Polymarket Changes Crypto Market Rules After Manipulation Study

Polymarket has changed how its short-dated crypto prediction markets settle after months of complaints and research pointing to possible manipulation around final price snapshots. According to CoinDesk, the platform is replacing single-price settlement points with time-weighted average prices, or TWAPs, after researchers identified 821 accounts that made $8.2 million in settlement windows they classified as likely manipulated.

The issue matters because short-duration prediction markets can be vulnerable when a contract resolves against a price that can be briefly moved in the underlying market. In the case studied, researchers from Stanford University and Singapore Management University examined roughly two months of five-minute bitcoin contracts and found unusually large Binance orders in the final seconds before settlement, followed by quick bitcoin price reversals.

The study did not prove trader intent or show that the same people placing spot-market orders also held Polymarket positions. Still, it found that, excluding market makers, 93% of losses in windows classified as manipulated were borne by retail traders, highlighting how market design can shift risk toward less sophisticated participants.

Under Polymarket’s new approach, five-minute markets will use a 30-second average, while 15-minute and four-hour markets will use a 60-second average. The platform said the pricing data will be delivered through Chainlink Data Streams and said it would add $1 million in liquidity rewards across affected markets during August to support the transition.

CoinDesk also noted that similar concerns have been discussed by onchain analysts before the study, including claims that traders could build a position on Polymarket and then move the Binance price during the settlement window. Rival Kalshi told CoinDesk it uses a regulated CF Benchmarks price index and a 60-second moving average, arguing that this makes short-lived price pushes harder and more expensive.