The market knew

In the first days of January, a quarter-million dollars of "Maduro captured" contracts changed hands on Polymarket at long odds — and then, days later, Venezuela's president was in U.S. custody and the contracts paid out.¹ The trader had not been lucky. He had been there. Master Sergeant Gannon Van Dyke, a Special Forces soldier, had helped plan the raid; he wagered about 33 thousand dollars in the week before it and walked away with roughly 410 thousand.¹ When reporters noticed the suspicious trading, he asked Polymarket to delete his account.

He is now charged with what the Commodity Futures Trading Commission calls its first insider-trading case on event contracts.¹ And he is not alone. The last few months delivered a run of these — a reckoning the category was always going to have, because a prediction market is, by construction, the most tempting place on earth for someone who already knows the answer.

The people who already knew

Run the cases together and a pattern jumps out. In every one, the winning trader's edge wasn't analysis. It was access.

• The soldier. Van Dyke knew Operation Absolute Resolve was coming because he was part of it. Classified knowledge of a military mission, converted into a 12-to-1 return on a public market.¹
• The engineer. Michele Spagnuolo, a security engineer at Google, allegedly used the company's confidential Year-in-Search data to bet — under the handle AlphaRaccoon — on who would be the most-searched person of the year — hitting 22 of 23 calls, an accuracy that is its own confession — and clearing about 1.2 million dollars before federal prosecutors charged him with commodities and wire fraud.²
• The operatives. Two people in Israel were indicted for using classified military intelligence to bet, with eerie precision, on the window in which Israel struck Iran — over 150 thousand dollars in profit on the timing of a war.³
• The pardons. And an anonymous account cleared roughly 300 thousand dollars calling four specific pardons the outgoing president would sign on his way out — a list only a handful of White House staff would ever have seen.⁴

Congress noticed. In May it opened a probe into both Polymarket and Kalshi over the fear that government employees — and members of Congress themselves — are sitting on exactly the kind of nonpublic information these markets price.⁵ The asset class spent two years arguing it was infrastructure, not a casino. It just found out what comes with being infrastructure: the same crimes the rest of finance has.

And look at where they happened. The Maduro bet, the Google-data bet, the Iran-timing bet, the pardons — every one played out on Polymarket, whose dominant exchange is operated out of Panama and sits outside the CFTC's reach. The platform admits it, in a notice at the bottom of its own site: the international book "is not regulated by the CFTC and operates independently."⁶ Polymarket has been trying to buy its way onshore — it acquired a licensed U.S. exchange, QCEX — but even with the license in hand, it still hasn't managed a full U.S. launch: the product sat invite-only for half a year, dropped its waitlist only in May, runs on iOS alone, and is being chased state to state by cease-and-desist orders and court injunctions.⁶ The offshore book it grew up on, meanwhile — the one with far more traffic — is still pseudonymous, still uncleared, still unwatched. That is the second pattern, and it's the sharper one: insider trading isn't an offshore temptation — the temptation is everywhere — it's an offshore outcome, because offshore is where nothing stands between the temptation and the payout.

The uncomfortable part: it's the mechanism

Here is the thing almost no one says out loud. The reason these trades worked is the reason prediction markets are useful in the first place. A market aggregates scattered information into a price because people who know things trade on them. The soldier who bet on Maduro's capture didn't break the machine — he was the machine, doing exactly what it's designed to reward: moving the price toward the truth before the public got there.⁷

So you cannot draw the line at "informed trading," because informed trading is the entire point. A market with no informed traders is just a random-number generator. The line has to fall somewhere more specific — and finance has spent a century working out exactly where. It is not how much you know. It's how you came to know it.

This isn't a fringe position. Robin Hanson — the economist who originated modern prediction markets and proposed governing by them — has argued for years that insider trading isn't the flaw in the design, it's the function: the quickest way to pull private knowledge into a public price is to let the people who hold it bet.⁸ The uncomfortable corollary is that you can't stamp it out without dimming the very signal you built the market to read. Which is why the line can't sit at knowing — it has to sit at how the knowledge was obtained.

A sharp analyst who reads the troop movements in open-source flight data and concludes a raid is coming has an edge, and the market is better for having her in it. A soldier who knows because he is holding the classified op order has committed a crime — the same crime, in spirit and increasingly in law, as a banker trading on a deal his client hasn't announced. The instrument is identical. What separates them is the provenance of the information: lawfully assembled versus stolen, misappropriated, or owed in confidence to someone else.

You can even see the crime in the math. An insider who knows a YES outcome is certain, and buys the contract at the market's price \(p\), earns a payoff multiple of \(1/p\). Van Dyke bought "Maduro captured" while the market still thought it a long shot — call it eight cents on the dollar — and \(1/0.08 \approx 12.4\), which is almost exactly his thirty-three thousand turning into four hundred ten.¹ And that same arithmetic is what gives the insider away:

Win once at long odds and you got lucky. Win a string of low-probability bets in a row — each one a near-certainty only because you'd seen the answer — and the odds of doing that honestly, \(p^{n}\), collapse toward zero. That collapsing number is a fingerprint. It is exactly what a surveillance system is built to notice.

