Original | Odaily Planet Daily (@OdailyChina)
Author | Golem (@web3_golem)
Recently, another insider trading scandal has been exposed at the U.S. White House.
A White House staffer made hundreds of thousands of dollars in profit from insider trading on prediction markets; the insider was revealed to be a long-time operator of Trump’s teleprompter. The employee has since been suspended and had their salary halted.
This prompt operator is the third insider disclosed by the U.S. Department of Justice to have used non-public information to profit significantly on prediction markets, following the special forces operative involved in the Maduro capture operation and the Google security engineer. （Related reading: After 4 Months, Polymarket Helped Trump Identify the Military Leak—But at What Cost?》《Looking at the Answers Before Submitting? Google Engineer Entangled in Polymarket Insider Trading Case》
Reported voluntarily by Kalshi, funds were frozen, but no criminal liability was incurred.
The main character is Gabriel Perez, who has been operating the teleprompter for Trump's speeches since 2016. Perez’s path to securing this job was quite dramatic: in 2016, Trump’s campaign team urgently needed a teleprompter operator and, while searching Google for “teleprompter,” stumbled upon Perez’s company, leading to his hiring by the Trump team.
Gabriel Perez
Although Pérez was hired by chance, over the past decade he has gradually become one of Trump’s closest aides. U.S. Politico even described Pérez as “Trump’s only trusted person,” noting that he frequently receives last-minute revisions to public speeches directly from Trump himself.
As a result, Pérez became one of the few individuals to receive Trump’s full speech drafts in advance and held final approval authority over nearly all of Trump’s prepared remarks. This power is not insignificant: Pérez’s official title in the White House is Deputy Assistant to the President and Technical Advisor, with an annual salary of $175,000—just $20,000 less than senior staff such as Chief of Staff Susie Wiles and Press Secretary Karine Jean-Pierre.
This salary already places one in the high-income bracket in the United States, but the greedy Pérez is still not satisfied.
As prediction markets gained popularity, countless players began betting on which specific words Trump would "mention" in a particular speech, and Pérez realized his "privileged" information could bring him even greater wealth.
CFTC investigators found that Perez placed bets on more than a dozen of Trump’s speeches over a period of approximately three months, earning over $100,000 in total profits. These included Trump’s prime-time speech in December last year, his address at the World Economic Forum in Davos, Switzerland, in January this year, his State of the Union address in February, and his speech at the Medal of Honor ceremony in March.
The U.S. president's statutory salary is $400,000, and with various allowances, the president earns approximately $569,000 annually. If Perez had not been caught, at his rate of earning $100,000 in three months, his annual income would exceed the president's salary, despite having less power.
But even with advance knowledge of the speech content, Perez could not consistently predict which words Trump would mention during the speech, as Trump often deviates from the script to improvise. When Trump skips a word that Perez has bet on, he immediately sells to cut losses. Trump himself acknowledged during a January speech at the Detroit Economic Club that he doesn’t look at the teleprompter 80% of the time.
Like the experiences of special forces operatives and Google security engineers, Perez's exposure stemmed from proactive reporting by a prediction market platform. Perez frequently used Kalshi for insider trading; starting in March of this year, Kalshi’s monitoring system detected unusual trades linked to specific words mentioned in Trump’s speeches, which drew attention to Perez.
Following the internal investigation, Kalshi swiftly froze over $90,000 in Perez’s account and transferred it to the Commodity Futures Trading Commission (CFTC). Upon learning of the incident, Trump called it “disgraceful,” personally decided to suspend Perez, and withheld his salary during the suspension.
Due to greed, Perez ultimately lost both his wife and his soldier—he failed to recover his profits from the prediction market and lost his original job. However, compared to special forces soldiers and Google security engineers, Perez was fortunate, as U.S. authorities did not bring criminal charges against him, so he does not face imprisonment.
During the investigation, the CFTC notified federal prosecutors in Manhattan, but the prosecutors declined to pursue a criminal investigation. According to knowledgeable sources, CFTC regulators have expressed willingness to reach a settlement with Perez and have discussed terms with him, resulting in a requirement for Perez to return his profits and cease similar transactions going forward.
Perez is just the beginning of clearing out those who mention market insiders.
Perez does not face jail time because prosecutors determined that Perez did not commit a criminal offense—he did not prematurely leak sensitive government information or pose a threat to national security. As Trump said, it was merely “disgraceful,” undermining the integrity of government officials.
In March this year, the White House warned staff not to place bets on prediction markets using non-public information, and White House spokesperson Davis Ingle said: “The White House has strict ethical guidelines, and we expect all staff and officials to adhere to them.”
But Perez is certainly not the only staff member in the White House who has profited from insider information; Trump, who openly runs a paid group for himself, has even less grounds to criticize this teleprompter operator（Related reading:Trump Starts Selling “Alpha” for $100,000 a Month）.
It’s no surprise Perez succumbed to the temptation—the “mention” markets in prediction markets are indeed the easiest to manipulate. When the cost of participation for insiders is driven to near zero while the potential returns are extremely high, it ceases to be a moral issue and becomes one of mechanism design. Even politicians who appear respectable and hold righteous positions cannot guarantee they will never cross that line when faced with financial incentives.
The "Mention" market allows users to bet on specific words, phrases, or topics that will be mentioned in public speeches. Compared to other events such as political elections or sports matches, the cost of cheating in "Mention" markets is extremely low. It’s not limited to individuals like Perez who can learn the speech content in advance— for the speaker themselves, cheating is as simple as saying one sentence, making the phrase "a word is worth a thousand gold" a tangible reality.
At this year’s Grammy Awards in February, host Trevor Noah said, “Welcome back to the Grammys,” then suddenly shouted, “Potato.” As everyone was confused, Trevor continued, “If you bet on me saying that word on Polymarket, you just hit the jackpot,” and congratulated the user “Noah 22.” In reality, “potato” was never an option in Polymarket’s prediction market for “What will be mentioned at the Grammy Awards,” and the user “noah-22” was entirely fictional.
The Grammy host shouted "potato" during the awards ceremony.
Some post-event analyses suggested this was a Polymarket marketing campaign, but it already demonstrated the party's ability to manipulate markets based on mentions.
Another more direct example: In October 2025, during Coinbase’s Q3 earnings call, just before the call ended, CEO Brian Armstrong said he noticed many people were betting on prediction markets about what he would mention during the call, so he opened Polymarket and read out every option aloud, resulting in all market outcomes ending with a 100% probability and a tie.
These are just two examples illustrating how individuals exert influence over the “mention” market; undoubtedly, many more people profiting from this practice remain hidden beneath the surface. However, as regulatory oversight of the market continues to strengthen, insiders in the “mention” market may eventually be fully eradicated—Perez is just the beginning.
Last month, Kalshi updated its policy to require users to disclose their place of employment. Bobby DeNault, Kalshi’s head of enforcement, explained that the rationale is: “If you have access to certain information due to your job or employment relationship and are legally obligated not to misuse it, you have a duty not to appropriate or exploit that information for personal gain.” Polymarket has not yet imposed such strict disclosure requirements on users, but as the prediction market sector intensifies its focus on compliance, stricter regulations from Polymarket are likely on the horizon.
From special forces operators and Google engineers to White House teleprompter operators, prediction markets are gradually cleaning up insider trading. At the same time, the market is undergoing a demystification of prediction markets—once thought to reflect collective wisdom, they are in fact merely ATMs for a select few insiders.
Although cracking down on insider trading makes prediction markets more compliant, it also moves them further away from the truth and closer to pure gambling.