Study Flags $8.2M Settlement Manipulation in Polymarket Bitcoin Markets

A new study of Polymarket’s five-minute Bitcoin contracts identified concentrated spot orders during the final seconds before settlement, followed by rapid price reversals once the contracts expired.

David Dai and Ruizhe Jia of Stanford University and Shihao Yu of Singapore Management University analyzed more than 60 million Polymarket fills alongside tick-level Bitcoin trades and order-book data from Binance. Their sample ran from July 1, 2025, through April 8, 2026, covering the periods before and after the five-minute contracts began trading on February 12.

Final-ten-second order flow on Binance increased by about 50% after the product launched. The increase was strongest when the contract remained close to an even outcome, where the late order-flow jump was 3.9 times larger than in other settlement windows. Much of the resulting Bitcoin price movement reversed within ten seconds.

The researchers classified 1,613 settlement windows as likely manipulated using unusually large final-ten-second order flow relative to activity earlier in the same cycle. The classification identifies a statistical trading pattern, not legally proven intent.

Binance Orders Could Shift the Settlement Benchmark

Polymarket’s five-minute Bitcoin markets pay $1 to the winning side based on whether Bitcoin finishes the interval above or below its opening price.

Settlement uses Chainlink Data Streams rather than the price from one exchange. Chainlink combines market data from multiple venues through independent data providers and node operators. The researchers found that Binance’s Bitcoin price remained close enough to the aggregated benchmark for a concentrated order placed on Binance to move the wider market through arbitrage activity.

Binance’s midpoint sat about 2.5 basis points from the Chainlink reference price and finished on the same side of the strike as the final settlement in roughly 85% of observed cycles. The study did not identify a compromise of Polymarket or Chainlink infrastructure. It instead describes a market-design weakness where traders holding a binary position may also trade the underlying asset used to determine its payout.

Researchers Attribute Most Losses to Retail

The study identified 821 wallets that participated in at least five likely manipulated cycles and earned at least $2,000 across those periods. The group represented 0.34% of approximately 243,000 traders and captured $8.22 million in aggregate profit.

Accounts classified as retail or other traders lost $7.61 million in the affected cycles and were net negative in 65.3% of them. Market makers earned approximately $620,000, with the analysis indicating that many withdrew liquidity as settlement approached rather than absorbing the other side of the trades. The 821-wallet cohort earned only about $90,000 across the remaining normal cycles.

The manipulation signature was substantially weaker in Polymarket’s 15-minute contracts. The researchers proposed longer settlement windows, averaged or randomized closing prices and position limits as possible safeguards.

The findings add a price-settlement issue to wider scrutiny of prediction-market integrity. A House Oversight investigation has already requested details on Polymarket’s identity checks, suspicious-trading controls and handling of users with access to nonpublic information.

The post Study Flags $8.2M Settlement Manipulation in Polymarket Bitcoin Markets appeared first on Crypto Adventure.