The $172,000 Insider Fine That Explains Kalshi's $22 Billion Bet

Gabriel Perez didn't need a model, a hunch, or a fast terminal. He needed the teleprompter.

Perez had run Trump's teleprompter since 2016 — the technical assistant whose job put the president's words on the screen before the president said them. "In his position," as the CFTC put it, "Perez had access to presidential speeches prior to those speeches being delivered," and he misappropriated that access. Between December and February, on Kalshi, the CFTC-regulated prediction market, he bet on which words and phrases Trump would use in upcoming addresses, and the regulator says he cleared $107,539.02. The settlement announced Friday ordered him to give that back, pay a $65,000 civil penalty, and stay off CFTC-regulated trading venues for three years. Rounding grabs the headline: $172,000.

The tabloid version of this story is a staffer with sticky fingers, gone for betting on his boss. The investment version is stranger and more useful: it is the CFTC's cleanest demonstration yet of a design flaw in the prediction-market boom, and of exactly why the industry's biggest players have to fix it. The numbers that matter are not in the fine.

How a script beats a market

Kalshi's "mention markets" are event betting in its purest form. A contract pays $1 if a given word or phrase appears in a given speech, and nothing if it doesn't. Will Trump say "Somalia"? Will he say "SAVE Act"? The State of the Union produced a cluster of these contracts that drew about $11.97 million in volume across 391,165 trades over the 46 days before the speech — and here's the tell: roughly 79 percent of that money arrived on the night of the speech itself. Before speech night, a mention contract is a thin pool of public guesswork, with average trades in the tens of dollars.

The thinness is the opportunity. If you have the text, you buy "yes" on the words you know are in it while the crowd is pricing your certainty at twenty or thirty cents; on speech night, when the whole market rushes in, the contract converges on its dollar and you collect the spread. No leverage, no model — certainty arbitraged across time. ABC News reported that Perez dominated trading in the speech contracts across five of the most market-moving moments of Trump's second term, and the CFTC's order notes that he traded one to seven markets per event, even selling positions when the president went off-script. His profit came from the other side of the book: the market makers and casual traders who had sold him those contracts at guess-prices. The only trader in the market who knew the right price for certain was the one trader the other side could never beat.

That is the paradox at the center of prediction markets. Their entire pitch is that they pay people to be right. The teleprompter operator is that pitch taken to its logical end — and the moment it works perfectly, the crowd that provides the other side of the trade can no longer trust the prices it is betting against.

Why the exchange turned him in

Here the scandal framing is backwards: the regulator did not catch Perez. The exchange did. Kalshi's onboarding and surveillance systems flagged the pattern, market makers had separately complained through whistleblower channels, and Kalshi interviewed Perez, froze his account with more than $90,000 of profit still in it, and handed the evidence to the CFTC, which thanked the exchange for its assistance.

That is not corporate charity. It is the incentive structure doing its job. Kalshi is a private company at the center of a boom — prediction-market monthly volume grew from about $1.2 billion in early 2025 to more than $20 billion by January 2026 — and it raised roughly $1 billion this spring at a $22 billion valuation. Its entire worth rests on one asset: the belief that its prices are honest. The firm is simultaneously fending off states that call it gambling — Nevada sued in February, Arizona's attorney general in March — while a friendly new CFTC chairman, Michael Selig, filed a brief in February arguing the venues are legitimate federally regulated markets. In that position, one unpunished story of "a White House insider traded on tomorrow's speech" is existential. A caught one is the best advertising the firm has ever bought. The CFTC even priced the settlement at a "substantial discount" on account of Perez's "exemplary cooperation" — a smaller fine, because a defecting insider is worth more than a jailed one.

Verdict: the $107,539 clawback is not the cost of the scandal. It is a rounding error next to a $22 billion valuation, and a discount next to what the scandal would have cost. The fine was never the point; the freeze-and-refer was.

He's a pattern, not an outlier

Do not read "case closed" from the cooperation discount. Perez is one node in a pattern federal prosecutors were already chasing. In April, a U.S. soldier, Gannon Ken Van Dyke, was charged with turning about $33,000 into $400,000 betting on the capture of Nicolás Maduro using classified information. In late August, authorities were reported to be preparing charges against a U.S. servicemember who made over $1 million wagering on military strikes, and a KPMG employee betting on corporate earnings beats was under scrutiny. Data firm Bubblemaps identified nine connected Polymarket accounts that took in $2.4 million betting almost exclusively on U.S. military actions, and Polymarket itself has referred dozens of accounts to the Justice Department.

And look at the specific product Perez played. In mid-August it was reported that federal regulators were investigating whether mention markets are, in one official's words, "potentially very easy to manipulate" — and that Kalshi had quietly shut down its sports mention markets, with a source saying they would not return "until further notice." The category that made the teleprompter play possible may be removed rather than merely policed. That is the adaptation loop running at full speed: contract design, exploit, investigation, deletion.

What it means for your money

Two practical reads, one for the prediction-market trade and one for everything else.

The prediction-market trade. Kalshi and Polymarket are private; the public-market handles are the companies plugged into them. Robinhood's deal put event contracts in front of roughly 27 million funded brokerage accounts, and NYSE owner Intercontinental Exchange has committed up to $2 billion to Polymarket at an $8 billion valuation. For an investor in that theme, the Perez case cuts both ways at once: bullish, because it proves the integrity machinery works and the regulated venues can turn a scandal into a moat; bearish, because the soldier, the servicemember, and the $2.4 million cluster all traded for months before the machinery caught up. The swing variable is speed — can the surveillance-and-disclosure layer thicken faster than the payoffs for knowing first grow? So far the payoffs are winning.

The general lesson. What Perez did is visible in every young, thin, information-heavy market: the person with early access to the answer is on the other side of your trade until coverage catches up. In a mature public stock market, the machinery that protects you — insider-trading law, surveillance, disclosure, exchanges that turn themselves in — is invisible, and its cost is built into the price. Prediction markets show you the early stage, where that machinery is being bolted on fine by fine, scandal by scandal. Treat their prices accordingly: as information, and simultaneously as a target for people who know more than you.

The teleprompter operator is now out of a White House job that paid $175,000 a year, banned for three years, and stripped of $172,539 — the profit already frozen by a platform he'd beaten for months. In one sense he's the story's smallest character: one insider, five speeches, one product line. But he is proof of how the whole edifice holds together — not because insiders don't try, but because the people who run these exchanges now have more at stake in catching them than the insiders had in not getting caught. Watch that ratio, not the scandal. It is what decides whether $20 billion a month of prediction markets becomes financial plumbing, or a casino with excellent surveillance.