Round numbers hold more orders, not the price

Prices cluster at round numbers, a pattern documented for decades and often read as evidence that round levels act as barriers. A paper posted this week points out that the evidence usually offered for a barrier is the same evidence used to show clustering, and that it has a mechanical explanation: more limit orders rest at round prices, deeper queues take longer to deplete, and so prices linger there without roundness itself doing anything.

The authors separate the two with a conditional question: given the same visible liquidity, does the price stay longer at a round level than at a non-round one? They call an affirmative answer residual stickiness. To measure it they work on the tick grid, counting the time spent at each level and the number of departures from it, and estimate the ratio of durations at matched entry depth with a proportional-hazards regression.

The data are 250 trading days in 2025 for five Nasdaq stocks — Apple, Cisco, Intel, SoFi Technologies and American Airlines — with median prices from 12.69 to 215.10 dollars and relative tick sizes from 0.5 to 7.9 basis points. Round levels attracted more resting liquidity in all five. Neither raw nor depth-adjusted durations showed greater persistence there.

Two stocks behaved differently. In Intel and SoFi, round levels showed a directional effect: prices departed sooner, and departures were systematically biased away from the round number. Both have comparatively large ticks relative to price, yet within each stock the effect did not strengthen as prices fell, which the authors read as a sign that relative tick size alone does not explain it. For anyone placing orders around round levels, the practical reading is that the queue is deeper there, not that the price is held.