Cards
Cards are the exception: trials that remember
Dice and wheels reset every round. A deck does not, and that single structural difference is responsible for most of what is interesting about card games.
Everything on this desk so far has rested on independent trials. A wheel spun a thousand times presents the same thirty-seven pockets on the thousand and first spin. Cards behave differently. A card dealt from a deck is not returned to it, so every card removed alters the composition of what remains, and with it every probability that depends on that composition.
The arithmetic of removal is elementary and its consequences are not. In a complete fifty-two card deck there are four aces, so the chance that the first card is an ace is four in fifty-two, or one in thirteen. If it is, the chance that the second is also an ace falls to three in fifty-one. If it is not, the chance rises to four in fifty-one.
Working, removal from a 52-card deck
P(first card is an ace) = 4/52 = 0.076923P(second is an ace | first was an ace) = 3/51 = 0.058824P(second is an ace | first was not) = 4/51 = 0.078431the deck's composition, not the dealer, moves the numberWhat conditional probability means at a table
The formal term for this is conditional probability: the probability of an event given that something else has already happened. In games built on dice or wheels, conditioning on the past changes nothing, which is why the gambler's fallacy is a fallacy there. In games built on a shrinking deck, conditioning on the past changes a great deal, and it does so in a way that is fully determined.
This is the structural reason that some card games admit skill in a sense that wheel games do not. The information available to a player genuinely improves as a deal proceeds, because the set of remaining possibilities has narrowed and everyone at the table can observe the narrowing.
What card counting actually measures
Card counting is frequently described as memorising a deck. It is better described as tracking a single summary statistic. In games where high cards favour the player and low cards favour the dealer, a counting system assigns a small positive or negative value to each card as it appears and keeps a running total. That total is a crude estimate of whether the undealt portion of the shoe is currently rich or poor in high cards.
Nothing about the count predicts the next card. It estimates a composition, and composition shifts expectation by fractions of a per cent. The technique is demanding, its effect is small, and it depends on rule details that vary between games: the number of decks in play, how deeply the shoe is dealt before reshuffling, and what the dealer is required to do on particular totals. Those details, not the counting system, set whether any edge exists at all.
How the games answered it
The structural response was to weaken the link between observed cards and future cards. Multiple decks were combined into a single shoe, which dilutes the effect of any one card's removal. Shoes were cut so that a substantial portion is never dealt, which means the composition estimate never becomes precise. Continuous shuffling machines return played cards to the shoe, returning the game to something close to independence and eliminating the phenomenon entirely.
Each of those measures is a mathematical intervention rather than a disciplinary one. They change the sample space so that dependent trials behave more like independent ones, which returns the game to the same footing as a wheel.
Games where the deck is the whole design
Draw and stud games, where players choose which cards to keep, add a further layer: the outcome depends on decisions made under incomplete information about other players' holdings as well as about the deck. Games of that shape are usually analysed as decision problems rather than as fixed-odds propositions, and the counting of hands is done with the same combinatorial tools that the lottery page applies to draws. The number of distinct five-card hands from a fifty-two card deck is a combinations problem, and every hand ranking in every such game is a statement about how many ways each hand can be made.
Working, distinct five-card hands from 52 cards
C(52,5) = (52 x 51 x 50 x 49 x 48) / (5 x 4 x 3 x 2 x 1)= 311,875,200 / 120= 2,598,960 distinct handshand rankings order these by how few ways each can be madeThe ranking of hands is therefore not a convention. It is a frequency table in disguise, arranged from the rarest arrangement to the commonest, and it would be the same in any culture that counted the same deck.