A new study claims that Polymarket’s five-minute Bitcoin contracts have become a wealth transfer machine. It shifted funds from individual bettors to a small number of manipulators, deteriorating Bitcoin’s spot price in the process.
The paper “Payment Operations in Prediction Markets” by David Dai, Ruizhe Jia and Shihao Yu from Stanford University and Singapore Management University studied products that did not exist before February 12, 2026.
On that day, Polymarket launched a binary contract that pays $1 if Bitcoin closes a 5-minute window from its inception, and $0 otherwise. New contracts started every 5 minutes for 24 hours.
Within a few months, over $4 billion was traded on Polymarket’s 5-minute and 15-minute crypto up/down markets, and the platform’s daily trading volume tripled. The Polymarket flaw occurred when a contract was settled against the Chainlink oracle, which averages Bitcoin prices on major spot exchanges.
Traders holding the contract can buy or sell real Bitcoin within seconds of expiration and drag its base price into the exercise period to win their bets.
Oracle’s blend of exchanges looked like a defense. Because it seemed like moving it would require moving many venues at once. The authors showed that it is not very protective. Binance, the largest cryptocurrency exchange, was about 2.5 basis points behind Oracle, and was tied almost 1-to-1 with Oracle. About 85% of the time, it ended on the same side as the strike solution. Pushing Binance’s price several basis points above the strike has resulted.
The pattern was in Binance’s data. After the 5-minute contract was initiated, net order flow in the last 10 seconds before the close of each trade was approximately 50% above the pre-opening level. The spikes were most rapid when the push was significant. In the 6% of cycles where we found the market to be roughly even, the jump was about 3.9x the rest.
We conceded a goal due to a reversal. Authentic information comes at a price. Manipulative pushing is not. Within 10 seconds, the price reversed by about a quarter in a nearly even cycle. The gains were concentrated in the thinner hours when dollar flows move the most, with 56% landing overnight and 44% over the weekend.
Who won and who paid in polymarket bets?
In a near-even cycle, a push against the favored side would flip a winner 65% of the time, compared to 41% in a regular trade. Even if one team had a 90-100% chance before the close, there was a 34% chance of reversing the outcome if there was a push, and a 1% chance of reversing the outcome in cycles without a push. Bets that were considered near-certain by the market ended up losing one in three times.
Since Polymarket settled on a public blockchain, the authors tracked each wallet. Only 821 traders fit the manipulator profile, or about 1 in 300 of the 243,000 people who traded the contract. They earned $8.2 million in the pushed cycle and broke even for the remainder of the cycle. 93% of the losses occurred in retail.
The authors ruled out hedging as a benign explanation. In binary contracts, there was little risk of hedging once one trade was almost certain, but these cycles were reversed on a push. And the trade arrived all at once in the last 50 seconds, rather than as a position built over the window.
treatment method
Amendments were due in the contract. No operations were performed on the 15-minute contract. This is because the longer window causes more regular trades to occur before the close, weakening the force of the fixed push. The stakes extended beyond cryptocurrencies. Nasdaq and Seabo have each filed with the SEC to register binary asset price contracts in their stock indexes, which pose the same risks to the larger market.
This article, Study found trader won $8.2 million in 5-minute Bitcoin bets on Polymarket, was first published in Bitcoin Magazine and written by Micah Zimmerman.
