The core idea
Card counting rests on one fact: in a game where cards are dealt from a deck and not shuffled back after each hand, the cards that have already appeared change what is left. The remaining deck has a different make-up from the full deck, so the probabilities of future cards are different too. Counting is a way of keeping a rough record of that make-up.
This article explains the mathematics. It is not instruction in how to count or play, and it does not suggest that anyone can expect to make money this way. Blackjack played in the ordinary way carries a house edge, meaning the expected result over many hands is a loss for the player, and gambling can cause real harm.
Why blackjack is the usual example
Many casino games reshuffle or use independent random events, so previous results say nothing about the next one. A roulette wheel has no memory of earlier spins, and slot results are generated independently. Blackjack is different because the cards dealt in one round are not returned until the shoe is reshuffled. The game therefore has dependence between rounds, which is the same principle as drawing without replacement.
That dependence is necessary but not sufficient. For counting to matter, the changing composition has to affect the player’s expected result, and in blackjack it does.
Not all cards are equal
In blackjack, aces count as 1 or 11, cards from 2 to 9 count at face value, and the 10, jack, queen and king all count as 10. In a single 52-card deck that means 16 ten-valued cards, a share of 16/52, about 30.8%.
A remaining deck rich in ten-valued cards and aces tends to help the player in several widely described ways: natural blackjacks, which usually pay more than an ordinary win, become more frequent; the dealer, who must follow fixed drawing rules, is more likely to bust when stiff hands meet ten-valued draws; and some player options, such as doubling down, are better supported. Small cards in the remaining deck work the other way: they help the dealer, whose forced drawing is more likely to reach a made hand.
The exact size of each effect depends on the rule set and requires detailed computation, so it is not given here. The point is only that the effect has a consistent direction.
A simple illustration
Take a single deck as a toy example. At the start, 16 of 52 cards are ten-valued, so the chance the top card is ten-valued is about 30.8%.
Now suppose 26 cards have been dealt and, in one scenario, 12 of the 16 ten-valued cards are gone. That leaves 4 among the 26 remaining, and the chance the next card is ten-valued is 4/26, about 15.4%. In another scenario, only 4 ten-valued cards have appeared, leaving 12 among the remaining 26, and the chance is 12/26, about 46.2%.
Both numbers come from the same deck at the same point; they differ only in what has already been seen. This is just conditional probability, and it is the entire mathematical basis for card counting: the probabilities vary with the history, within a game whose cards are not replaced.
Turning composition into a running tally
Tracking every card exactly is hard to do mentally, so counting systems simplify. The general approach assigns each rank a small value (such as plus one for some low cards, zero for middle cards and minus one for tens and aces) and keeps a running total as cards appear. A higher total signals that a larger share of the remaining cards are the ones that favour the player. When several decks are in use, the total is typically adjusted for how many decks remain, because a given surplus matters more in a small remaining pile than in a large one.
Simplification means loss of precision. A running tally is a rough summary of composition, not a full record. Different systems trade accuracy against ease of use, and none is perfect.
What the shift actually amounts to
Here the picture needs to be kept in proportion:
- The shift is small. Even when the remaining composition is favourable, the change in the player’s expectation is typically a modest fraction of a bet. Under normal rules the starting position is still a house edge, and counting may only move the odds partly or temporarily to the other side.
- Outcomes remain random. A favourable count raises the chance of good outcomes slightly. It does not mean the next hand wins. Even with the odds shifted, losing runs of many hands are normal, and ideas like volatility and return and risk of ruin describe how large the swings can be relative to the edge.
- Variance dominates. A small edge needs a very large number of hands before results reflect it, and the law of large numbers says only that averages settle in the long run, not that any short stretch will.
- Conditions matter. The effect depends on how many decks are used, how deep into the shoe dealing goes before reshuffling, and the rules of the particular table. Many games are set up so that the effect is reduced.
Practical and human limits
The mathematics being sound does not mean the activity is practical. Keeping an accurate tally while playing is demanding, and errors remove the small advantage. Casinos are private businesses and commonly watch for players they believe are counting, and they may ask such players to stop or leave. Reshuffling more often, using continuous shuffling machines and other measures are designed to weaken the effect. Large swings in results also strain any bankroll and the player’s judgment, which is why bankroll concepts and chasing losses matter more than the counting itself.
For most people, the practical lesson is not about counting at all. It is that deck composition affects probabilities, and that a game with a built-in disadvantage stays a losing proposition on average unless unusually favourable conditions are met.
Where this leaves the question
Card counting is mathematically possible because blackjack deals from a deck without replacement and because the cards differ in how much they matter to the result. That is a statement about how probabilities respond to information. It is not a statement that anyone will win. If gambling is causing stress, debt or harm, local support services can offer confidential help, and nobody should borrow money to play or try to win back losses.