Researchers Find Evidence of Manipulation in Bitcoin Contracts on Polymarket

Researchers from Stanford University and Singapore Management University have identified signs of manipulation in the settlement of five-minute Bitcoin contracts on Polymarket. This is detailed in their paper, Settlement Manipulation in Prediction Markets.

The authors analyzed over 60 million on-chain records, snapshots of the platform's order book, and high-frequency spot market data. They identified 821 wallets that earned $8.22 million during periods of unusual activity, while retail and other unclassified participants lost $7.61 million.

How the Alleged Scheme Works

The five-minute Polymarket contract was launched on February 12, 2026. It pays out $1 if the Bitcoin price at the end of the window is at least equal to the starting value, and $0 otherwise.

To determine the outcome, the platform uses the BTC/USD price feed from Chainlink. New contracts are initiated every five minutes, around the clock.

Bitcoin price movement within a completed five-minute contract. Source: Polymarket.

The researchers described the following alleged scheme: a trader first buys a contract betting on the rise or fall of Bitcoin, and in the last seconds before settlement, executes large spot trades in the desired direction.

This short-term market movement affects the price recorded by the oracle. The contract settles in favor of the trader, after which prices partially revert to their previous levels.

After the launch of the five-minute markets, the directed flow of orders in the last ten seconds of the window increased by about 50% compared to the period before their introduction. Price movements began to reverse within the next ten seconds.

Theoretical sequence of manipulating the settlement price of the contract. Source: Settlement Manipulation in Prediction Markets.

The authors considered this pattern a sign of temporary price manipulation rather than trading based on new information, as price movements triggered by significant news typically last longer.

In cycles where the outcome remained close to even odds, the spike in activity was approximately 3.9 times stronger than normal. The average volume of directed spot trades in the last ten seconds of the allegedly manipulated cycles reached $1.7 million, compared to $68,000 in others.

About 56% of these episodes occurred during nighttime hours in UTC, while 44% occurred on weekends. The researchers explained this concentration by lower liquidity during these times, meaning less capital was needed to shift prices.

821 Wallets Earned $8.22 Million

The researchers classified 1,613 cycles as allegedly manipulated—representing the top 10% of observations based on the intensity of directed trades in the last seconds. They identified wallets that participated in at least five such cycles and made a cumulative profit of at least $2,000 as likely manipulators.

A total of 821 wallets met these criteria, accounting for 0.34% of approximately 243,000 addresses that traded the contract. They earned $8.22 million in anomalous cycles and only $90,000 in others.

The retail/other category lost $7.61 million in the first instance, which corresponds to about 93% of the profits of the alleged manipulators. In the overall distribution of losses, retail and other unclassified participants accounted for 70.5%, likely manipulators in unsuccessful cycles for 20.2%, and market makers for 9.3%.

The authors cautioned that a wallet address does not necessarily correspond to an individual; one participant could use multiple accounts. Furthermore, the study does not prove the intent of the address owners.

The classification is based on participation in anomalous cycles and the profits earned. The researchers also did not establish a direct link between accounts executing spot trades and wallets receiving payouts on Polymarket.

Even Nearly Resolved Outcomes Changed

The authors highlighted cycles where Polymarket had already assessed the probability of one side winning at 90–100%. In directed movements against the favorite, the opposing side won in 34.2% of the allegedly manipulated cycles. Without a recorded spike, such a result occurred in only 1% of cases.

In more balanced cycles, where the favorite's probability was 50–60%, directed movement against them led to the opposing side winning in 65.4% of cases, compared to 41.4% without anomalous activity.

The researchers also considered an alternative explanation—hedging by market makers. However, their average maximum exposure was around $1,400 per cycle, while the directed spot flow reached approximately $1.7 million.

Additionally, positions on Polymarket were built gradually over five minutes, whereas large spot trades were concentrated just before settlement. The authors deemed this difference in scale and timing incompatible with typical hedging.

Polymarket to Change Settlement Mechanism

A Polymarket representative told Bloomberg that the platform uses several independent price oracles to aggregate data. The company also plans to transition some markets to a mechanism that considers prices over a longer period rather than a single point in time.

In fifteen-minute contracts, the authors found a significantly weaker spike in activity and did not observe a pronounced price reversal after settlement. They believe that increasing the duration of the window reduces the likelihood that a brief spot trade can change the outcome.

Among other protective measures, the researchers suggested calculating based on the average price over a period, randomizing the moment of fixation, limiting position sizes, and imposing additional costs for trades made just before closing.

In April, researchers questioned the "wisdom of the crowd" on Polymarket. They estimated that 3.14% of the platform's users, along with market makers, earn over 30% of the profits.