High Stakes: Traders Rake in $8.2 Million from Lightning Bitcoin Bets on Polymarket!

A recent study has unveiled critical vulnerabilities within Polymarket’s five-minute Bitcoin contract, asserting that it functioned as a mechanism for wealth transfer, redirecting funds from retail bettors to a select group of manipulators. This process not only facilitated illicit gains but also negatively impacted Bitcoin’s spot price. The comprehensive paper, titled “Settlement Manipulation in Prediction Markets,” was authored by David Dai, Ruizhe Jia, and Shihao Yu from Stanford University and Singapore Management University, meticulously examining a product that emerged on February 12, 2026.

On its launch date, Polymarket introduced a binary contract designed to pay $1 if Bitcoin’s price closed a five-minute window above its opening price, and $0 otherwise. A new contract would commence every five minutes, operating around the clock. Within just a few months, these five- and fifteen-minute crypto up/down markets on Polymarket saw over $4 billion in trading volume, significantly tripling the platform’s daily activity.

The fundamental flaw identified in Polymarket’s design stemmed from the contract’s settlement mechanism, which relied on a Chainlink oracle. This oracle averaged Bitcoin’s price across several major spot exchanges. Critically, a trader holding the contract possessed the ability to influence this reference price in the crucial closing seconds by executing substantial buy or sell orders of real Bitcoin, thereby dragging the oracle’s average price across the predetermined strike price and winning the bet.

While the oracle’s blend of multiple exchanges initially appeared to offer a robust defense against manipulation—suggesting that moving the price would necessitate influencing many venues simultaneously—the authors demonstrated this was not the case. Binance, being the largest cryptocurrency exchange, maintained a close correlation of approximately two and a half basis points with the oracle and often moved nearly one-for-one with it. Consequently, Binance’s price finished on the same side of the strike as the final resolution about 85% of the time, meaning a relatively small push that nudged the Binance price a few basis points past the strike was often sufficient to determine the outcome.

Empirical evidence supporting this manipulation was clearly visible in the Binance data. Following the launch of the five-minute contract, the net order flow in the final ten seconds before each contract’s close surged by about 50% above pre-launch levels. This spike was most pronounced and impactful in the approximately 6% of cycles that the market initially judged as near-even, where the jump was nearly 3.9 times greater than in other cycles. A telling sign of manipulation, rather than genuine information, was the rapid reversal of the price within ten seconds, receding by about a quarter in these near-even cycles.

The timing of these manipulative pushes also offered insights, clustering predominantly during periods of lower liquidity when a dollar of order flow could move the price the most significantly. Approximately 56% of these pushes occurred overnight, and 44% happened on weekends. The impact on contract outcomes was substantial: in near-even cycles, a push against the favored side flipped the winner 65% of the time, compared to 41% in normal trading. Even in scenarios where one side held a 90-to-100% chance of winning before the close, a manipulative push reversed the outcome in 34% of these cycles, starkly contrasting with only a 1% reversal rate in cycles without such pushes. This meant a bet that the market considered near-certain still lost one time out of three due to manipulation.

Leveraging the public nature of the blockchain, the authors meticulously traced each wallet involved. Their analysis revealed that only 821 traders fit the manipulator profile, representing a tiny fraction—about one in three hundred—of the 243,000 individuals who traded the contract. These manipulators collectively gained $8.2 million specifically from the pushed cycles, while breaking even in the remaining cycles. A staggering 93% of the losses incurred during these manipulated cycles fell upon retail investors.

The researchers also diligently ruled out hedging as an innocent explanation for these trading patterns. They noted that a binary contract offers minimal exposure to hedge once one side becomes near-certain, yet these were precisely the cycles where pushes were most effective in flipping outcomes. Furthermore, the manipulative trades arrived in a single, concentrated burst within the final fifty seconds, rather than as positions gradually built over the entire contract window, further discrediting the hedging hypothesis.

The study concluded by proposing a clear remedy: extending the contract’s horizon. They observed that manipulation was entirely absent from the fifteen-minute contracts because a longer window naturally incorporates more ordinary trading activity before the close, thereby diminishing the effectiveness of any fixed manipulative push. The implications of this research extend beyond the cryptocurrency market, reaching traditional finance. Both Nasdaq and Cboe have filed with the SEC to list binary asset-price contracts on equity indices, which, if not carefully designed, would carry the same inherent risks of manipulation onto much larger and more significant markets.