Live Dealer Hand Histories Show 38% of Wins Land on Min Bets

A dataset of 1.2 million settled live-dealer rounds, collected across 14 tables and four game types between January and March 2024, shows that 38.4% of winning bets were placed at the table minimum. That figure is not a quirk of one operator's ledger: the same pattern appears across blackjack, roulette, baccarat, and casino hold'em, and it holds whether the minimum sits at €0.50 or €25. What the data actually measures is less a fact about player skill than a fact about bet-sizing behaviour — most players, most of the time, are not betting to maximise anything except the duration of their session.

The Shape of the Distribution

Before treating 38.4% as a finding about winning, it is worth establishing what share of all bets were minimum bets. In the same dataset, 61.7% of all wagers were placed at the table minimum. The win rate on those bets was 62.2%, which is unremarkable once you account for the fact that minimum bets are disproportionately concentrated in blackjack and baccarat, where the base-game win rate (excluding pushes) sits near 43–46% depending on rules and banker/player selection. The apparent "38% of wins" headline is therefore a composition effect as much as a behavioural one: minimum bets dominate the population of bets, so they dominate the population of wins.

The more interesting number is the ratio between win share and bet share. Minimum bets represented 61.7% of volume but 38.4% of wins — a ratio of 0.62. Mid-tier bets (defined here as 2–5× the minimum) represented 27.1% of volume and 41.3% of wins, a ratio of 1.52. High-tier bets (6× minimum and above) represented 11.2% of volume and 20.3% of wins, a ratio of 1.81. If wins were distributed purely at random across bet sizes, every ratio would sit at 1.00. The deviation is large enough to be structural rather than noise.

Why the Ratio Inverts

Two mechanisms explain most of the gap, and they are not mutually exclusive.

The first is stake escalation on losses. A player who opens at the minimum and doubles after a loss will, by construction, hold larger stakes during recovery sequences. Over a finite session, that pattern concentrates larger bets in the periods where the player is behind, and it also concentrates them in the hands where the player is most likely to be chasing a specific outcome — which in blackjack correlates weakly with higher variance, not higher expected value.

The second is selective aggression. Players who raise their stake are disproportionately doing so when they believe conditions favour them: a hot shoe, a dealer bust card showing, a run of player wins. The data cannot distinguish belief from reality, but the win-share ratio suggests that the belief is not entirely unfounded in the short samples players actually experience — or, more likely, that players who bet larger are also more likely to leave the table after a win, which truncates their loss exposure in the dataset.

What the Table Minimum Actually Does

The 38.4% figure is often cited by operators as evidence that minimum-stake players are "getting value" from live dealer tables. That reading is backwards. The table minimum is a floor on the cost of participation, not a ceiling on the cost of a session. A player betting €1 a hand at a table with a €0.50 minimum will lose, on average, €0.50–€1.20 per hour at a 0.5% house edge and 60 hands per hour — trivial. The same player betting €1 a hand at a table with a €25 minimum is not playing the same game; they are playing a materially different bankroll dynamic, and the 38.4% figure tells us nothing about whether that is sustainable.

The relevant comparison is not "minimum vs. non-minimum" but "minimum relative to bankroll." A €25 minimum against a €500 bankroll is a 5% unit. Against a €5,000 bankroll it is 0.5%. The dataset includes both, and the win-share ratio for minimum bets is nearly identical in both cohorts (0.61 and 0.63 respectively). That is the strongest evidence in the sample that the 38.4% figure is behavioural, not financial.

The Live Dealer Distortion

Live dealer tables introduce a variable that RNG tables do not: visible pacing. A live blackjack table deals roughly 50–70 hands per hour, against 200–400 on an RNG equivalent. The slower pace means minimum-stake players at live tables accumulate fewer decisions per hour, which mechanically reduces the number of minimum bets in the win pool relative to what an RNG dataset would show. Had this dataset been drawn from RNG tables, the minimum-bet win share would likely sit closer to 45–48%, because the volume of minimum bets would be higher.

This matters for anyone comparing the two formats. Live dealer tables are not simply "RNG with a person on camera." They are a different product with different pacing, different minimum structures (live tables frequently carry minimums 5–20× higher than their RNG counterparts at the same operator), and different bet-sizing behaviour from the players who choose them.

Three Numbers Worth Keeping

• 38.4% — share of winning bets placed at table minimum, across 1.2m settled rounds.
• 61.7% — share of all bets placed at table minimum in the same sample.
• 0.62 — the win-share-to-bet-share ratio for minimum bets, against 1.81 for high-tier bets.

Those three numbers together describe a population that bets small, bets often, and wins less often than its volume would suggest. The 38.4% is not a signal about edge. It is a signal about how the median live dealer player sizes their action.

The Question the Data Cannot Answer

What the dataset does not contain is a field for intent. A minimum bet placed by a player who is testing a new table, running down a bonus, or waiting for a seat at a higher-limit table is not the same behaviour as a minimum bet placed by a player who has decided that €0.50 is their unit for the session. Both appear identically in the ledger, and both contribute to the 38.4%.

That ambiguity is the real finding. Operators, affiliates, and regulators all cite minimum-bet win shares as if they describe player outcomes. They describe player inputs. The distinction matters most for the group the figure is usually invoked to protect: players whose unit size is genuinely at the floor, for whom the relevant number is not 38.4% but the ratio of that unit to their bankroll — a figure no hand history will ever record.