Counting is the foundation
Almost every probability in blackjack comes from one idea: when every outcome is equally likely, the probability of an event is the number of favourable outcomes divided by the total number of outcomes. The difficulty is not the division; it is counting the outcomes correctly. That branch of maths is called combinatorics, and the two-card opening hand is a perfect small example.
Ordered or unordered?
Start with a single 52-card deck. Deal two cards one after another.
- The first card can be any of 52. The second can be any of the remaining 51. That gives 52 x 51 = 2,652 ordered sequences, where “ace of spades then king of hearts” differs from “king of hearts then ace of spades”.
- In blackjack the order of the first two cards does not change the hand, so each hand has been counted twice. Dividing by 2 gives 2,652 / 2 = 1,326 unordered hands.
This count is written as “52 choose 2”. The general rule for choosing k cards from n is n! / (k! x (n - k)!), but for two cards the shortcut n x (n - 1) / 2 is easier. Either way, 1,326 is the number of different two-card hands that can appear from one deck, and each is equally likely when the deck is well shuffled.
Not every kind of hand is equally common
A common mistake is to list hand types and treat each as equally likely. Compare two examples.
- Ace with a ten-valued card. There are 4 aces and 16 ten-valued cards, so 4 x 16 = 64 hands.
- Two with a three. There are 4 twos and 4 threes, so 4 x 4 = 16 hands.
An ace-ten hand is four times as common as a two-three hand, purely because ten-valued cards (ten, jack, queen, king) are so numerous. This is why the totals in blackjack are not evenly spread, and why probabilities of hands such as 20 or 21 are higher than a quick glance at the card values might suggest.
Pairs follow the same pattern. A specific rank offers 4 cards, and choosing 2 of them can be done in 6 ways, so a specific pair has 6 hands, and with 13 ranks there are 13 x 6 = 78 pairs overall. Dividing, 78 / 1,326 = 1 / 17. The article on the pair odds goes into that figure and its variations.
Checking that the counts add up
Because totals must reconcile, you can test your counting. Split the 1,326 hands into pairs and non-pairs:
- Pairs: 78
- Non-pairs: choose two different ranks from 13 (13 x 12 / 2 = 78 rank combinations), then one of 4 suits for each card (4 x 4 = 16 ways), giving 78 x 16 = 1,248
78 + 1,248 = 1,326. The books balance. If a count of your own does not reconcile like this, an error has crept in somewhere, and checking totals in this way is an important habit.
Naturals as a counting problem
A natural is any hand made of an ace and a ten-valued card. We already counted 4 x 16 = 64. Divided by the total:
64 / 1,326 = 0.04827, or about 4.83%
The treatment of how this changes with deck count belongs in the article on the natural blackjack chance. For now, notice how little machinery was needed: one product, one division.
What changes with more decks
With d decks the total number of two-card hands becomes 52d x (52d - 1) / 2. For two decks that is 104 x 103 / 2 = 5,356 hands. A natural needs one of 8 aces and one of 32 ten-valued cards, which is 8 x 32 = 256 hands, and 256 / 5,356 is about 4.78%.
Notice that duplicate cards make the language of “distinct” hands trickier: two identical king of spades cards from different decks produce hands that look the same on the table but count separately. Calculations that treat cards as drawn from an endless supply, the infinite deck shortcut, sidestep this by using fixed card probabilities instead, and they are accurate enough for many purposes.
Beyond two cards
The same method scales up, although the numbers become large quickly. Three cards from a single deck can form 52 x 51 x 50 / 6 = 22,100 hands, and a five-card hand, which is the kind of hand used in poker, can form 2,598,960. The real challenge in blackjack is not the initial count but the fact that the number of ways a round can unfold grows rapidly with every draw, because the player makes decisions and the dealer keeps drawing to a rule.
For that reason, full-round results are not worked out by hand. They are computed by programs that either enumerate everything systematically or draw samples; the family of approaches includes Monte Carlo methods and models of the dealer’s draws as a Markov chain. These are covered under their own headings.
What counting cannot do
Combinatorics gives exact frequencies for a freshly shuffled deck, and nothing more. It does not tell you the next card, it does not reverse the house edge, and it does not turn a short session into a predictable one. Treat these calculations as a way of understanding the game, not of controlling its results.
Gambling can be harmful and outcomes are uncertain. If it is causing you difficulty, a local support service may be able to help.