The deck has a memory, even when the wheel does not
When a blackjack dealer pulls cards from a shoe, the odds of the next card being an ace are not the same as they were at the start of the shoe. Every card dealt removes one possibility from what remains. That is dependent probability in action: the outcome of one event changes the pool from which the next event is drawn. Understanding which games work this way matters more than most players assume, because it changes what you can know mid-game and what you cannot.
Not every casino game behaves like this. A roulette ball does not remember where it landed last spin, and a slot machine's random number generator does not consult its own history. The distinction between dependent and independent events is one of the few structural differences in casino math that reaches the player's side of the table.
What dependent probability means in a casino context
In probability theory, events are dependent when the outcome of one changes the probability of another. In a casino, that almost always means sampling without replacement. A card dealt from a deck or shoe is removed from the remaining pool. A number drawn in keno or bingo is not drawn again in the same game. The pool shrinks, so the probabilities shift with every draw.
Independent events work the opposite way. A roulette wheel has the same 37 or 38 pockets on every spin, regardless of what happened before. A pair of dice has the same 36 equally likely combinations on every roll. A slot spin is resolved by an RNG that generates outcomes independently of the previous spin. The device resets to the same state each time.
The practical difference is whether past outcomes can tell you anything about future ones. In an independent game, they cannot. In a dependent game, they sometimes can, but only if you are tracking the right information and only to the extent that the game's rules allow you to use it.
Games built on dependent probability
Card games dominate this category because cards are drawn from a finite pool without replacement until a reshuffle. The most prominent examples are blackjack, baccarat, poker, and their variants. In each, the composition of the remaining deck or shoe changes as cards are dealt, which shifts the probability of future cards.
Blackjack is the clearest case. If the shoe starts with a known number of decks, the proportion of high cards to low cards changes as play proceeds. A run of small cards leaving the shoe increases the proportion of tens and aces remaining. That shift changes the probability of a natural blackjack, of busting on a hit, and of the dealer busting. Published basic-strategy charts assume a full shoe; they derive from a specific rule set, and they do not adjust for what has already been dealt unless the player is counting cards, which is a separate skill and one that venues actively watch for.
Baccarat also uses a shoe, and card removal changes the probabilities of banker, player, and tie outcomes slightly from hand to hand. The shifts are far smaller than in blackjack because the player makes no decisions about drawing cards. The effect exists mathematically, but it is not something a typical player can act on in a meaningful way.
Poker, whether played against the house or against other players, is built on dependent probability. Your hole cards are removed from the deck, which changes the odds of what your opponents hold and what the community cards will be. The dependence here is not just about the deck; it is also about the information you have. Knowing your own cards lets you update the probability of every other hand, which is precisely what makes poker a game of incomplete information.
Keno and bingo are also dependent-probability games. Numbers are drawn without replacement, so each draw changes the odds of the next. A player who has matched several numbers has a different probability of matching the next one than they did at the start, because the pool of undrawn numbers is smaller.
| Game | Dependent or independent | What changes as play proceeds | Can the player use the information |
|---|---|---|---|
| Blackjack | Dependent | Composition of remaining shoe | Yes, through card counting or basic awareness |
| Baccarat | Dependent | Composition of remaining shoe | Only marginally; no drawing decisions |
| Poker | Dependent | Known hole cards removed from deck | Yes, through hand reading and odds calculation |
| Keno / bingo | Dependent | Numbers drawn without replacement | Only for tracking current game progress |
| Roulette | Independent | Nothing; wheel resets each spin | No |
| Craps | Independent | Nothing; dice reset each roll | No |
| Slots | Independent | Nothing; RNG resolves each spin independently | No |
How dependence changes the house edge
Dependent probability does not automatically make a game better or worse for the player. The house edge is built into the rules and the payouts, not into whether events are dependent. A game with dependent events can still carry a large house edge, and an independent game can carry a small one.
What dependence changes is the amount of information available during play. In a game of independent events, no amount of observation changes your expected loss on the next bet. In a game of dependent events, the remaining pool can shift your expectation, but only if the game gives you a way to act on that shift.
Blackjack is the standard example. The house edge on a given hand is not a single fixed number for the entire shoe. It moves as the composition of the shoe changes. A shoe rich in high cards favours the player in specific ways, primarily because the dealer busts more often on stiff hands under common rule sets. A shoe rich in low cards favours the house. A player who tracks that composition and adjusts bet size or strategy is responding to dependent probability. A player who does not is still playing a game where the long-run edge belongs to the house, just with less information.
