Auto-Rollover Toggles Cut Manual Rebuys 29% in 500 Tables
A 500-table study of cash-game poker and live-dealer blackjack found that enabling an auto-rollover toggle reduced manual rebuys by 29% over a twelve-week observation window. The effect held across stake bands from €0.50/€1 to €25/€50, though it was weakest at the lowest limits, where players appear to treat the rebuy prompt as a deliberate pause rather than a friction point. The finding matters less for the headline number than for what it implies about default architecture: the toggle does not change how often players go broke, only how often they have to intervene in the process.
What the toggle actually does
An auto-rollover toggle, in the sense used here, is a persistent account-level setting that permits the client to draw a pre-authorised top-up from a segregated balance the moment a table stake falls below a defined threshold. It is not a credit line, not a marker, and not a bonus mechanism. It is closer to a standing instruction than a product feature.
This distinction is important because the term is often conflated with auto-rebuy in slot play, where the mechanic simply re-triggers a spin sequence. In table games the operational question is different: the player is mid-session, often mid-hand in the case of poker, and the decision to top up is a discrete interruption. Removing that interruption is the entire point.
The study cohort was drawn from 500 tables across four operators, split roughly 60/40 between poker (no-limit hold'em and pot-limit Omaha) and live-dealer blackjack variants. Observation ran from 3 February to 27 April 2025. Tables were assigned to treatment or control by operator, not by player, which introduces some clustering risk but avoids the selection bias that would come from letting players self-select into the toggle.
The 29% figure, unpacked
Manual rebuys per table-hour fell from 4.31 in the control group to 3.06 in the treatment group — a 29.0% reduction, with a 95% confidence interval of 26.4% to 31.5%. The absolute number is modest. The relative number is the one that gets quoted, and it deserves scrutiny.
What the data does not show is a corresponding 29% rise in automated top-ups. Automated rollovers occurred at 2.14 per table-hour in the treatment group, meaning total top-up events — manual plus automated — fell by roughly 7% relative to control. Players with the toggle on did not simply shift the same behaviour to a new channel; they topped up somewhat less often in aggregate. Whether that reflects better bankroll discipline or a subtler form of disengagement is not resolved by this dataset.
Where the effect concentrates
The reduction was not uniform. Segmenting by stake band produces a clearer picture than the aggregate figure.
| Stake band | Manual rebuy reduction | Automated rollover rate |
|---|---|---|
| €0.50/€1 | 11% | 0.9/hr |
| €2/€5 | 27% | 1.8/hr |
| €5/€10 | 34% | 2.6/hr |
| €25/€50 | 31% | 2.2/hr |
The €0.50/€1 band behaves differently for reasons that are probably not technical. At micro-stakes, the minimum buy-in is small enough that many players deliberately short-buy and treat the rebuy prompt as a natural break point — a chance to review session length or walk away. Automating that prompt removes a soft commitment device that some players were using, consciously or not.
The €5/€10 band shows the strongest effect, which aligns with the intuition that this is where session continuity matters most: stakes high enough that interruption is costly, low enough that the top-up amounts are routine.
Blackjack versus poker
Live-dealer blackjack tables showed a larger reduction (33%) than poker tables (24%). This is counterintuitive at first, since poker players face more frequent rebuy decisions per hour and one might expect more friction to remove. The likely explanation is that poker players already have well-developed rebuy habits — many multi-table and treat top-ups as routine housekeeping — whereas blackjack players more often play single-table and experience each rebuy as a genuine interruption.
There is also a regulatory dimension. In several jurisdictions, live-dealer blackjack is subject to stricter session-continuity rules than poker, and the toggle had to be implemented differently to comply. The study did not isolate this variable, which is a limitation worth flagging.
The friction argument, and its limits
The headline interpretation — that auto-rollover removes friction and therefore reduces the number of manual interventions — is almost tautologically true. If you automate a task, people perform it manually less often. The interesting question is whether the reduction in manual rebuys corresponds to any change in player welfare or operator economics.
Three findings bear on this.
First, average session length in the treatment group rose by 8.4%, from 71 minutes to 77 minutes. This is a meaningful increase and cuts against the idea that the toggle simply makes existing sessions smoother. Sessions got longer.
Second, voluntary session-ending behaviour — players actively closing the client rather than timing out — fell by 12% in the treatment group. Again, this is consistent with reduced friction, but it is also consistent with reduced opportunities for reflection.
Third, and most importantly for the responsible-gambling question, the study found no statistically significant difference in the proportion of players who exceeded a self-set deposit limit. That proportion was 6.2% in control and 6.4% in treatment, well within the confidence interval. The toggle did not appear to push players past their own guardrails, at least not at the aggregate level.
What the operators did with the data
Two of the four participating operators have since made the toggle opt-in at account creation rather than buried in settings. One has kept it opt-in but added a mandatory seven-day cooling-off period before the toggle can be disabled. The fourth has taken no action, citing the unresolved questions about session length.
These are product decisions, not research findings, and they reflect different weightings of the same evidence. The operator that added the cooling-off period is effectively treating the toggle as a commitment device in its own right — once on, hard to turn off — which is a defensible reading but not one the data directly supports.
The question the study does not answer
The 29% figure is robust within its own terms. Manual rebuys fell, the effect was consistent across operators, and the confidence interval is tight. But the study measured rebuys, not outcomes. It cannot say whether the players who rebought less often were better off, worse off, or simply doing the same thing through a different interface.
The more uncomfortable possibility is that auto-rollover works precisely because it removes a moment of friction that some players were using — deliberately or instinctively — as a checkpoint. If that is true, then the 29% reduction is not a neutral efficiency gain but a transfer of decision-making from the player to the system. The study's own finding that session length rose 8.4% while voluntary session-ending fell 12% is at least consistent with that reading.
What would resolve it is a longitudinal design tracking the same players across toggle states, with outcome measures beyond rebuy counts: net position, session-ending triggers, and self-reported control. Until that exists, the 29% number will keep getting quoted, and the question of what it actually represents will keep getting deferred.