The alleged $400,000 insider bet by a Green Beret on a Venezuelan raid initially seemed like an isolated incident. However, a new report by the Anti-Corruption Data Collective (ACDC) indicates it may be just the tip of the iceberg. Analyzing over 435,000 markets and $54.4 billion in cumulative volume on Polymarket from January 2021 to mid-March 2026, the nonprofit research group found that low-probability bets on military and defense outcomes have success rates that are difficult to attribute to skill or luck alone. Typically, such 'longshot' bets succeed about 14% of the time across political markets, but in military-linked contracts, success rates have reached as high as 50% in some cases.
The authors of the report note that markets tied to specific government policies, including military and defense and foreign affairs, are particularly challenging to forecast using publicly available information. This makes them more vulnerable to information asymmetries, including insider trading or specialized knowledge. The gap between informed and uninformed traders may be widest in these markets, creating an environment where a small group can consistently outperform others, not just by reacting faster, but by possessing more knowledge. Polymarket emphasizes its market surveillance teams and cooperation with the Department of Justice, particularly in the Venezuela case, and prohibits trading on confidential knowledge on its platform, as does Kalshi.
The report's findings contribute to a growing body of research pointing in the same direction. A working paper from the London Business School and Yale found that roughly 3% of traders are responsible for most price discovery on Polymarket. Separate analysis by blockchain analytics firm Solidus Labs showed that profits are even more concentrated, with less than 1% of wallets capturing about half of all gains. The ACDC report aims to shed light on the source of this edge.
It examines the June 2025 U.S. strikes on Iran as a case study, where Polymarket listed date-specific contracts on whether a strike would occur. The strike happened at 18:40 ET on June 21.
In the hours leading up to it, 19 longshot bets totaling $164,292 were placed across contracts that ultimately resolved to YES. Eight wallets shared about $1.8 million in profits, with one wallet taking nearly $500,000. The operation was designed by the Pentagon to be unreadable from the outside, utilizing decoy bombers and long-range stealth aircraft to avoid detection. Despite these measures, a small number of traders placed large, well-timed bets on the outcome.
This pattern is not limited to a single event. Across Polymarket's military and defense category, the report found that in five of the six two-hour windows before market resolution, winning longshot bets outnumbered losing ones, contrary to what market prices would suggest. While longshot bets can outperform due to other reasons such as mispricing or shifts in public expectations, the consistency of these patterns, especially in markets tied to military decisions, suggests that some participants may be operating with information advantages not available to others.
The ACDC, being a nonprofit research group funded through the Fund for Constitutional Government, does not have a surveillance product to sell, unlike Solidus Labs, whose recent Polymarket analysis also serves as a marketing case for the platform it licenses to Kalshi. The ACDC's recommendations include implementing identity verification for bettors, conditional payouts on suspicious wagers, restrictions on markets whose outcomes are decided by small groups, and limits on how granular contracts can become.
The report concludes by calling for 'an evidence-informed debate about whether the public should be betting on these outcomes at all.'