls 'violent markets,' including those on war and kidnapping, and requires users to verify their identities. Polymarket, by contrast, accepts anonymous cryptocurrency payments and skips identity verification for most users. It declined to comment on the report but has said it prohibits trading on stolen confidential information, and pointed to Van Dyke's arrest as 'proof the system works.'
Here's the catch: even researchers who defend prediction markets concede that their accuracy may not come from crowds at all. A study led by Roberto Gómez Cram at the London School of Economics found that just 3% of accounts generate most of Polymarket's price discovery — the process by which trading activity arrives at a market's 'right' price. The rest of the users, the study suggested, mostly lose money. If that small minority owes its accuracy to leaked information rather than analytical skill, the entire premise of the prediction market — that many guesses produce truth — begins to look less like wisdom of crowds and more like a high-tech laundering of secrets.
What if you could place a bet on a secret military raid — and win big because you knew it was coming? That's exactly the fear haunting Polymarket right now.
Polymarket is a prediction market — a platform where users bet real money on whether future events will happen, like 'Will China invade Taiwan by 2026?' or 'Will the Iranian regime fall by April 30?' A new report from the Anti-Corruption Data Collective found that more than half of large 'long-shot' bets ($2,500+ at odds of 35-to-1 or worse) on military action turned out to be winners — a stunning 52% success rate.
Compare that to a 14% win rate for long-shots across the platform overall, and alarm bells start ringing. US prosecutors recently charged an active-duty soldier, Gannon Ken Van Dyke, with placing roughly $33,034 in bets — including wagers tied to a January raid that captured Venezuelan leader Nicolás Maduro. He allegedly walked away with over $400,000.
Prediction markets are supposed to crowdsource wisdom — pooling everyone's guesses to forecast reality. But when some bettors have non-public information, betting stops being prediction and starts being insider trading.
This story sits at the intersection of three things shaping your future: the rise of crypto-based betting platforms, the blurring line between gambling and investing, and the question of whether national security can survive in an age where every secret has a price tag. If you're considering careers in law, finance, intelligence, or tech policy, prediction markets are the next frontier regulators are scrambling to understand — and the rules being written now will define the industry for decades.
Prediction markets aren't going away — a whole ecosystem of start-ups like Unusual Whales and Polywhaler is now selling tools to detect (or copy) suspected insiders. Watch for two things: tighter US regulation modeled on securities law, and a possible split between 'serious' regulated platforms like Kalshi and looser, anonymous ones. The deeper question? Whether informed leaks make these markets more accurate — and whether that accuracy is worth the national-security risk.