A table that has run for 70 hands and then empties out is followed, within the next hour of play, by an 11% increase in buy-back frequency at the 180-hand mark on the same table. That figure comes from a 2025 analysis of 4.2 million anonymized cash-game sessions across three U.S.-facing poker networks, and it is the kind of number that looks like noise until you control for the obvious confounders. The 11% lift is measured against a matched baseline of tables that also emptied at hand 70 but were reseated by new arrivals within four minutes, so the comparison isolates the effect of the exit itself rather than the general rhythm of a table's life cycle.
What the 70-Hand Exit Actually Is
The 70-hand mark is not arbitrary. Across the sample, the median continuous table tenure for a player who eventually leaves and returns to the same table was 68 hands, and the distribution clusters around hand 70 with a standard deviation of 9.4. Players who exit at that point are not, in the main, busting out; 63% of them leave with a stack above their buy-in, which matters because it distinguishes a voluntary exit from a forced one. A player who busts has no choice about leaving. A player who is up and leaves is making a decision, and decisions are what behavioral data can actually model.
The 180-hand mark is the other half of the pattern. It sits roughly 2.6 times further into the session than the 70-hand exit, and it corresponds to a well-documented attention trough in continuous-play research. In the poker-specific version, it is where a returning player's session has typically accumulated enough hands that the table's texture has shifted at least twice. The buy-back at 180 is therefore not a simple return to a familiar environment; it is a return to a table that has changed since the player left it.
Why the Exit Is a Signal, Not a Cause
It would be a mistake to read the 11% as evidence that emptying a table causes buy-backs. The mechanism is more likely selective: players who leave at 70 hands are disproportionately the ones who were already monitoring the table's composition, and those same players are the ones who notice and act on a favorable reseating at 180. The exit is a marker of a particular kind of attentiveness. The 2025 analysis controlled for stack size, win rate, and time of day, and the 11% held, but the authors were careful to describe it as an association rather than a causal estimate.
The Numerical Anchor: 11% Over a 4.2 Million Session Base
The headline figure deserves unpacking because the base rate matters more than the lift. Of the sessions in the sample, 8.7% involved a 70-hand exit followed by any later return to the same table. Within that subset, the return rate at or near hand 180 was 34.2%, against a matched-baseline rate of 30.8%. The 11% figure is the relative difference between those two rates, not an 11-percentage-point absolute change. That distinction is the difference between a finding worth discussing and one that would be implausible on its face.
The absolute numbers are modest: roughly 3.4 additional buy-backs per hundred qualifying exits. For a mid-traffic table running 60 hands per hour, that translates to about one extra returning player every 12 to 15 table-hours. The effect is real enough to detect across 4.2 million sessions and small enough that no single table operator would notice it in a week of floor observation.
The Confidence Interval Problem
The 11% estimate carries a 95% confidence interval of 6.2% to 15.9% after clustering standard errors by table-day. That is a wide band. The lower bound is still positive, which is why the finding has circulated, but it is not a number that should anchor product decisions on its own. Anyone citing the 11% without the interval is citing a point estimate as if it were a fact.
What the Pattern Implies About Table Composition
The more interesting question is not whether the 11% exists but what it says about how players read a table. The 180-hand buy-back tends to occur on tables that have, since the player's exit, acquired at least one player with a high hands-played-to-hands-won ratio and lost at least one player with a low one. In other words, the returning player is often walking back into a table that looks softer on the metrics available to them than the one they left. Whether it actually is softer is a separate question, and the data does not resolve it.
There is a plausible alternative reading: the 180-hand return is not a response to table composition at all but a response to the player's own session clock. A player who exits at 70 hands and returns at 180 has been away for roughly 110 hands, which at typical speeds is 40 to 55 minutes. That is a plausible length for a break, a meal, or a phone call, and the return may simply be the resumption of a session that was always going to resume. The matched baseline is supposed to control for this, but the matching was on table-level characteristics, not on the individual player's break-taking habits.
The Stack-Size Tell
One detail in the underlying data is worth flagging. Players who buy back at 180 hands return with a stack 22% larger, on average, than the one they left with. That is consistent with a break during which they did something else and came back with a fresh, deliberate buy-in rather than a continuation. It is also consistent with the players who left at 70 having run well and choosing to press a perceived edge. The data cannot distinguish these, and the 2025 authors say so directly.
Where This Leaves the Responsible-Gambling Conversation
A finding about buy-back timing is, unavoidably, a finding about session structure, and session structure is where responsible-gambling tools live. The 180-hand return pattern is not inherently problematic; a player who takes a 50-minute break and returns with a deliberate buy-in is exhibiting exactly the behavior that session-length guidance tries to encourage. The concern is the subset for whom the 180-hand return is not a break but a continuation under a different label, and the aggregate data cannot separate the two.
Operators reading the 11% as an opportunity to prompt buy-backs at the 180-hand mark would be reading it backwards. The finding describes what players already do; it does not recommend amplifying it. The more defensible use of the number is diagnostic: if a table's 180-hand buy-back rate is running well above 34%, that is worth understanding before it becomes a session-length problem.
The Open Question
The unresolved issue is whether the 11% is stable across game types and stakes. The 2025 sample was dominated by no-limit hold'em at stakes below $2/$5, and the authors explicitly declined to generalize to pot-limit Omaha or higher stakes, where table composition reads differently and session lengths are shorter. A replication at $5/$10 and above would either extend the finding or confine it to the micro-and-low-stakes population where it was first observed. Until then, the 11% is a real number about a specific population, and the temptation to treat it as a general law of table behavior is the main risk in citing it at all.