July 23, 2026 · 9 min read · blackjack · house edge · penetration · table selection

Hands per hour: blackjack's real number

Card counters talk about the count. Casinos talk about decisions per hour. Only one of those two groups reliably makes money, and the difference is not an accident of temperament — it is a difference in which term of the equation they treat as the interesting one.

An edge is a rate. It only becomes money when something is repeated. Double the repetitions and you double the result, and the direction of that doubling depends entirely on whose edge is running. The house edge is always on. Yours is on for maybe fifteen percent of the shoe, if the penetration is good. That asymmetry is the whole subject of this post.

One number, measured: hands per hour by occupancy

The most-cited data on this comes from Kilby, Fox and Lucas, Casino Operations Management, second edition, Table 13.1, titled 'Correlation between Players per Table and Decisions per Hour.' Wizard of Odds reproduces it. It reports blackjack hands per player per hour as a function of how many people are sitting at the table, and the drop-off is steeper than most players expect.

Two caveats before the numbers, because they matter more than the numbers do. Kilby footnotes the blackjack row to a study conducted in an Australian casino where seven decks of an eight-deck shoe were dealt before shuffling — that is 87.5% penetration, which is deeper than almost any American six-deck game — and he warns that hands per hour is acutely sensitive to the number of decks used, shuffle time, and the number of decks dealt before shuffling. So treat the absolute figures as that study's, not as a universal constant for your local pit.

What survives the caveats is the ratio, and Kilby says so explicitly: each player at a full table will receive approximately one-quarter the hands per hour they would receive heads-up, regardless of the casino's dealing procedure. The arithmetic in the table agrees — 209 divided by 52 is 4.02. Whatever your game's true pace, sitting down at a full table cuts your personal hand rate by roughly four.

1 player: 209 hands per hour.
2 players: 139 hands per hour each.
3 players: 105 hands per hour each.
4 players: 84 hands per hour each.
5 players: 70 hands per hour each.
6 players: 60 hands per hour each.
7 players: 52 hands per hour each.

The formula the casino actually uses

Robert Hannum, writing in the UNLV Gaming Studies Research Center's Casino Math Guide, gives the industry's standard calculation for what a player is worth: Earning Potential = Average Bet x Hours Played x Decisions per Hour x House Advantage. Four terms, all multiplied. There is no hierarchy among them. A player who plays twice as fast is worth exactly as much extra to the casino as a player who bets twice as much.

Hannum works it twice. A baccarat player betting $500 per hand for 12 hours at 60 hands per hour, against a 1.2% house advantage, is worth $4,320 — that is 500 x 12 x 60 x 0.012. The same $500 bettor playing double-zero roulette at 60 spins per hour is worth about $19,000, because 5.3% is roughly four and a half times 1.2%. Same bet, same hours, same pace; only the edge changed, and the value moved proportionally.

Now hold the bet and the edge still and move only the pace, using the occupancy table above and Bill Zender's 0.69% working house edge for a six-deck basic-strategy game. The result is the same 4x swing, in the player's cost. Nothing about the game changed. The player made no mistakes. The seat count did all of it.

Full table, 52 hands/hour, $25 average bet: 52 x 25 x 0.0069 = $8.97 per hour.
Six players, 60 hands/hour: 60 x 25 x 0.0069 = $10.35 per hour.
Three players, 105 hands/hour: 105 x 25 x 0.0069 = $18.11 per hour.
Heads-up, 209 hands/hour: 209 x 25 x 0.0069 = $36.05 per hour.

Why the casino's win looks so much bigger than its edge

For the twelve months ending December 31, 2025, Nevada's nonrestricted locations won $1,198,992,000 at blackjack — the Gaming Control Board still calls it Twenty One — across an average of 1,947 tables at 110 locations, at a reported win percent of 12.86%. December 2025 alone: $96,798,000 on 1,912 units at 12.18%, with all table, counter and card games together winning $480,390,000 at 14.74%. The calendar-year blackjack win was down 10.51% against the prior period.

That 12.86% is not the house edge, and reading it as one is the most common misreading of Nevada's revenue reports. NGCB's table-game win percent is win divided by drop, where drop is the money exchanged for chips at the table — it is a hold figure. Hannum draws the distinction cleanly: hold percentage is win over drop, while the actual house edge is win over handle, the total amount wagered. His illustration is roulette, where he puts Nevada's hold at about 24% against a true double-zero edge of 5.3%, and notes drily that no honest casino keeps 24% of the money bet on the wheel.

So why is the hold figure nearly twenty times the 0.69% per-hand edge? This next part is our reasoning rather than Hannum's — he says outright that many factors drive drop and hold and declines to unpack them. But the dominant mechanism is obvious once stated: you buy in for $200, and that same $200 gets wagered, won, lost and re-wagered dozens of times before you stand up. The edge applies to each pass. Kilby states the structural half of this directly, observing that as occupancy rises, hold falls — because each seated player is getting fewer hands.

The continuous shuffler is a speed product

A continuous shuffling machine returns discards into the shoe as play proceeds, so the game never stops for a shuffle. Counterintuitively, this slightly lowers the house edge, because it removes the cut-card effect. Wizard of Odds measures the reduction by deck count: 0.113% for one deck, 0.063% for two, 0.034% for four, 0.020% for six and 0.014% for eight. For the six-deck game most readers will face, that is two hundredths of a percentage point in your favor.

