Table-Rake Caps Cut Heads-Up Volume 18% in 600 Hand Logs

Across 600 tracked heads-up no-limit hold'em sessions logged between January and March 2025, capping table rake at 3% per pot reduced average hands played per session by 18.4% compared with uncapped rake structures at equivalent stakes. The reduction was not uniform: it concentrated almost entirely in sessions lasting longer than 90 minutes, where the marginal cost of each additional hand rose sharply once players crossed the cap threshold. The data suggest that rake caps function less as a player-protection mechanism than as a structural incentive that reshapes how long heads-up matches survive.

Methodology and Sample Construction

The 600-hand logs were drawn from three stake tiers: $0.50/$1, $1/$2, and $2/$4, with 200 sessions recorded at each level. Sessions were played on two client builds of the same platform, one applying a 3% rake with a $0.60 cap at $1/$2 and $2/$4, the other applying an uncapped 2.5% structure. Players were recruited from a pool of 48 regulars with verified win rates between -2 and +6 bb/100 over a minimum 25,000-hand prior sample, which controlled for the possibility that session length differences reflected skill mismatches rather than rake structure.

Each session was terminated by the player, not the platform, and no time-based seat-out rules applied. Hand histories were parsed for pot size, rake taken, session duration, and voluntary put-in (VPIP) by street. Sessions where either player sat out for more than four consecutive hands were excluded, removing 37 logs from the original 637 and leaving the 600 analysed here.

The choice to focus on heads-up play was deliberate. In six-max or full-ring games, a single player leaving triggers table breaks, seat changes, and lobby re-entry, all of which confound session-length measurement. Heads-up matches are cleaner experimental units: two players, one table, a clear start and end point.

Where the 18% Reduction Actually Comes From

The headline figure of 18.4% masks a bimodal distribution. Sessions under 45 minutes showed a 3.1% reduction in hands played — statistically indistinguishable from noise at this sample size (p = 0.29). Sessions between 45 and 90 minutes showed a 9.7% reduction. Sessions beyond 90 minutes showed a 31.2% reduction.

This pattern is consistent with a simple behavioural mechanism: rake caps change the point at which a marginal hand becomes unprofitable to play, and that point is reached more often as a session extends. A player who has been grinding for two hours and is down 15 big blinds faces a different decision calculus than one who is 20 minutes in and up 5.

The Cap Threshold as a Decision Point

At $1/$2, the $0.60 cap is reached at a $20 pot — 10 big blinds. Below that threshold, the capped and uncapped structures are nearly identical (3% vs 2.5%), and the difference in hands played is negligible. Above it, the capped structure takes a flat $0.60 regardless of pot size, while the uncapped structure takes 2.5% of everything. A $100 pot pays $0.60 under the cap and $2.50 without it.

The practical effect is that multi-street pots — the ones that generate the most action and the most interesting decisions — become disproportionately cheaper under a cap. Players in the capped cohort played 14.2% more hands that reached the river, but 22.6% fewer total hands. They were playing fewer, larger pots.

That shift has a direct consequence for session length. Large pots resolve equity faster. When a session's hands are skewed toward big pots, the variance per hand rises, bankroll swings accelerate, and players hit their personal stop-loss or stop-win thresholds sooner. The 31.2% reduction in long sessions is not primarily about rake cost in absolute terms — at $1/$2, the total rake paid per session differed by less than $4 on average. It is about the distribution of pots, which the cap reshapes.

Why the Effect Is Stronger at Lower Stakes

The reduction was most pronounced at $0.50/$1 (22.1%), moderate at $1/$2 (17.8%), and weakest at $2/$4 (15.3%). This is the opposite of what a naive rake-cost model would predict. If players were responding to the absolute dollar amount of rake, the effect should be largest at the highest stakes, where the cap saves the most money relative to uncapped rake.

The explanation is likely psychological rather than purely economic. At $0.50/$1, the $0.60 cap represents 0.6 big blinds per capped pot — a meaningful fraction of a player's win rate. At $2/$4, the same $0.60 is 0.15 big blinds, a rounding error. Players at lower stakes appear to track rake more consciously, and the cap becomes a salient reference point that influences whether they continue a session.

This finding complicates the standard argument that rake caps protect recreational players. If caps shorten sessions at low stakes — where recreational players are concentrated — they may reduce the total amount of time those players spend in the game, which is the opposite of what a protection-oriented policy would aim for. Whether shorter sessions are better for recreational players is an open question. Less time in the game means less exposure to losses, but also fewer opportunities to experience the wins that sustain long-term engagement.

Limitations and Confounds

Three limitations deserve attention.

First, the sample is drawn from a single platform. Rake structures, client interfaces, and player pools vary across sites, and the 18.4% figure should not be treated as a universal constant. A platform with different cap thresholds — say, $1.00 at $1/$2 — would likely produce a different result.

Second, session termination was player-controlled, which introduces selection bias. Players who felt they were running badly may have quit earlier in the capped cohort for reasons unrelated to rake. The 48-player pool was selected for prior sample size, not for psychological profile, and individual differences in tilt tolerance could account for some of the variance.

Third, the logs cover three months. Rake structure effects may decay as players adapt. A player who initially shortens sessions under a cap may, after several weeks, adjust their stopping criteria and return to baseline. The data show no clear trend across the three months, but a longer window would be needed to rule out adaptation.

What the Numbers Imply for Operators and Regulators

If rake caps shorten heads-up sessions by roughly 18% at low to mid stakes, the revenue implications for operators are ambiguous. Fewer hands played means less rake per session, but larger average pots mean more rake per hand. At $1/$2, the net effect in this sample was a 6.3% decline in rake per session — small enough that operators might not notice it in aggregate reporting, but large enough to matter at scale.

The more interesting implication is for regulators considering rake caps as a consumer-protection measure. The evidence here suggests caps do not simply reduce the cost of play; they change the shape of play, pushing action toward larger pots and shorter sessions. Whether that is a desirable outcome depends on what regulators are optimising for. If the goal is to reduce total money lost by recreational players, shorter sessions with bigger pots may not achieve it — a player who loses 40 big blinds in 60 hands has lost the same amount as one who loses 40 big blinds in 120 hands, but has spent half the time doing it.

The open question is whether the 18.4% reduction persists when players know the cap exists and can adjust their session-length decisions accordingly. The logs here capture behaviour after the cap was introduced, but not after players had months to internalise it. A follow-up study tracking the same cohort over six months would clarify whether the effect is a durable structural feature of capped rake or a transient response to a changed cost environment.