Algorithms keep time by the clock, and the clock marks :00 and :30
Posted: Sat Oct 31, 2026 9:49 am
Look at tick activity in one-minute buckets across a day and small spikes appear at the top of the hour and at the half hour. Orders cluster there because plenty of automated systems are scheduled to run at round times: rebalancing routines, time-weighted execution algorithms, scheduled reports, bar-close strategies.
The effect is modest on its own, but it adds up. Spreads can tick wider for a second or two. Slippage on market orders at exactly 00 seconds is often a bit worse than at 07 seconds, as covered in discussions of bar-open entries.
Bar-close strategies are a notable culprit. Thousands of retail and institutional systems trigger on the close of a five-minute or fifteen-minute bar, and they all send orders at the same instant.
For a human scalper, avoiding the first few seconds of each round-number time slot costs almost nothing and removes some of the worst fills.
It also changes how breakouts at those times should be read. A spike that appears exactly on the minute or the quarter hour is more likely to be mechanical, and less likely to carry information.
Plotting average tick count or spread against the second within each minute reveals the effect clearly on most pairs.
Does the data show a spike in activity at round clock times?
The effect is modest on its own, but it adds up. Spreads can tick wider for a second or two. Slippage on market orders at exactly 00 seconds is often a bit worse than at 07 seconds, as covered in discussions of bar-open entries.
Bar-close strategies are a notable culprit. Thousands of retail and institutional systems trigger on the close of a five-minute or fifteen-minute bar, and they all send orders at the same instant.
For a human scalper, avoiding the first few seconds of each round-number time slot costs almost nothing and removes some of the worst fills.
It also changes how breakouts at those times should be read. A spike that appears exactly on the minute or the quarter hour is more likely to be mechanical, and less likely to carry information.
Plotting average tick count or spread against the second within each minute reveals the effect clearly on most pairs.
Does the data show a spike in activity at round clock times?