The Polymarket prediction market experienced an unusual surge in activity, which, as it turned out, had a dark side. The launch of ultra-short five-minute contracts on Bitcoin's price movement not only attracted retail traders but also created a perfect mechanism for redistributing funds in favor of a narrow group of manipulators. A new study from Stanford University and the Singapore Management University sheds light on this scheme, which distorted not only the prediction market but also the spot price of Bitcoin.

Starting February 12, 2026, when Polymarket launched its binary contract, trading activity on the platform skyrocketed. In just a few months, the total trading volume for these contracts exceeded $4 billion, and the platform's daily turnover tripled. The mechanics were simple: the contract paid $1 if Bitcoin closed a five-minute window above its opening price, and $0 otherwise. A new contract was launched every five minutes, around the clock.

The Oracle's Achilles' Heel

The key vulnerability, according to researchers, lay in the settlement mechanism. The contract relied on the Chainlink oracle, which averaged the Bitcoin price across major spot exchanges. A trader holding the contract could, in the final seconds of the window, buy or sell actual Bitcoin on Binance, shifting the reference price in the desired direction and guaranteeing a win. The protection of averaging across multiple exchanges proved illusory: Binance, the largest among them, deviated from the oracle by only 2.5 basis points and moved almost in sync with it.

Manipulators applied pressure specifically through Binance. In 85% of cases where the contract was close to equilibrium, the price on Binance ended up on the same side of the threshold as the final calculation result. A price nudge of a few points—and the outcome was decided. A characteristic price pullback after the calculation confirmed the artificial nature of the movement: within ten seconds, the price recovered about a quarter of its shift. The peak of manipulation activity occurred during quiet hours—56% at night and 44% on weekends.

Numbers and Profiles of Manipulators

The results were striking. In cycles close to equilibrium, a nudge against the favorite changed the winner in 65% of cases, compared to 41% during normal trading. Even when one side had a 90–100% chance of winning, an attack reversed the outcome in 34% of cases. Researchers identified only 821 wallets fitting the manipulator profile—roughly one in three hundred among 243,000 participants. In cycles with nudges, they earned $8.2 million, while others broke even. Meanwhile, 93% of all losses were borne by retail traders.

The authors dismissed the possibility of harmless hedging: when hedging, a binary contract carries almost no risk when one side is close to winning, yet manipulators specifically flipped such cycles. Moreover, trades were executed in a single burst in the last 50 seconds, rather than accumulating gradually.

A Lesson for Markets and a Solution

Notably, no manipulation was observed on the fifteen-minute contract: a longer window allowed more regular trades to pass through, diluting the single nudge. According to the authors, the solution lies precisely in the contract duration. However, the implications extend far beyond cryptocurrencies. Exchanges Nasdaq and Cboe have already filed applications with the SEC for binary contracts on stock indices, risking transferring the same problem to significantly larger markets.

Expert opinion: This story is a vivid example of how innovations in DeFi, despite their transparency, can be turned against unsophisticated users. Polymarket, being a public blockchain, made all transactions visible, but that did not stop manipulators. The market urgently needs smarter oracles and mechanisms resistant to the "last mile"—otherwise, we will see this scheme repeated on traditional exchanges, where the stakes will no longer be $4 billion, but many times greater.