In the chaos of a presidential rally, where every word is scripted and every pause calculated, we found a winter soul. Josh Perez, a White House teleprompter operator, turned his access into a personal arbitrage machine on Kalshi, the CFTC-regulated prediction market. Over multiple speeches, including the State of the Union, Perez placed bets on whether Trump would mention specific words like "tariffs" or "border," winning over $100,000. He even withdrew his positions mid-speech when he knew the word was not coming—a real-time information advantage that no automated algorithm could match.
This is not a story about a rogue employee. It is a story about the fundamental tension between market efficiency and ethical governance in a decentralized world.
Kalshi, founded by Tarek Mansour and Luana Lopes Lara, operates as a designated contract market under CFTC oversight. Unlike Polymarket's permissionless chain, Kalshi requires KYC and employer disclosure—the latter introduced just last month after this incident came to light. The platform's "Mentions" market allows users to wager on whether a public figure will utter specific terms during an event. It is a niche that seems harmless until someone with privileged information steps in.
When Kalshi's monitoring team flagged Perez's trades and reported them to the CFTC, they did what any compliant platform should. Yet the question remains: why did the detection come only after tens of thousands of dollars in profit had been extracted? The answer lies in the inherent asymmetry of prediction markets: they are designed to aggregate public information, but they cannot, by themselves, distinguish between a well-reasoned forecast and insider knowledge.
Based on my experience auditing governance models in DeFi—specifically a notorious protocol where whale wallets bypassed consensus through a flawed quadratic voting mechanism—I see a similar pattern here. Code is law, but conscience is the compiler. The Kalshi incident reveals that technical safeguards, such as employer disclosure and trade patterns analysis, are necessary but insufficient. They are reactive filters, not proactive gatekeepers.
The core insight is this: prediction markets rely on an assumption of fair play that breaks down the moment a participant gains non-public access. The "Mentions" market is especially vulnerable because it rewards speed of execution based on real-time knowledge. Perez did not need to know the speech's full content; he only needed to know what was on the teleprompter at that moment. That is a form of high-frequency insider trading that cannot be coded away.
What Kalshi did right—flagging and reporting—is commendable, but it also underscores a deeper structural issue. The platform's employer disclosure rule, implemented after the fact, is like adding a lock after the burglary. The real solution lies in rethinking market design itself. Could we weight each participant's position by their access to privileged information? Or require that all bets be placed before the event's start, eliminating mid-event entry? These are not technical patches but governance interventions.
The contrarian angle: Many will argue that this scandal proves prediction markets are inherently corrupt and should be tightly regulated. I disagree. This event actually validates the regulatory framework. Kalshi self-reported, the CFTC is negotiating a settlement, and Perez faces consequences. That is a functioning system. The real blind spot is our over-reliance on code as a substitute for ethical judgment. We treat governance as a vote, but it is a vigil.
The temptation is to automate detection—use AI to flag suspicious trades, analyze speech transcripts in real time, cross-reference with employee databases. But automation alone cannot adjudicate intent. What if Perez believed his trades were permissible because he was not trading company stock but rather "trivia" about speech content? The line between forecast and inside information is blurry precisely because prediction markets are designed to reward accurate predictions, regardless of the source.
Silence in the bear market is where truth compiles. In a bull market of prediction market hype, we must resist the allure of pure automation. The Kalshi case reminds us that human oversight—a governance architect's role—is not a bottleneck but a necessary filter. We need hybrid governance: algorithmic monitoring to generate alerts, followed by human ethical review to determine intent. This is not less efficient; it is more resilient.
The takeaway is not that prediction markets are broken. It is that we must build them with the humility to admit that code cannot replace conscience. As I wrote in my 2017 audit of EtherSwap: "Code is not law if power is centralized." Here, the power is information, and the law is the market's integrity.
Forward-looking thought: If CFTC formalizes rules for prediction market insider trading, it could set a precedent for all event contracts—from sports to elections. The winners will be platforms that embed ethical governance from day one, not as an afterthought. The question is: when your competition offers speed and anonymity, can you afford to prioritize fairness? The answer, for those who believe in the long-term value of trust, is yes.