The George Santos Settlement Shows What Prediction Markets Can't Fix

The media is treating the George Santos CFTC settlement like a cautionary tale about one disgraced former congressman who tried to game a system and got caught. Santos agreed on July 31 to pay $35,000-$17,500 in penalties plus disgorgement of the $17,569.98 he profited-and accept a three-year ban from trading on prediction markets. The CFTC said he engaged in "manipulative activity" betting on Kalshi that he would not attend the State of the Union, after publicly saying he would.

The real story is the $17,570. That is how much profit someone risks federal investigation, a multi-year trading ban, and lasting public humiliation to make. The incentive structure of prediction markets rewards information arbitrage at every scale. You don't need a Fortune 500 edge. You just need to know something the market doesn't.

The insider trading paradox

The fundamental problem with prediction markets is not a compliance gap. It is a structural paradox: the participants who are most valuable to the market are the ones it must exclude.

Prediction markets are supposed to aggregate dispersed information into a single price. Their entire theoretical value proposition is that people with private knowledge will trade on it, compressing that information into an observable probability. Robin Hanson, who first proposed prediction markets as a forecasting mechanism decades ago, has long explained this tradeoff. The social value of prediction markets derives from financially incentivizing insiders to reveal confidential information. But the moment insiders trade on material non-public information, the market becomes what regulators call "manipulative" and what the rest of the financial system calls illegal insider trading.

The paradox is not a theoretical exercise. Over the past six months, it has produced a sequence of enforcement actions that reads like a stress test:

• The January charging of U.S. Army master sergeant Gannon Van Dyke by the DOJ for using classified information about a military operation to capture Nicolás Maduro to profit over $400,000 on Polymarket.
• Kalshi's April suspension and fining of three congressional candidates-Ezekiel Enriquez in Texas, Matt Klein in Minnesota, and Mark Moran in Virginia-for betting on their own election outcomes.
• Kalshi's referral of Santos to the DOJ and CFTC after detecting his trades in February, followed by the July settlement.

In each case, the person with the most direct knowledge of the event was also the person trading on it. The soldier knew about the operation. The candidates knew whether they'd win. Santos knew whether he'd show up. These aren't edge cases. They are the natural attractors of the market structure.

The compliance response is a patch, not a solution

Kalshi has been scrambling to build institutional-grade guardrails. In March 2026, it announced new rules designed to prevent politicians from trading on their own campaigns and athletes from trading on their own sports leagues. In June, it went further, mandating employment disclosures for users trading on "sensitive" contracts, launching a whistleblower portal, and assigning risk scores to markets based on factors like corporate performance metrics, national security implications, and manipulation potential.

Kalshi's head of enforcement, Robert DeNault, has been the most vocal advocate for market integrity in the industry. He flagged Santos's trades, referred him to federal authorities, and said the company would pursue its own enforcement action and reimburse affected traders. He called the congressional candidates' wagers "political insider trading" and characterized the Moran case-which involved a candidate betting on his own candidacy before announcing his run-as a violation of the principle that "direct decision makers" shouldn't trade on their own outcomes.

All of this is to Kalshi's credit. But compliance measures built on top of a market where insider information is the primary driver of price discovery are inherently reactive. Employment disclosures require traders to self-report their conflicts. Risk scores are Kalshi's judgment call about which markets are sensitive. Whistleblower portals depend on other traders noticing something suspicious. These are detection mechanisms, not prevention mechanisms. They shift the problem from "can anyone exploit insider knowledge" to "can we catch the people who do."

The CFTC has been pushing in the same direction. It issued a staff advisory in February 2026 emphasizing that exchanges must ensure contracts are not "readily susceptible to manipulation" and reiterating that insider trading is prohibited under existing anti-fraud authorities. In March, the agency followed up with guidance that Kalshi and Polymarket both cited when updating their rules. Congress has introduced proposals to bar government officials from trading on information obtained through their roles.

The participant ecology

Prediction markets lack the natural participant balance that sustains other derivative markets. In wheat futures, farmers are naturally long and cereal manufacturers are naturally short. The market exists because real economic actors need to transfer risk, and speculators provide liquidity in between. The participants who stand to profit the most from insider information are constrained by the existence of other participants who are trading for genuine economic reasons.

Prediction markets don't have that structure. There are no natural hedgers buying "yes" on the Santos State of the Union contract because they had economic exposure to his attendance. There was no cereal manufacturer equivalent offsetting the soldier who knew about the Maduro operation. The entire market consisted of people guessing about an outcome-and one person who knew.

When a market has no natural buyers on one side, it is entirely composed of people trading information. And the person with the best information always has an incentive to trade before the market prices it in. That is not a bug in the system. That is what gives the system value.

Santos himself acknowledged this on his podcast in March, after the social media backlash began: "I guess people lost money. Some people made unexpected money. That's to show you how fragile these markets are." He's wrong about the fragility. The markets aren't fragile. They're working exactly as designed. The person with private information traded on it and profited. The market mechanism did its job.

The problem is that the market's job conflicts with the legal and regulatory framework it now operates within.

What changes the view

Prediction markets could resolve this paradox in one of two ways. They could accept insider trading as a feature and build markets where private information is legally disclosed through trading-a direction that would require fundamentally rewriting how the CFTC and DOJ approach commodities fraud. Or they could accept insider trading as a cost and build compliance infrastructure dense enough to make it unprofitable-which would require restricting participation so heavily that the information-aggregation function deteriorates.

Neither path has been chosen. Instead, we're in the middle ground where Kalshi's DeNault says "for more serious matters, we refer cases to the CFTC or DOJ," implying a spectrum of enforcement severity. But the Santos settlement, the congressional candidate bans, and the soldier's indictment all suggest that the current approach is case-by-case detection followed by retroactive punishment. That is not a sustainable governance model for a market whose growth depends on the perception that it produces honest prices.

Verdict: The Santos settlement is not about George Santos. It's about the structural impossibility at the center of prediction markets. The participants who provide the most value to these markets-people with private, outcome-affecting knowledge-are the same participants who legally shouldn't be trading. Employment disclosures, risk scores, and whistleblower portals are compliance bandages on a market structure where insider information is both the product and the threat. Until the industry and its regulators confront that paradox directly, enforcement will remain a game of whack-a-mole, and the $17,570 profit that Santos made will look less like a crime and more like a feature.