A teleprompter operator's insider betting scheme has exposed a major flaw in prediction market oversight. Gabriel Perez, who controlled what Donald Trump would read during speeches, allegedly wagered on the content of at least 12 speeches over three months on Kalshi, a CFTC-regulated prediction exchange. Over $10 million traded on these markets.
Perez held a massive informational edge. As the operator controlling the teleprompter display, he knew Trump's exact remarks before delivery. He exploited this position to place bets on specific speech outcomes, essentially betting on information only he possessed. CFTC investigators documented the scheme and flagged the trades.
Trump called the situation a "disgrace" and placed Perez on unpaid administrative leave. The incident raises hard questions about market integrity at Kalshi, which has aggressively expanded political prediction markets in recent years. Kalshi operates under CFTC approval as a designated contract market, meaning regulators granted it authority to list binary options on political events.
The scandal mirrors insider trading concerns in traditional financial markets, except prediction markets operate with lighter regulatory touch. Traders bet on outcomes they can actually influence or predict through nonpublic information. Kalshi's systems apparently failed to detect or flag suspicious activity from someone with direct access to speech content.
This matters beyond the specifics. Prediction markets have gained credibility among serious bettors and institutions as legitimate price-discovery mechanisms. But this breach shows the markets remain vulnerable to manipulation by insiders with structural advantages. A teleprompter operator shouldn't be able to accumulate positions based on advance knowledge of presidential remarks.
The CFTC now faces pressure to tighten oversight of political prediction markets. Kalshi will need to implement controls detecting trades from individuals with material nonpublic information tied to market-moving events. The $10 million
