NYT journalist Adam Iscoe on the hidden world of prediction market sharps — and why Wall Street keeps losing to them

Prediction market sharps are beating Wall Street — and Jeff Yass admits it

Adam Iscoe spent five years at The New Yorker before joining Notion. His most recent reporting focus was prediction markets, and the central finding is blunt: a small group of individual traders — a few hundred people at most — are consistently taking money from both casual users and institutional funds.

Jeff Yass, co-founder of Susquehanna, told Iscoe directly that SIG is getting beaten by these sharps. SIG has tried to hire them, but at least one declined the offer. The reason is instructive: individual sharps can use techniques — scraping websites, running unconventional models — that corporate compliance policies at a firm like SIG would prohibit. One trader Iscoe profiles built models predicting Rotten Tomatoes scores for films and has made seven figures doing it, from a $600 Lenovo laptop, without an Ivy League degree or a Wall Street contact.

Another source Iscoe identifies only by the screen name Frozen, a grad student, turned $200 into $500,000. Frozen's own description of the market structure: a large rotating group of people join, bet, and lose, while a couple hundred sharp traders consistently win. That's most of the story.

“There's a group of a couple hundred guys winning, and that's the whole story... he self-described as sort of like a dipshit from the Midwest. I didn't go to an Ivy League school. And I'm able to outcompete Wall Street with a $600 Lenovo laptop. I asked Jeff Yas, who co-founded SIG, and he said, 'We are just getting taken for a ride. We're just getting crushed by these sharps.'”

Market structure

The average participant is getting decimated. The sharps sit above them, using informational edges built through genuine research — modelling weather data, canvassing voters, scraping public sources — rather than insider access. Above the sharps sits institutional money, which is losing to the sharps despite superior resources, primarily because of the constraints that come with operating at scale inside a regulated firm.

Iscoe distinguishes sharps from bad actors. He spent time in Texas with a group called the MAGA QB Club, roughly ten people who knocked on nearly a thousand doors to model a Democratic primary and got it right. That kind of ground-level information gathering, he argues, is what makes markets more efficient. The Venezuela case — a military-connected individual apparently trading ahead of a strike — is a different category, one the Justice Department and CFTC are actively prosecuting.

Regulatory window

The Trump administration has been favorable to Kalshi, Polymarket, and a more permissive CFTC posture toward prediction markets generally. Both Kalshi and Polymarket told Iscoe they are building AI systems to detect insider trading. Whether that enforcement is real or performative is an open question. The Council on Foreign Relations is among the bodies working through what sensible regulation might look like.

Iscoe flags a meaningful risk: if there is a change of administration in 2028 or 2029, the regulatory environment could shift. In the near term, the markets are expanding faster than oversight.

One documented case Iscoe mentions: a Polymarket trader was found betting on Google search results while employed at Google, giving him access to the exact data the market was pricing — a clear terms-of-service violation caught by an independent watchdog and reported by the Journal.

Iscoe left journalism to join Notion, convinced he couldn't develop the standing to cover AI credibly from the outside. His plan there is to ask questions and tell stories about what AI looks like in the economy — less marketing, more reporting from inside the industry.