A new study of Binance futures data finds that Bitcoin perpetual futures trading surges every 15 minutes, with the first ten seconds of each quarter-hour carrying far more trades and volume than an ordinary moment. The pattern holds across six cryptocurrencies, but the researchers found the predictable price move is too small to cover trading fees.
Bitcoin perpetual futures trade calmly most of the time, then burst into activity the instant the clock strikes a quarter-hour. Researchers documented 26% more trades and 32% more dollar volume during the first ten seconds of these boundaries than during the same window at an ordinary minute, with absolute price moves running 26% larger.
Korean policy researcher Chan Kim and Peter Reinhard Hansen of the University of North Carolina found the pattern in an August 2026 study covering six Binance futures markets: Bitcoin, Ethereum, XRP, Solana, Dogecoin, and Cardano. They examined trade records from Jan. 1, 2021, through Oct. 31, 2024, spanning 1,400 days of continuous trading.
Bitcoin leads the pattern by size
Bitcoin averaged 1.54 million daily trades and $14.58 billion in contract volume during the sample, while Cardano averaged roughly 290,000 trades and $544 million over the same rhythm. The researchers say that consistency across such different market sizes points to a shared cause across trading systems rather than a quirk of one token.
The researchers link the bursts to chart intervals that many trading systems follow. When a 15-minute candle closes, technical indicators recalculate and automated strategies receive fresh signals at the same moment, pulling many systems toward the same clock boundary.
A tell in trade size
To spot machine activity, the authors counted how often trade sizes ended in round numbers, since algorithms using leverage or a volatility-based formula tend to produce less round quantities than human traders. For Bitcoin trades eligible to end in at least two zeros, the round-size share fell by 0.20 standard deviations at the top of the hour, five times the drop seen at an ordinary minute. A separate check on Bybit data showed a similar structure on another exchange.
Predictable, but not profitable
A rolling model using data available before each quarter-hour picked the correct price direction 56.6% of the time, with an out-of-sample R-squared of 3.4%. Yet the average gross return came to just 0.51 basis points per trade before fees, roughly 51 cents on a $10,000 trade. Binance's base taker fee during the sample was 5 basis points, meaning the predicted move covered only a tenth of that single charge.
The pattern still carries a longer signal: when buyer-initiated volume exceeded seller-initiated volume at a boundary, that imbalance was associated with returns over the next four to 12 hours. Market makers, the authors note, could still use the finding to demand a wider spread or reduce how much they offer during those ten seconds.
Source: CryptoSlate
Trading involves risk.
Risk Disclaimer
