A new study argues that Polymarket’s five-minute Bitcoin contract turned into a mechanism for transferring wealth from retail bettors to a small group of traders who could influence settlement, while also worsening spot-market price quality.
The paper, Settlement Manipulation in Prediction Markets, was written by David Dai, Ruizhe Jia, and Shihao Yu of Stanford and Singapore Management University. It examined a product that only began trading on Feb. 12, 2026, when Polymarket launched a binary market that paid $1 if Bitcoin closed a five-minute window above where it opened, and $0 if it did not. A new contract opened every five minutes, around the clock.
The settlement design is at the center of the study
Within months of launch, Polymarket’s five- and fifteen-minute crypto up/down markets handled more than $4 billion in volume and tripled the platform’s daily activity, according to the article.
The study says the weakness came from how the contracts were settled. Polymarket used a Chainlink oracle that averaged Bitcoin prices across major spot exchanges. Under that setup, a trader holding the contract could buy or sell actual Bitcoin in the last seconds before expiry, drag the reference price across the strike, and change the outcome of the bet.
Using a blended price from several exchanges looked like a safeguard on paper, because it seemed to require moving multiple venues at once. The authors said that protection was limited. Binance, the largest crypto exchange, sat roughly 2.5 basis points from the oracle and moved almost one-for-one with it. It ended on the same side of the strike as the final resolution about 85% of the time. In practice, pushing Binance a few basis points past the strike was often enough to carry the result.
Final-seconds flow jumped after launch
The paper said the pattern was visible in Binance trading data. After the five-minute contract went live, net order flow in the last 10 seconds before each close rose about 50% above pre-launch levels.
The effect was strongest in the moments when a small push mattered most. In the 6% of cycles where the market viewed the odds as close to even, the jump in order flow was about 3.9 times the level seen in the rest of the sample.
Prices then snapped back. That reversal is a key part of the study’s argument: information-driven price moves usually remain in the market, while manipulative pushes tend to fade quickly. Within 10 seconds, prices reverted, and in near-even cycles the pullback was about one-quarter of the earlier move.
The pushes also clustered during thinner trading hours, when each dollar of flow had more effect on price. The study found that 56% occurred overnight and 44% happened on weekends.
821 traders matched the manipulator profile
The authors also examined who gained and who absorbed the losses.
In near-even cycles, a push against the favored side flipped the outcome 65% of the time, compared with 41% in normal trading. Even when one side had a 90% to 100% chance before the close, a push against it reversed the result 34% of the time. In cycles without a push, that figure was just 1%. Bets the market treated as near-certain still lost roughly one time in three when a last-second move appeared.
Because Polymarket settles on a public blockchain, the researchers traced individual wallets. They identified 821 traders who fit the manipulator profile, about one in 300 among the 243,000 people who traded the contract. In the pushed cycles, that group took $8.2 million. In the rest, they were roughly break-even.
The study said 93% of the losses fell on retail traders.
The authors also rejected hedging as an innocent explanation. A binary contract leaves little exposure to hedge once one side is already close to certain, yet those were the same cycles where a push could still reverse the result. The trades also arrived in a single burst during the final 50 seconds rather than being built gradually over the full window.
The paper points to longer horizons as the fix
According to the study, the issue did not appear in the 15-minute contract. A longer window captured more ordinary trading before settlement, making a fixed-size push less effective.
The article says the implications may extend beyond crypto. Nasdaq and Cboe have each filed with the U.S. Securities and Exchange Commission to list binary asset-price contracts tied to equity indexes, which the authors said could carry the same risk into larger markets.
This report first appeared in Bitcoin Magazine and was written by Micah Zimmerman.