Price clustering on cryptocurrency order books at a US-based exchange

We investigate the price clustering effect in cryptocurrency limit order books, where traders tend to place orders in round numbers. Analyzing 10-minute snapshots of five USD-denominated cryptocurrencies over 35 weeks (January–August 2020), we find that the frequency of specific cent components (e.g., 00, 50) increases significantly at higher price levels. Subsample analyses reveal that this effect strengthens with larger previous cumulative dollar depth and larger price distance from the best price. Furthermore, near-best-price and far-from-best-price dollar quotes exhibit a weak negative and strong positive price impact, respectively, confirming the informativeness of the clustering effect. These findings remain robust when considering superstitious two digits, cluster undercutting, last digits, excluding stale quotes, and additional price levels.