And it is still a bad trade, because the same source concludes that for the basic-strategy player a CSM produces a greater expected loss on an hourly basis. The machine gives back 0.020% of edge and takes back far more than that in hands. Zender's numbers point the same way from the casino's side of the desk: he puts the six-deck basic-strategy edge at 0.6873% with a cut card and 0.61873% against a continuous shuffler, a difference of under seven hundredths of a point — trivially small next to the extra rounds you will play.

The manufacturers have never been coy about which half of that trade is the product. Shuffle Master's Form 10-K for the fiscal year ended October 31, 2010 told investors that continuous shuffling reduces or eliminates card counting and shuffle tracking, and that because the shuffler works while the game is played, dealer shuffling downtime is significantly reduced, 'with the potential for a corresponding increase in playing time and win for the casino.' The company's foundational patent, filed in 1998 and granted in 2001, states the premise even more plainly: 'From the perspective of casinos, shuffling time reduces the number of wagers placed and resolved in a given amount of time, thereby reducing revenue.' The Wizard's advice to counters facing one of these is unambiguous — it isn't worth the bother.

Penetration is where a counter's money actually comes from

Penetration is how deep into the shoe the dealer goes before shuffling. It matters more than any other single variable a counter can shop for, and the Wizard of Odds Hi-Lo tables make the case in numbers. All figures below are six decks, dealer stands on soft 17, surrender allowed, double after split, resplit to four hands including aces, playing the Illustrious 18 plus Fab 4 indexes. That rule set is friendlier than a typical Strip shoe, so read the figures as relative, not as a promise.

The comparison that should reorganize your priorities is the third bullet. A modest 1-to-5 spread with five of six decks dealt beats an aggressive 1-to-10 spread with only four decks dealt. Bet spread is the thing counters obsess over and the thing pit staff watch for; penetration is free, invisible, and worth more. The Wizard's threshold is a two-part condition worth quoting properly: six-deck games are worth counting if the rules are favorable and at least 75% or so of the cards are dealt.

Compare that to the return on memorizing more index numbers. Going from the Illustrious 18 plus Fab 4 to the complete index set, at a 1-to-15 spread with five decks dealt, moves the edge from 1.147% to 1.182% — a gain of 0.035 points, against the 0.569 points that deeper penetration delivered. The page says as much: knowing the top indexes gives the counter 80% to 85% of the value of knowing every index number in a six-deck game. Effort spent on table selection returns roughly sixteen times what the same effort spent on memorization does.

1-to-15 spread, 4 of 6 decks dealt (67%): 0.578% player advantage.
1-to-15 spread, 4.5 decks dealt (75%): 0.834%.
1-to-15 spread, 5 decks dealt (83%): 1.147% — nearly double the 67% figure.
1-to-5 spread, 4 decks dealt: 0.157%. Same spread, 5 decks dealt: 0.469% — triple.
1-to-10 spread, 4 decks dealt: 0.368% — less than the 1-to-5 spread with 5 decks dealt.

Other players cost you rate, not edge

It is a common belief that other players at the table 'eat your cards' and hurt your results. Wizard of Odds rejects the first half of that and accepts the second, and the distinction is exactly the subject of this post. Measured in house edge, the number of other players does not matter — they consume good cards when the count is high and bad cards when it is low, and it washes out. But in expected wins per hour, when you have an advantage it helps to have fewer players, because you play more hands in the same amount of time.

Norm Wattenberger's simulations put a number on the consumption side. In a six-deck game his published output has the dealer averaging 2.78 cards per round heads-up and the player about 2.71, ranging from 2.53 to 3 cards per person per round across true counts from -13 to +13. It is not a fixed constant — with two players his dealer figure rises to 2.91 — so the shoe does not deplete in neat multiples. But the direction is clear: more players means the cut card arrives sooner in wall-clock terms and your share of the deal shrinks.

Run the counter's side of Hannum's formula and the stakes become concrete. With a 1-to-15 spread at 83% penetration — a 1.147% advantage — and a $25 average bet across the ramp, a full table pays 52 x 25 x 0.01147 = $14.91 an hour, while heads-up pays 209 x 25 x 0.01147 = $59.93. Same skill, same bankroll, same shoe. Four times the money, because of who else sat down.

Speed is the part you can actually train

Most of what determines a counter's hourly rate is procurement, not skill: find deep penetration, favorable rules, a hand shuffle rather than a CSM, and an empty table. None of that is something you get better at by practicing. What is trainable is holding an accurate count while the game moves at the pace those good conditions produce — which is the pace that makes the conditions worth having in the first place.

That is a genuinely different skill from counting at your kitchen table. Heads-up at a deep six-deck shoe, you may see something in the neighborhood of 200 decisions an hour: a card every second or two, plus playing decisions, plus index deviations, plus keeping a running deck estimate for the true count conversion. Counting perfectly at your own pace and counting perfectly at the dealer's are not the same accomplishment, and only the second one is worth anything.

None of this argues that counting is a good way to make money — the edges above top out near 1% under rules better than you will typically find, and Zender estimates the average real-world player is actually facing about 1.5% because most people do not play accurate basic strategy. What it does argue is that speed is a multiplier on whatever edge is present, in both directions, and that the multiplier is the part of the equation almost nobody rehearses. If you are going to learn any of this, learn it at tempo.

The count is the easy half. Keeping it accurate at a real dealer's pace is the half that takes reps — drill it with the speed turned up until the number stays right without effort.

Practice counting at speed →

Share this post — the link unfurls with its own card.