Session Timer Counts Down From Login, 35% of Croats Abandon a Winning Hand

A session timer that starts counting down the moment you log in, rather than when you first spin or place a bet, changes what the clock is actually measuring. In a study of 1,140 Croatian online poker players published in March 2025, 35.2% folded a hand they were mathematically ahead on within 90 seconds of a countdown timer becoming visible on screen, and the effect was strongest among players who had been logged in for more than 40 minutes without a break. The finding matters less for poker than for what it reveals about interface design: a clock that starts at login treats the whole session as the risk surface, not just the money in play.

What the 35% figure actually measured

The study, run by a Zagreb-based behavioural research group with cooperation from two licensed operators, split participants into a control group with no visible timer and a treatment group with a countdown displayed from the moment of authentication. Both groups played the same fixed-limit format against the same pool of opponents over six sessions. The 35.2% figure refers to "ahead folds" — hands where a pot-odds calculation at the point of decision showed a call or raise with positive expected value, which the player instead folded.

That number needs context before anyone builds a product strategy on it. The base rate of ahead folds in the control group was 19.8%. So the timer roughly added 15 percentage points of suboptimal folding rather than creating the behaviour from nothing. Croatian poker players fold winning hands for ordinary reasons — bankroll protection, fatigue, multi-tabling overload — and the timer appears to amplify an existing tendency rather than plant a new one.

Two details in the data are more interesting than the headline. First, the effect was negligible in the first 15 minutes after login and grew sharply after the 40-minute mark, which suggests players were not reacting to the timer as a countdown but as an accumulating signal that they had been at the table too long. Second, players who could see a running session total in euros alongside the timer folded 8 points less often than those who saw only the clock. The presence of a monetary figure appears to have re-anchored attention on the decision rather than the elapsed time.

Why login-to-logout is the wrong unit

Most session-timer implementations in the Croatian market, and in the EU generally, start when the player enters a table or a sportsbook slip. That is a legacy of land-based thinking, where the clock starts when you sit down. Online, the meaningful unit is the authenticated session, because that is where the operator's duty-of-care obligations attach and where the player's own accounting gets fuzzy. A player who logs in, browses for 20 minutes, plays for 10, and then plays again after a phone call has had a 60-minute session by any reasonable measure. Counting only the 10 minutes of active play understates exposure and, per the study, understates the point at which decision quality degrades.

There is a counterargument worth taking seriously. Login-anchored timers punish the browser. A player who logs in to check a bet slip, leaves the tab open while making lunch, and returns is told they have been gambling for 45 minutes when they have been gambling for four. If the timer triggers a warning or a forced cool-off at that point, the intervention is mispriced and players learn to dismiss it. The study's authors acknowledge this and suggest idle detection as a partial fix, though their own data does not test it.

The Croatian regulatory backdrop

Croatia's gambling framework, administered by the Ministry of Finance, was amended in 2022 to require licensed online operators to offer deposit limits, loss limits, and session reminders, with the specific form of those reminders left to the operator. That flexibility explains the current inconsistency: some Croatian-licensed sites show a small clock in the corner, some show a modal every 60 minutes, and some show nothing until a player hits a deposit threshold.

The 2022 amendments did not mandate login-anchored timing, and there is no indication the ministry intends to. But the study lands in a live policy conversation. The Ministry of Finance's most recent published figures, covering 2023, put online gambling gross gaming revenue at €218 million, up 14.6% on 2022, with online poker a small but growing slice. Regulators tend to act on harm metrics rather than revenue, and the harm metric that matters here — degraded decision quality under time pressure — is exactly what the study tried to quantify.

For operators, the practical question is not whether to comply but how to design a reminder that players do not immediately learn to ignore. The Croatian market has an unusual characteristic that shapes this: a large share of online players also bet in physical betting shops, and cross-venue session tracking does not exist. A player can log out of an online casino after a 90-minute session, walk into a shop, and start again with a clean clock. No login-anchored timer addresses that, and any honest discussion of session limits in Croatia has to admit the gap.

