Ask a casual player what card counting buys and the answer sounds like more winning hands. The counter watches the shoe, knows when it turns rich, and suddenly the wins stack up. As far as the numbers go, that picture is mostly wrong.
We measured it directly. The same simulation behind our article on how the player's edge moves with the true count, about 414 million blackjack rounds dealt under fixed rules with a flat one-unit bet, can answer a second question for free: at each count, how many rounds did the player actually win, push, or lose?
The short answer: the count barely changes how often you win a hand, and it moves the average result of a round a good deal more.
The simulation
The setup matches the earlier piece exactly. Six decks, dealer stands on soft 17 and peeks, blackjack pays 3:2, double on any two cards, double after split, split to four hands, one card to split aces with no resplitting, no surrender. The shoe is dealt to a cut card at 75 percent penetration. One player bets one flat unit on every round and plays basic strategy with no count-based plays. Before each round the true count is taken as the Hi-Lo running count divided by the exact decks remaining, then floored to a whole number.
New in this article is the classification. Every round is labeled by its net result after any doubles and splits: a win when the net result is positive, a push when it is zero, a loss when it is negative. A round that splits into three hands and nets one unit won counts as one win, not two wins and a loss.
Here is the full table, with the average result alongside for reference:
| Floored true count | Rounds won | Rounds pushed | Rounds lost | Average result, percent of initial bet |
|---|---|---|---|---|
| -6 or lower | 42.36% | 8.30% | 49.34% | -4.23 |
| -5 | 42.77% | 8.40% | 48.82% | -2.81 |
| -4 | 42.92% | 8.49% | 48.59% | -2.20 |
| -3 | 43.06% | 8.56% | 48.37% | -1.65 |
| -2 | 43.19% | 8.64% | 48.16% | -1.12 |
| -1 | 43.31% | 8.72% | 47.97% | -0.63 |
| 0 | 43.42% | 8.80% | 47.79% | -0.18 |
| +1 | 43.54% | 8.89% | 47.57% | +0.34 |
| +2 | 43.65% | 8.99% | 47.36% | +0.86 |
| +3 | 43.75% | 9.09% | 47.16% | +1.32 |
| +4 | 43.86% | 9.18% | 46.96% | +1.77 |
| +5 | 43.90% | 9.32% | 46.78% | +2.14 |
| +6 or higher | 44.03% | 9.58% | 46.39% | +2.96 |
| All rounds | 43.35% | 8.77% | 47.88% | -0.43 |
What the numbers show
Start with the extremes. At -6 or lower the player wins 42.36 percent of rounds and loses 49.34. At +6 or higher they win 44.03 and lose 46.39. From the bottom row of the table to the top, the win share moves by less than 1.7 points. Our piece on how often each true count occurs covers how often each row comes up.
The middle of the table tells the same story, and it is where most rounds live. From -2 to +2 the share of rounds won rises from 43.19 to 43.65 percent. That is 0.46 points spread over four counts, or about 0.12 points per count. Losses fall from 48.16 to 47.36, which is 0.80 points, 0.20 per count, and pushes rise from 8.64 to 8.99. Over exactly the same four counts, the average result moves from -1.12 to +0.86 percent of the initial bet: 1.98 points.
The gap between those two figures is the whole article. The average result moves roughly four times as many points as the win share, 1.98 against 0.46. At +4 the player still loses more rounds than they win, 46.96 against 43.86, and yet averages +1.77 percent. The count moves the average result of a round far more than the share of rounds won.
21 TrainerFree on iPhone and Android. Instant right or wrong feedback with a clear explanation on every hand.
Get the app
Where the rise comes from
Start with how an average can be positive while wins are a minority of outcomes: winning rounds are not all worth one unit. A natural pays 3:2, a doubled hand that wins pays two units, and split hands can pay several at once. Losses mostly cost exactly one unit, with doubled and split losses as the exceptions. The +4 row is the proof: with fewer rounds won than lost, its average can only be positive if the average winning round pays more than the average losing round costs.
Frequency still does part of the work. From -2 to +2 the table shows 0.46 points more wins and 0.80 points fewer losses. If every win paid exactly one unit and every loss cost exactly one, those two shifts alone would move the average by 0.46 plus 0.80, or 1.26 points. The measured move is 1.98, so the other 0.72 exists only because some rounds win or lose more than one unit.
