Recently, there has been an interesting story about anomalies in the prediction markets worth noting. Six accounts on Polymarket are suspected of insider trading and managed to make around $1.2 million in a highly suspicious manner.

What makes it interesting is their pattern. According to analysis from Bubblemaps, all these accounts were created in February, funded within 24 hours before the U.S. attack on Iran occurred, and immediately bought large positions in the prediction market for that strike. One account even bought over 560,000 shares at 10.8 cents per share, then cashed out nearly $560,000 when the market settled at $1. All others bought 150,000 shares at 20 cents and gained six-figure profits. All of this happened just a few hours before the attack was reported in Tehran.

The most suspicious part is that none of them had any other activity on the platform—only these bets. Bubblemaps even created visualizations showing these six wallets clustered together and funded through similar channels. Out of 2 million Polymarket users, only 10 exhibited this pattern, but the results are highly significant.

The attack itself had a severe impact on the market. Bitcoin dropped significantly, while oil futures on Hyperliquid rose 5%. Trading volume in the February 28 contract reached nearly $90 million, part of over $529 million wagered on related markets since December.

This is not the first case. The week before, Kalshi, a ( competitor of Polymarket), already took action against two users for insider trading, including a visual editor from Beast Games suspected of trading based on unpublished event results. Kalshi banned them for two years and fined over $20,000. They also have more than a dozen active investigations.

The CFTC has issued warnings that insider trading in prediction markets could violate U.S. laws. Regulators see exchanges as the first line of defense in preventing this.

The most ironic case is another where Polymarket traders are suspected of insider trading on a market designed to catch insider trading itself. When ZachXBT revealed he would release an investigation about the crypto platform (Axiom), some people clearly already knew the answer. They immediately started aggressive betting on Polymarket contracts about which company would be mentioned, and Lookonchain identified 12 wallets doing this before the announcement.

This shows that as blockchain adoption grows, the available metadata for machine learning models also increases. Privacy based on obfuscation structurally degrades. Prediction markets are becoming a new playground to test how effective regulatory enforcement is in the crypto era.