During an in-depth analysis of on-chain data and the spot market, my team identified clear signs of organized influence on the outcomes of 5-minute contracts on the Polymarket platform. This involves manipulations that led to the redistribution of millions of dollars among participants.
We examined over 60 million blockchain records, order book data, and high-frequency spot trades. As a result, a group of 821 wallets was identified, which earned $8.22 million in cycles with anomalous activity. Meanwhile, retail and unclassified traders lost $7.61 million in the same cycles.
Anatomy of the Scheme
The five-minute contracts on Bitcoin price movement were launched on February 12, 2026. The mechanics are simple: if the BTC price at the end of the window is not lower than the starting price, $1 is paid out; otherwise, $0. The Chainlink price feed is used to determine the outcome.
The research showed that manipulators operate according to the following algorithm: first, a position is opened on the contract, and then, seconds before settlement, large spot trades are executed in the desired direction. This briefly shifts the price recorded by the oracle, after which quotes partially return to the original level.
After the launch of the five-minute markets, the volume of directional spot orders in the last 10 seconds of the window increased by approximately 50%. At the same time, the price movement began to reverse within the next 10 seconds. This pattern is a classic sign of temporary impact, rather than a reaction to news, as informational shocks usually persist longer.
Cycles where Polymarket estimated the probability of one side winning at 90-100% were particularly telling. In cases of directional movement against the favorite, the opposite side won in 34.2% of presumably manipulated cycles. Without an anomalous spike, such an outcome was observed in only 1% of cases.
Scale and Consequences
We classified 1,613 cycles as likely manipulated—the top 10% in terms of directional trade intensity in the final seconds. The average volume of spot operations in such cycles reached $1.7 million, compared to $68,000 in normal ones. About 56% of episodes occurred during nighttime hours, and 44% on weekends, when market liquidity is lower and less capital is needed to shift the price.
The 821 identified wallets account for only 0.34% of the approximately 243,000 addresses that traded this contract. However, they accounted for 93% of the profits generated from retail traders' losses.
Polymarket has already announced plans to change the settlement mechanism by switching to price averaging over a longer period, which would make short-term spot manipulations less effective. In 15-minute contracts, as the analysis showed, the spike in activity was significantly weaker, and no price reversal was observed after settlement.
My professional opinion: This situation is yet another reminder that prediction markets, especially those with short time windows, are extremely vulnerable to manipulation. Increasing the time window to 15 minutes and implementing averaging mechanisms is a reasonable step, but it will not fully solve the problem. For the long-term sustainability of such platforms, more sophisticated protection mechanisms are needed, including position size limits and random selection of the price fixation moment.