The one cause our simulation measures directly is naturals. The edge article's table puts the share of rounds with a player natural at 4.40 percent at -2 and 5.33 at +2, a rise of 0.93 points. The dealer's naturals were not recorded, but by symmetry they should rise by the same amount over the same counts, since both hands are dealt from the same shoe. The payoffs are not symmetric: the player's natural wins one and a half units while the dealer's natural costs one. Netting those out, the extra naturals are worth roughly 0.93 times 0.5, about 0.47 points, roughly a quarter of the 1.98 point rise. Treat that as an estimate resting on the symmetry assumption, not a measurement.
The remaining three quarters of the rise is not broken down by cause in this simulation. Better doubling hands and more frequent dealer busts are two reasons commonly cited, and both are plausible: more tens remaining helps a doubled hand and hurts a dealer who must draw to a stiff. We report both as commonly cited and leave the numbers out.
Rounds, hands and other people's numbers
These figures are round results, and the definition is worth stating honestly, because other sources quote slightly different shares. Our piece on how often a blackjack hand pushes gives the commonly published push rate as "roughly 8 to 9 percent of hands in standard six-deck games where the dealer stands on soft 17", about one hand in twelve. Our simulation's overall push share, 8.77 percent of rounds, sits inside that range.
The two are not measured the same way, so any gap between them is one of definition and rules, not error. A hand-level rate counts each split hand separately, while a split round here is one pooled result, and the push article notes that the soft 17 rule, and to a small degree the number of decks, move the push rate too. Change those choices and the shares shift slightly. The comparison above does not depend on which definition is used, because every row of the table uses the same one.
What this means for learning to count
The practical takeaway is about expectations. A counter does not experience more winning hands in any way they could feel at the table. Nobody senses 0.12 points of win rate per count: a hundred rounds at +2 yield on average about half of one extra win compared with a hundred at -2. The advantage shows up only in the average over a great many rounds: slightly fewer losses, slightly more wins and pushes, and some results worth more than one unit.
Counters act on that shift in the average through two channels. The bet channel: our piece on bet spreads explains that a spread is a ratio of largest bet to smallest, and that most of a shoe is dealt at neutral or unfavorable counts. The play channel: our piece on index plays explains that an index is a true count threshold at which a specific basic strategy decision flips. Both are described as mechanism only, not betting advice, and neither turns counting into income.
Keep the scale honest while you learn. A single round swings by about 1.15 betting units per Wizard of Odds, as our article on session length and results unpacks, which dwarfs every average result in the table above. Counting with your head is legal in the United States, casinos are private businesses that may refuse service to suspected counters, and no technique removes risk from the game.
21 Trainer's Pro upgrade teaches counting on that honest scale: four counting systems, Hi-Lo, KO, Omega II and Zen, index deviation training covering the Illustrious 18 and the Fab 4, all with virtual chips and nothing at stake.
Learn to countFree on iPhone and Android. The Pro upgrade adds four counting systems and deviation drills with virtual chips.
Download
Frequently asked questions
Does card counting help you win more hands?
Barely. In our simulation the share of rounds won rose from 43.19 percent at a true count of -2 to 43.65 percent at +2, about 0.12 points per count, and the share lost fell from 48.16 to 47.36, while the average result moved almost 2 points over the same range. The count moves the average result of a round far more than the share of rounds won.
What percentage of blackjack hands does a counter win?
In our simulation the basic strategy player won 43.35 percent of all rounds, pushed 8.77 percent and lost 47.88 percent, and no count row reached even 44.5 percent wins. Definitions matter: these are round results, with every split pooled into one result and pushes counted in the total. Sources that count split hands separately quote slightly different shares.
How can you have an edge while losing more hands than you win?
Because winning hands are not all worth the same. A natural pays 3:2, and doubled or split hands can win more than one unit, while most losses cost exactly one. At a true count of +4 the player still lost more rounds than they won, 46.96 against 43.86, and yet averaged +1.77 percent of the initial bet.
Why do other sources quote different blackjack win rates?
Definitions and rules differ. Our table counts each round once, pooling split hands into a single net result, and reports pushes as their own share of the total. Other sources may count each split hand separately or use other rules, such as a dealer who hits soft 17 or a different number of decks, and either choice shifts the shares slightly. Neither approach is an error; they measure different things.
Can you practice card counting without risking money?
Yes. The Pro upgrade in 21 Trainer teaches counting as a skill with virtual chips and nothing at stake, covering four systems, Hi-Lo, KO, Omega II and Zen, plus index deviation training. Counting with your head is legal in the United States, but casinos are private businesses that may refuse service, and no technique removes risk.