Intraday Trading Volume Patterns of Market Index ETFs

Abstract

Using minute-by-minute trading volume data from five ETFs (SPY, DIA, IWM, DDM, and DOG) over the period from 2013 to 2021, this paper examines the intraday trading puzzle and uncovers patterns beyond the typical U-shape. Even after controlling for potential spurious effects—such as market open/close and Wednesdays—we find persistent, albeit with reduced magnitude, statistically significant trading volume spikes at 5-minute intervals across all ETFs. Hence, while open and close, as well as Wednesdays, contribute to the pattern, they do not fully explain it. The spikes are more pronounced during negative return periods, and our models suggest increased randomness in the distribution of trading activity over time. While these predictable spikes may stem from algorithmic trading biases, we propose they could be a legacy of older algorithms influencing newer ones as they feed from historical data. The results suggest that a puzzling trading pattern persists even for highly traded ETFs; hence, it provides important findings for investors exposed to these securities.