Why a deck is different from a die
A die has no memory. Every roll has the same chance of landing on a six, whatever happened before. A deck of cards behaves differently. Each card you draw is gone, so the composition of what remains changes with every draw. If you hold a 60-card deck with four copies of a key card and you draw three cards without seeing one, the remaining 57 cards still contain all four copies, and your next draw is slightly more likely to find one than the first was.
This is called drawing without replacement, and it is the central idea behind every deck calculation. It means draws are not independent events in the strict sense: what you have already seen changes what is possible next. The good news is that the maths is still manageable, because the deck is a fixed, known collection.
The simplest case: one card, a few draws
Suppose you want one particular card, and the deck holds exactly one copy. If the deck has 60 cards and you draw 7, the chance that your copy is among them is simply 7 out of 60, about 11.7 percent. After 10 draws it is 10 out of 60, about 16.7 percent. With a single copy the odds grow in a straight line with the number of cards seen.
That straight line is useful as a sanity check. If someone tells you a single card has a 30 percent chance of appearing in a seven-card hand from a 60-card deck, you know the figure is wrong.
Several copies: asking about at least one
Most real questions involve more than one copy. “How likely am I to see at least one of my four copies?” is easier to answer indirectly. Work out the chance of seeing none, then subtract it from one. This is the same trick used for the probability of at least one success in any repeated trial.
To find the chance of seeing none of four copies in seven cards from a 60-card deck, count the hands that avoid them. There are 56 other cards, so the number of avoiding hands is the number of ways to choose 7 from 56, and the total number of hands is the number of ways to choose 7 from 60. The ratio is roughly 0.60. Subtracting from one gives about 0.40, so you see at least one copy around 40 percent of the time.
Two things are worth noticing. First, four copies in seven cards does not give you anything close to a guarantee. Second, the answer relies on counting outcomes correctly, which is why a firm grasp of permutations and combinations matters. A hand is a set of cards, so the order you drew them in does not matter, and combinations are the right tool.
The hypergeometric idea
The method above is a special case of a general rule. If a deck has N cards, K of which are the type you care about, and you draw n cards, the chance of drawing exactly k of that type is:
(ways to choose k of the K) multiplied by (ways to choose n minus k of the other N minus K), divided by (ways to choose n of the N).
That is the hypergeometric distribution, and it is simply a careful count of favourable hands over all hands. Nothing in it requires memorising a formula if you understand the counting: pick the good cards, pick the rest, divide by every possible hand.
A worked example. In a 60-card deck with four copies of a card, what is the chance that your opening seven contains exactly two of them? Choose 2 of the 4 copies (6 ways) and 5 of the other 56 cards (3,819,816 ways), then divide by the number of seven-card hands from 60 cards (386,206,920). That works out to about 0.059, or roughly 6 percent.
How the numbers move when the deck changes
Deck builders usually care less about one calculation than about how the answer shifts when the list changes. A few patterns hold in most card games:
- Adding copies of a card has diminishing returns. Going from one copy to two in a 60-card deck raises the chance of seeing it in seven cards from about 11.7 percent to about 22 percent. Going from three to four raises it from about 31.5 percent to about 40 percent.
- Drawing extra cards helps, but modestly. With four copies, an eighth card lifts the chance of at least one copy from about 40 percent to about 44.5 percent.
- A smaller deck concentrates everything. The same four copies in a 40-card deck give about a 55 percent chance of at least one in seven cards, compared with about 40 percent in 60 cards.
This last point is one reason why formats with smaller minimum deck sizes feel more consistent. The tradeoff between reliable draws and strong but riskier cards is a design question in its own right, and it is covered in the discussion of deck consistency.
Expected counts versus actual hands
Sometimes you do not need the full distribution, only an average. If K of N cards are of a given type and you draw n cards, the expected number of that type is n times K divided by N. With 17 resource cards in a 40-card deck, a seven-card hand contains on average 7 times 17 divided by 40, which is 2.975.
The average is useful, but it can mislead. Nobody ever holds 2.975 cards. In that same deck, a seven-card hand contains exactly three resource cards about 32 percent of the time, and contains zero or one about 10.5 percent of the time. An average tells you where the middle sits, not how often you land far from it. This is the usual caution about expected value: it summarises, but it does not describe any individual game.
Common mistakes
Treating draws as if cards were replaced. The quick calculation (7 draws, 4 out of 60 each time) overstates or understates the answer because it ignores that the deck shrinks. It is often a decent approximation for rough work, but it is not exact.
Expecting the deck to correct itself. The deck has no memory of what you wish it would do. If you have drawn five blank cards, the remaining cards are simply the rest of the deck. The feeling that something is “due” is the gamblers fallacy, and it applies to shuffled decks just as it does to dice.
Forgetting that the shuffle must be fair. All of this assumes each ordering of the deck is equally likely. A poorly mixed deck can behave differently, which is the subject of the research on shuffle randomness.
Trusting small samples. A handful of games where a card never appears says very little. Sample size matters before you decide that a calculation is wrong or that a list needs rebuilding.
A practical workflow
When you want to test a deck decision, it helps to follow the same few steps each time:
- Write down N, the deck size, K, the number of relevant cards, and n, the cards you will have seen by the turn you care about.
- Decide whether the question is “at least one”, “exactly k” or “at least k”.
- Calculate the chance of the complement when that is easier, then subtract from one.
- Check the result against the single-copy straight-line estimate to catch obvious errors.
- Treat the answer as a guide to typical frequency over many games, not as a prediction for the next one.
Spreadsheets handle these counts easily, and a Monte Carlo method, which means shuffling and drawing a simulated deck many times, is a useful independent check on your hand calculation.
The calculations are only part of deck building. Two decks with identical draw odds can still perform differently because of what the cards do, how fast the format moves, and what opponents are playing. Probability tells you how often you will find a card; it does not tell you whether that card was worth including.