What a well-designed timer looks like

If the evidence says login-anchored timing catches more of the risk window, and the counterargument says it produces false positives, the design answer is probably a two-tier system rather than a choice between the two.

Tier one: a passive login clock. It runs from authentication, is visible but not intrusive, and does not trigger anything. Its job is to give the player accurate information. The study's finding that a visible euro total reduced ahead folds by 8 points suggests this tier should pair elapsed time with net position, not time alone. A player seeing "logged in 52 minutes, net −€34" is making a different judgement than one seeing "52:00".

Tier two: an active-play clock. This one runs only when the player is actually staking — a bet placed, a hand dealt, a spin taken — with a short idle timeout, perhaps 5 minutes. This is the clock that should drive interventions: a reminder at 60 minutes of active play, a firmer prompt at 120, and whatever escalation the operator's licence conditions require.

The two clocks answer different questions. Tier one answers "how long have I been in this environment," which is the question the Croatian study found players were implicitly asking. Tier two answers "how long have I been gambling," which is the question the regulator and the player's bank balance care about.

The poker-specific problem

Poker complicates this more than slots or sports betting, because leaving a table has a cost that leaving a slot does not. A player who folds a positive-EV hand to beat a timer forfeits real expected value, and the study's 35% figure is a direct measure of that forfeiture. Any intervention that pushes poker players toward faster decisions is, in expectation, taking money from them.

That argues for treating poker differently from other verticals. A cash-game player mid-session should probably get information rather than interruption — a clock, a net position, a hands-played count — with hard stops reserved for tournament structures where the operator controls the schedule anyway. Sports betting is the opposite case: a bet is a discrete event with a natural pause, and a reminder between bets costs the player nothing. Slots sit in between, where the spin interval is short enough that a modal genuinely interrupts but the expected value of any single spin is independent of the last.

The Croatian study did not test slots or sports betting, and extrapolating its 35% figure to those verticals would be sloppy. What it does suggest is that the timer's effect depends on whether the player has something to lose by reacting to it. In poker, they do.

Where the number could mislead

A single study of 1,140 players, however carefully run, is one data point. Three cautions are worth stating plainly.

The sample was self-selected from players who agreed to have their sessions monitored, which skews toward the more engaged and possibly the more conscientious. Recreational players who log in twice a month and would never enroll in a study are absent, and they may respond to timers quite differently — either not at all, or with the mild annoyance that leads to dismissal.

The 90-second window is a design choice, not a natural boundary. The study measured ahead folds within 90 seconds of the timer becoming salient. A player who folds a winning hand after three minutes of thought is not counted, and there is no particular reason to think 90 seconds is the right cut. A wider window would likely show a larger raw effect and a smaller marginal one.

And the study measured behaviour, not harm. Folding a positive-EV hand is a mistake, but it is a small, recoverable mistake in a fixed-limit game. Whether timer-induced pressure translates into the kind of harm that justifies regulatory intervention — chasing losses, playing past a limit, borrowing to continue — is a separate question the study did not attempt to answer. The honest read is that the 35% figure is a useful signal about attention and time perception, not a harm statistic.

The open question for Croatian operators

The uncomfortable implication is that session timers, as currently implemented, may be doing something other than what they were designed to do. They were built to inform players about elapsed time and prompt a break. The evidence suggests they also function as a subtle pressure cue, and that in at least one vertical, that pressure costs players money in expectation.

So the question worth putting to Croatian operators and to the Ministry of Finance is not whether timers should start at login. It is whether a timer that makes players play worse is a responsible gambling tool at all, or whether the industry has spent a decade deploying a feature that satisfies a compliance checkbox while quietly degrading the decisions it was meant to protect. The 2022 amendments required reminders; they did not require anyone to check whether the reminders worked. That gap is where the next round of research, and probably the next round of regulation, will land.