Interesting developments in the prediction markets. I recently learned about an incident on Polymarket that clearly demonstrates how serious the insider problem is on decentralized platforms.

Six accounts made about $1.2 million by almost perfectly predicting the U.S. strike on Iran on February 28. According to blockchain analysts from Bubblemaps, these wallets were funded literally 24 hours before the event and immediately bought "Yes" positions on the U.S. strike market. No activity outside of this contract. One of them invested only $56,000 in the position and earned $560,000 in profit. This no longer looks like a coincidence.

What surprised me is that this isn't the first such case. It turns out insiders are actively working on a competing platform, Kalshi. This week, they penalized a visual effects specialist from the MrBeast show who traded based on information about the show's results. The fine was over $20,000 and a two-year ban. Kalshi has already conducted around 200 investigations of similar cases.

Regulators have clearly become concerned. The CFTC issued a warning that insider trading of event contracts violates U.S. law. The commission's chairperson said that exchanges themselves are the first line of defense against such violations.

And here’s the funniest part — insiders have started trading even on markets created specifically to catch them. ZachXBT announced an investigation into a crypto platform, and immediately a contract was created on Polymarket about which company would be named. Lookonchain identified 12 wallets that clearly knew the answer and actively bet on Axiom even before the results were published. Insiders have learned to even play on the very information about insider investigations.

This shows that the problem is much deeper than just individual violators. When prediction markets have such volumes — Polymarket traded nearly $90 million on this contract — the risk of informational asymmetry becomes systemic. Platforms will have to take a more serious approach to verifying and monitoring activity.