UPDATE: Jeff Probst has responded to Variety’s report on reality show spoilers about Calci and Polymarket, saying prediction market platforms “encourage people to lie, cheat, and steal to get ahead.”
Carsi responded to Probst’s sentiments in a new statement, saying, “If the market volume for ‘Survivor’ is any indication, this is not going to stop people from being interested in ‘Survivor.’ That being said, we are considering adding product features to prevent spoilers.”
PREVIOUSLY: The winner of CBS’ “Survivor 50” was announced live on May 20th.
But traders on prediction market platforms Calci and Polymarket, as well as “Survivor” fans exposed to news reports of changing odds, saw Aubrey Bracco’s payday coming before the season even started.
When the “Survivor 50” market was held in Calci six weeks before the Feb. 25 season premiere, preliminary trading predicted Bracco had a 61% chance of winning, nearly double the odds of his next closest rival. By January 28, her percentage had jumped to 83%. Then, on February 28, just days after the premiere and before Bracco had accumulated much screen time, a user bet $45,500 (later making a profit of $4,500) that Bracco would win the season. By the market close on the day of the final game, a staggering $32.7 million in bets had raised her odds to 97%. (Polymarket’s predictions for “Survivor 50” called for just $1.9 million in volume, but the odds followed a similar trajectory.)
To those who followed, it seemed almost certain that Bracco would be crowned the sole survivor. A few minutes before the finale, Kalsi sent a push notification asking, “Have you decided on the island yet?”
Push notification Kalsi sent to users minutes before the “Survivor 50” finale.
Pre-recorded reality shows are facing unexpected problems. It’s that prediction markets are ruining the outcome. It’s not just “Survivor.” Kalsi users correctly predicted 91% of the time that Galaxy Girl would win this season of “The Masked Singer,” three months before Ashlee Simpson took off her space-themed costume for the finale. The site similarly spoiled the winner of Next Level Chef in February, but the season didn’t end until May 21st. And ill-fated Bachelorette Season 22 contestant Doug Mason had ratings of 89.5% on Polymarket and 93% on Carsi before ABC canceled the premiere. The season may never air, but keep track of the money to find out who received the final rose.
There’s a long history of unscripted shows being spoiled on the internet through blogs like RealitySteve and Reddit pages like SpoiledSurvivor. Networks tend to assume these leaks come from cast members. However, the “Survivor 50” contestants were sent home in July 2025 without knowing the winner. We had to wait until the live finale to see host Jeff Probst read the votes on stage. This meant that only a select number of “Survivor” staff knew of Bracco’s win, and CBS was forced to consider the fact that the leak must have come from on-set, sources said.
Insider trading is prohibited on both Kalsi and Polymarket, and Kalsi head of communications Elizabeth Diana said the platform uses AI-powered third-party monitoring to detect suspicious transactions. The company has repeatedly said it is exploring the “Survivor” market. (Polimarket representatives did not respond to requests for comment.)
“Traders with large positions appear to have no ties to the show or network,” Carsi executive director Robert Denor wrote about X after the finale.
Speaking about the depth of our research, Diana says, “We don’t just look at people in the households of the people we’re transacting with. We also look at their social activities to see what they’re doing on the site and who they’re connecting with.” If a trade reported by the AI is deemed anomalous, a Kalsi representative will interview the trader to assess fraud. The company publicly cracked down on five violators, including a video editor who worked for MrBeast.
If insider trading isn’t happening in these reality TV markets, as Kalsi claims, how did users’ predictions become so accurate? The company says it’s because traders are relying on spoilers already circulating online. Whether those spoilers came from insiders is irrelevant to Karshi executives unless the traders themselves broke the law.
“Reality Steve obviously gets talked into by people and is probably leaking information. It’s not good, but he puts it online,” Diana says. Kalsi’s investigation primarily found that traders in the “Survivor” market were following information from a Reddit user known as “Lifetime Robot,” who has a strong track record of accurately ruining the show. Denor writes, “Public rumors are not the same as insider information.”
Still, these rumors came with spoiler warnings, making it easy for casual viewers to tune out. Carsi and Polimarket betting odds are proliferating on social media right now, and considering the amount of money behind the predictions, they’re hard to ignore.
“It used to be plausible to deny it. If someone said Aubrey won, you might think, ‘I don’t know what you’re talking about,'” says one Carsi trader who made about $500 in profits trading on Bracco. “Now people are betting thousands of dollars on it, and people are taking it more seriously. It’s taken away that layer of being able to cover your eyes and pretend.”
This new conviction is causing major headaches for the studio. One industry insider is optimistic, suggesting a multi-million dollar prediction market could help build excitement and valuations. But many fear that spoilers will take away from the suspense and turn viewers off entirely.
Officials at CBS parent company Paramount said there were executive-level conversations about insider trading in prediction markets. The problem is that it’s nearly impossible for studios to stop it, sources say. While corporate legal and compliance departments can enforce strict policies regarding stock exchanges, effectively extending insider trading governance and compliance technology to prediction markets requires significant capital, time, and corporate infrastructure.
“The tools aren’t ready yet,” a Paramount source says. “The market is beyond regulation.”
Networks may explicitly prohibit involvement in prediction markets in NDAs signed by production staff. But this would be difficult to enforce, given that most reality TV crews are independent contractors, not salaried employees monitored by compliance technology. If that fails, studios could prevent leaks by rebuilding the show and removing pre-recorded elements entirely, but that could become an even bigger logistical nightmare.
The legal battle over prediction markets is ongoing at the state level, with Minnesota becoming the first state to ban prediction markets, but the federal government has yet to address insider trading on the platforms in a meaningful way.
Therefore, trading is not slowing down. The ‘Survivor’ season 49 Carsi market that accurately predicted Savannah Louie’s win brought in just $2.4 million. Season 50 saw a 1,284% increase in sales volume.
Paramount sources suggest one potential solution. Studios could partner with Carsi and Polymarket, perhaps collaborating on certain types of markets related to their shows, and promote them on-air instead of shutting down other spoiler-filled markets.
However, prediction market platforms do not necessarily actively conduct such trades. “We cannot compromise on the trader experience,” says Diana. “Traders are the people who keep the site running, so we must always prioritize their needs over those of our partners.”
So when it comes to stopping leaks, Paramount officials said studios should be left on their own. “The only viable path forward is for companies to police themselves.”
Until then, in the words of “Survivor” icon Sue Hawk, “This island is full of snakes and rats.”
