A two-beat time-bank alert — roughly 1.6 seconds of audible warning before the clock expires — produces measurably better fold decisions in session-90 play than the industry-standard five-beat alert, according to hand-history analysis of 41,300 tournament folds across three buy-in tiers. The effect is not large, about 3.4 percentage points fewer timeouts per hundred folds, but it is consistent across sites and stakes, and it runs against the intuition that more warning time is always better.
This matters more than it sounds. Time banks were introduced to protect players who need to think, and they do. But the alert that precedes expiry has quietly become a behavioral variable of its own, and almost no operator treats it as one. Most platforms inherited a five-beat countdown from early tournament software and never revisited it. The data suggests that choice is costing players — and, in a roundabout way, the operators who rely on those players finishing events rather than timing out of them.
What the Alert Actually Does
A time-bank alert is a cue, not a control. It fires when a player's remaining reserve drops below a threshold, usually expressed in seconds or in software "beats." The five-beat convention is common because it maps cleanly onto a five-second visual countdown and because it gives a distracted player a fair chance to return attention to the table.
The problem is that five beats is long enough to invite deliberation. A player who hears the first alert and still has four beats left will often use them — re-examining a marginal call, re-running a range, second-guessing a read. That deliberation is not free. In the session-90 sample, 61.8% of folds made after a five-beat alert consumed at least three additional beats, and those folds were 4.1 percentage points more likely to be followed by a timeout later in the same session. The alert appears to train a habit of using the reserve rather than preserving it.
Why two beats behaves differently
A two-beat alert functions closer to an interrupt than a warning. It arrives late enough to be actionable and early enough that the player has no room to reopen the decision. In the sample, players receiving a two-beat alert folded 3.4 percentage points less often at expiry and, more tellingly, made 8.7% fewer "reversal folds" — folds where the player had already moved chips toward the pot and pulled back. Reversal folds are the costliest kind of hesitation, both in time and in table image.
The Session-90 Problem
Session-90 refers to the ninety-minute mark of a tournament session, a point at which cumulative fatigue, rising blinds, and shrinking stacks converge. It is not a precise physiological threshold; it is a convenient marker that corresponds to the third or fourth level after late registration closes in most mid-stakes schedules.
The sample was drawn specifically from this window because the baseline timeout rate is high enough to measure. Across the 41,300 folds, the overall expiry rate was 6.2%. In session-90 hands specifically, it rose to 9.4% under five-beat alerts and 6.0% under two-beat alerts. That gap — 3.4 points — is the headline figure, and it held within 0.6 points when the sample was split by buy-in tier ($22, $109, $530) and by site.
The confound worth naming
Players who reach session 90 are not a random sample. They are survivors, and survivors tend to be more patient, more bankrolled, and more practiced at clock management. Any claim about alert timing has to survive the objection that the two-beat group simply contained better players.
The analysis attempted to control for this by matching on hands-per-minute, prior timeout rate, and stack depth at the ninety-minute mark. Matching reduced the measured effect from 4.6 to 3.4 points but did not eliminate it. That residual is the honest number, and it is small enough that a skeptic could reasonably attribute part of it to unobserved skill. The direction, however, did not flip in any subgroup.
What Operators Get Wrong
The five-beat default persists for reasons that have little to do with player outcomes. It is baked into legacy tournament clients, it is documented in older rule sets, and changing it requires a client update that touches every table. Operators also worry, reasonably, that shortening the alert will increase timeouts and generate support tickets.
The data suggests the opposite. Timeouts fell in every two-beat cohort measured. The mechanism is plausible: a shorter alert interrupts rather than invites, and an interrupted player acts. The five-beat alert, by contrast, gives a player just enough runway to talk themselves out of a decision they had already made correctly.
There is a caveat for cash games and for players using assistive software, where the alert interacts with HUD refresh cycles and can fire before a stat panel updates. Two beats may be too tight in those environments. The finding here is specific to tournament play, and specifically to the session-90 window.
A note on responsible design
None of this argues for removing time banks or compressing them to the point of stress. A two-beat alert still leaves the full reserve available; it changes only when the player is told the reserve is running low. Players who need more time should have it, and any operator tightening alerts should pair the change with a clear, opt-out-able setting rather than a silent default. Session-90 fatigue is a real phenomenon, and a design that pushes players to act faster late in a long session is not automatically a kindness. The case for two beats rests on fewer timeouts and fewer reversal folds, not on speed for its own sake.
The Number That Should Travel
The cleanest way to state the finding is this: over 41,300 folds, a two-beat alert reduced session-90 expiry from 9.4% to 6.0%, a 36% relative reduction, with the effect surviving matching at 3.4 absolute points. That is a modest, replicable result from a single observational dataset, and it should be treated as a hypothesis worth testing rather than a settled design rule.
The open question is whether the effect compounds. If a two-beat alert reduces timeouts by 3.4 points per session-90 window, does a player who experiences it across twenty tournaments develop better clock habits, or does the benefit wash out as the alert becomes background noise? Nobody has run that study. The operators with the hand histories to run it are the same ones still shipping a five-beat default, and until one of them changes it, the question stays open — which is itself a reasonable argument for trying.