A Stanford and Singapore Management University study found that Polymarket’s five-minute Bitcoin contract functioned as a wealth transfer mechanism, with a small group of traders systematically manipulating the settlement oracle to win bets that would otherwise have lost.
The contract, launched February 12, 2026, paid $1 if Bitcoin closed above its opening price within the five-minute window, $0 otherwise. A new contract opened every five minutes around the clock. The contract settled against a Chainlink oracle that averaged Bitcoin’s price across major spot exchanges including Binance.
Researchers David Dai of Stanford, Ruizhe Jia of Singapore Management University, and Shihao Yu of Singapore Management University found that 821 traders out of 243,000 total participants fit the manipulator profile. These traders executed large Bitcoin purchases or sales in the final seconds to move the Chainlink reference price across the strike price, reversing outcomes in their favor.
The manipulators extracted $8.2 million in pushed cycles, while breaking even in non-pushed cycles. Retail traders absorbed 93% of the losses. The ratio of manipulators to total traders was approximately one in three hundred.
Mechanics of the Manipulation
The vulnerability stemmed from the oracle’s reliance on spot exchange prices in the closing seconds. Binance typically finished within 2.5 basis points of the oracle and landed on the same side of the strike as final settlement 85% of the time, making it the primary reference point for manipulation.
Net order flow spiked 50% in the final ten seconds after launch. In near-even cycles where the outcome was uncertain, order flow jumped 3.9 times higher than in the rest of the window. Within ten seconds of manipulative pushes, price reverted by about a quarter in near-even cycles.
Pushes occurred with notable timing patterns: 56% happened overnight and 44% on weekends, periods when markets were thinner and price movement required less capital. In near-even cycles, pushes flipped the winner 65% of the time, compared to a 41% flip rate in regular trading without manipulation.
When Bitcoin’s probability of finishing above the strike reached 90-100%, a push reversed the outcome 34% of the time. In cycles with no push, reversals occurred just 1% of the time.
Market Scale and Structural Implications
Five- and fifteen-minute crypto markets on Polymarket traded more than $4 billion within months of the five-minute contract’s launch. The fifteen-minute contract showed no manipulation because longer windows diluted the effect of fixed-size pushes.
The study titled “Settlement Manipulation in Prediction Markets” reveals a structural vulnerability in oracle-dependent contracts where real-world price movement can be engineered by concentrated capital in the final settlement window.
Regulatory Developments
Nasdaq and Cboe each filed with the SEC to list binary asset-price contracts on equity indices. The filings suggest similar five-minute or fifteen-minute binary contracts could be deployed on traditional assets, potentially exposing equity markets to the same manipulation vector unless oracle design is hardened against final-window pushes.
Frequently Asked Questions
How much did the manipulators extract, and who lost?
The 821 traders extracted $8.2 million in pushed cycles while breaking even in non-pushed cycles, and retail traders absorbed 93% of the losses.
How did the manipulation work?
The contract settled against a Chainlink oracle averaging Bitcoin's price across spot exchanges including Binance; traders made large trades in the final seconds to move the reference price across the strike, reversing outcomes.
What is the broader risk?
Five- and fifteen-minute crypto markets on Polymarket traded more than $4 billion within months, and Nasdaq and Cboe each filed with the SEC to list similar binary contracts on equity indices.