Kalshi catches its own

Which brings us to the case most of the headlines missed, and the one that matters most. In February, Kalshi — the regulated, onshore exchange — publicly disciplined a trader named Artem Kaptur. Kaptur was a video editor on MrBeast's team, and he had been trading the markets on what would happen on MrBeast's own channel: outcomes he could see coming because he was editing them.⁹

His trades were small — about four thousand dollars. What flagged him wasn't the size; it was the shape. Kalshi's surveillance caught what it called "near-perfect trading success on markets with low odds" — the \(p^{n}\) fingerprint made visible. The exchange fined him 15 thousand dollars, clawed back his profits, suspended him for two years, and referred the matter to the CFTC. MrBeast's company fired him and said employees are barred from trading their own markets.⁹

And Kaptur was no fluke. The same surveillance flagged former Congressman George Santos betting on whether he'd turn up to the State of the Union, and caught three congressional candidates wagering on their own races — the purest insider trade there is, betting on yourself.¹⁰ By its own account, in the first quarter of 2026 Kalshi blocked more than a hundred suspicious trades, opened over a hundred and fifty investigations, and made twenty referrals to law enforcement. That is not a venue getting lucky. It is a venue running a surveillance desk.

The difference is structural, not cultural. Kalshi runs as a CFTC-designated contract market with its own clearinghouse — Kalshi Klear — and a surveillance team whose entire job is to watch the order flow for that collapsing \(p^{n}\) and freeze the accounts that light up.¹¹ Polymarket's offshore book has none of it: no U.S. license, no clearinghouse behind the trades, no surveillance mandate, and accounts a soldier can simply ask to delete. One venue is built to catch its insiders; the other is built never to have to know.

Polymarket

Offshore · where the scandals are

• Dominant exchange run from Panama
• Not CFTC-regulated — "operates independently"
• No clearinghouse behind the trades
• Pseudonymous accounts — one asked to be deleted
• No surveillance mandate

Kalshi

Licensed · catches its own

• CFTC contract market — KalshiEX
• Its own clearinghouse — Kalshi Klear
• KYC identity on every account
• In-house surveillance for the fingerprint
• A regulator to hand the file to

Notice the contrast in how each was caught. The Polymarket cases surfaced the loud way — a suspicious whale, reporters digging, a federal indictment, a trader trying to delete his account after the fact.¹ To Polymarket's credit, it flagged Van Dyke's activity and passed it to the Justice Department.¹ But Kaptur was caught the quiet way: a known, identity-verified account, watched in real time, stopped by the venue itself before it became a federal case. That is not a difference of luck. It is a difference of plumbing: a venue that knows who its traders are and watches what they do, against one that does neither.

The other way a market betrays you

Insider trading attacks the price. There's a second failure that attacks the payout — and a market on the same platform is about to test it. Polymarket is running a market asking whether Michael Saylor's Strategy will sell Bitcoin by May 31. More than 175 million dollars rides on it.¹²

Strategy may well sell before the deadline — but the contract turns on something subtler than the sale: it keys on when a sale is confirmed, not when it occurs, and a corporate sale is typically disclosed only in a later SEC filing. If the disclosure lands after May 31, a market that everyone "knows" should pay YES could resolve NO on the wording alone — and you can already hear the dispute that would follow, a YES holder accusing the platform of rewriting the criteria after the fact.¹² Whichever way it lands, the lesson is the one we keep coming back to: the wording of a market is not paperwork, it's the product, and a resolver that looks like it's improvising — even once, even defensibly — spends trust it can't easily rebuild.¹³

So the integrity problem has two halves, and they want different fixes. The price is corrupted by stolen information; you answer it with surveillance and law. The payout is corrupted by ambiguous resolution; you answer it with rules crisp enough that there's nothing to improvise, set before a dollar trades. Most of the trust in this category will be won or lost on those two fronts, not on the app.

$410K

a soldier's payout on a raid he'd planned

1st

CFTC insider charge on event contracts

$15K

Kalshi's fine on the insider it caught itself

What it actually means

It's tempting to read a run of scandals as a verdict against the whole idea. We read it the other way. Equities had insider trading and settlement fights for the better part of a century before the surveillance and the case law caught up; the existence of those crimes is not evidence that stocks were a scam — it's evidence that they mattered enough to steal from. Prediction markets just compressed that arc into a few quarters. The CFTC bringing its first event-contract insider case, Congress holding hearings, an exchange disciplining its own trader — that is not a category dying. That is a category being taken seriously.

The honest hard part is that the edge-versus-crime line will never be perfectly clean. Push enforcement too hard and you scare off the informed traders who make the price worth reading; too soft and the venue becomes a laundromat for stolen information. The answer isn't a brighter line; it's better plumbing on both sides of the one finance already drew — identity and surveillance to catch the provenance, and resolution discipline so the payout is never the thing in dispute. The venues that build it will look, increasingly, like regulated exchanges — licensed, cleared, watched. The offshore books that don't will keep breeding the scandals, and keep finding out about their best traders from the federal indictment.