Bet Limits at 3 Losses Cut Session Spend 28% in 600 Accounts

A 600-account study of real-money online casino players found that capping deposits after three consecutive losing sessions reduced total session spend by 28.4% over a 90-day observation window, compared with a matched control group operating without the cap. The effect was measured on net deposits rather than turnover, and it held after adjusting for stake size, game type, and baseline spend. The intervention is a soft limit — players could raise or remove it after a 24-hour cooling period — which makes the size of the reduction the more notable result.

Study design and the 28.4% figure

The trial ran from 3 March to 31 May 2024 across six licensed operators in three jurisdictions (Malta, Gibraltar, and Ontario), covering 600 accounts split evenly into treatment and control arms. Randomisation was stratified by historical monthly deposit volume and by primary vertical (slots, live dealer, sports). The treatment arm received a default rule: once a player recorded three consecutive sessions with a net loss, further deposits were blocked for 72 hours unless the player actively overrode the block.

The headline number is a 28.4% reduction in mean net deposits per account over 90 days: €412 in the treatment arm against €575 in control. Median figures moved less dramatically — a 19.1% reduction — which tells you the effect was concentrated in the upper tail of the spend distribution. That is where you would expect a loss-triggered cap to bite, since heavy-spending accounts are more likely to hit three consecutive losing sessions within a short window.

Two design details matter for interpretation. First, sessions were defined by operator-side inactivity thresholds of 15 minutes, not by player-declared session boundaries. Second, "consecutive" was evaluated per calendar day, so a player who lost on Monday, Wednesday, and Friday triggered the block on Friday evening. Both choices are defensible but not neutral; a stricter definition of consecutiveness would likely have produced a smaller and less frequent intervention.

Who the cap actually stopped

Roughly 61% of treatment-arm accounts triggered the block at least once during the window. Of those triggers, 44% were overridden within the 24-hour cooling period. The reduction in spend was almost entirely driven by the 56% of triggers that were not overridden — a reminder that default-on interventions work partly through friction and partly through genuine intent to stop. Players who overrode the block spent at rates statistically indistinguishable from control, which suggests the cap filtered rather than reformed behaviour.

The intervention had no measurable effect on session frequency: treatment accounts opened 11.2 sessions per month against 11.6 in control, a difference within noise. The spend reduction came from smaller average deposits per session (−22.7%) rather than from players logging in less often.

Why a loss-triggered cap behaves differently from a deposit limit

Standard deposit limits are set ex ante and apply uniformly. A loss-triggered cap is conditional: it activates on realised outcomes, which means it targets the specific state — chasing — that most harm-reduction frameworks are trying to interrupt. That is the theoretical case for the design, and the 28.4% figure is consistent with it.

But the mechanism has a well-known failure mode. Players who anticipate the trigger can pre-load deposits before a session they expect to lose, or split play across multiple operators. The study could not observe cross-operator behaviour, and the six participating operators shared no identity graph. A player with accounts at three of them could have triggered the cap at one and continued at another. If that substitution happened at scale, the true reduction in total gambling spend is smaller than 28.4% — possibly substantially so.

There is a second, subtler issue. The cap is triggered by net loss over a session, which includes wins earlier in the session offsetting later losses. A player who is up €200 and then loses €250 across three sessions may not trigger the block at all, even though the behavioural pattern — escalating stakes after losses — is present. Gross-loss triggers would catch more of these cases but would also fire far more often, and the override data suggests players tolerate only so much friction before they route around it.

Operational costs and the override question

Operators in the trial reported implementation costs mainly in customer support: the override flow generated a 3.1× increase in contacts per 1,000 active accounts during the first month, settling to 1.4× by month three. That is a real cost, and it falls disproportionately on smaller operators without automated identity verification.

The override rate is the number regulators should focus on. A 44% override rate means the cap, as deployed, functioned more as a speed bump than a barrier for nearly half of triggers. Whether that is acceptable depends on the policy objective. If the goal is to reduce aggregate spend, the study supports the design. If the goal is to protect a specific subpopulation of at-risk players — the ones most likely to override — the data here are silent, because the study did not stratify by problem-gambling screen score. That is the most significant gap in the design and the one most likely to be addressed in follow-up work.

Responsible-gambling practitioners will recognise the tension. Interventions that are easy to override get overridden by exactly the players who most need them; interventions that are hard to override drive those players to unregulated operators or to accounts the operator cannot see. The 28.4% figure sits somewhere in the middle of that trade-off, and the study does not tell us which side of it is preferable.

What the number does and does not establish

The result is robust to the obvious confounders — stake size, game mix, tenure, and baseline spend — but it rests on a 90-day window during a period with no major sporting calendar disruption. Whether the effect decays as players learn the rule is unknown; the trial's own monthly breakdown shows the reduction shrinking from 34.2% in month one to 21.6% in month three, a trend that should give anyone extrapolating to a 12-month horizon pause.

The more interesting question is whether the trigger threshold is doing the work or whether any friction at the point of deposit would produce a similar effect. A control arm with a random deposit pause — unrelated to losses — would isolate the loss-trigger mechanism from the general friction of a 72-hour block. Until that comparison exists, the study supports the claim in the title but not the causal story most readers will infer from it: that targeting losses specifically is what produced the 28.4%. It may simply be that interrupting a deposit at the moment a player is losing is enough, whatever the stated reason.