Baccarat shows the opposite end of the spectrum. The probabilities of banker and player do shift as cards leave the shoe, but the player has no decisions to make after the bet is placed. The dependence exists, but it is not actionable in any practical sense. The banker, player and tie bets carry different house edges, with the tie bet the highest, and those edges hold over the long run regardless of what has been dealt.
Expected value under dependent and independent events
Expected value is the average result of a bet if it were repeated a very large number of times under the same conditions. The formula does not care whether the events are dependent or independent. What matters is the probability of each outcome and the payout attached to it.
As an example, suppose a game returns 95% of wagers to players over a very large number of rounds. The house edge is then a small fraction of each wager, because RTP and house edge sum to 100% for the same game under the same rule set. If a player wagers 100 units, the expected loss is 5 units. That arithmetic holds whether the game is a slot, a roulette spin, or a hand of blackjack played against a full shoe.
The difference is that in a dependent game, the probability of each outcome can change from round to round. In blackjack, the probability of being dealt a natural changes as the shoe composition shifts. In an independent game, the probability is the same on every round. The expected value calculation is identical in form; it is the inputs that move.
This is why short sessions look nothing like long-run averages. A player who wins 40 units in an hour of blackjack has not disproved the house edge. They have experienced variance. The edge is a long-run statistical average, not a per-session guarantee. A player who loses 60 units in an hour of roulette has not been cheated. They have experienced the same variance in the other direction. The house edge operates over a very large number of spins or hands, and any single session is a small sample.
Decisions per hour and total exposure
One of the least understood consequences of game structure is how many decisions a player makes per hour, and what that does to total exposure. A game with fast rounds and many bets per hour multiplies the house edge across more wagers. A game with slow rounds and few bets per hour does the opposite.
As an example, suppose a player wagers 10 units per round on a game with a 5% house edge. If the game produces 60 rounds per hour, the player wagers 600 units and the expected loss is 30 units. If the same game produces 300 rounds per hour, the player wagers 3,000 units and the expected loss is 150 units. The edge did not change. The exposure did.
Dependent games often involve player decisions that slow the game down. Blackjack hands take time because the player must decide whether to hit, stand, double, or split. A slot spin resolves in seconds. The speed difference does not change the edge on any individual bet, but it changes how quickly the edge accumulates. A player who wants to reduce total expected loss can do so by betting less, playing fewer rounds, or choosing a game with a lower house edge. A player cannot do so by changing the order of their bets.
Why progression systems do not work on dependent games
A progression system is a betting pattern that changes the wager size based on previous outcomes. The most common is the martingale, where the player doubles the bet after every loss. The idea is that a win will eventually recover all previous losses plus a small profit. The idea is wrong.
Progression systems do not change the house edge. They change the distribution of outcomes. A martingale player wins small amounts most of the time and loses a very large amount occasionally. The expected loss is the same as if the player had made the same total wagers at a flat bet size. The system redistributes variance, not edge.
This is true whether the game uses dependent or independent events. A martingale on roulette does not work because the wheel has no memory. A martingale on blackjack does not work because the edge is still built into the rules, even when the shoe composition shifts. No betting pattern changes the built-in edge of a negative-expectation game. The only things under the player's control are bet selection, game choice, speed of play, session length, bankroll sizing, and stopping rules.
Dependent probability can create situations where the player has more information than the house would prefer, which is why blackjack card counting exists as a concept and why venues watch for it. But that is not a progression system. It is a response to changing probabilities, and it requires tracking the shoe accurately, adjusting strategy correctly, and managing bankroll through the swings. It does not remove the house edge; it shifts the edge in specific shoe states, and only for a player who can execute it perfectly.
What a player can control
The distinction between dependent and independent probability matters because it tells you what observation can and cannot do. In an independent game, watching the wheel or the reels is entertainment, not information. In a dependent game, the remaining pool carries information, but only sometimes actionable information, and only within the limits of the game's rules.
The controls that matter are the same across both categories. Choose bets with a lower house edge where the game offers a choice. Play fewer rounds per hour if the goal is to reduce total exposure. Set a bankroll limit before the session and treat it as the cost of the entertainment. Stop when the limit is reached. None of these controls change the edge on any individual bet, but they change the total amount wagered, and that is the only number the player fully controls.
FAQ
Is blackjack the only casino game with dependent probability?
No. Any game where items are drawn from a finite pool without replacement has dependent probability. That includes baccarat, poker, keno, bingo, and some table-game side bets. Roulette, craps, and slots are the main independent-probability games.
Does dependent probability mean a game is beatable?
Not by itself. Dependent probability means the odds shift as the pool shrinks. Whether a player can act on that shift depends on the game's rules. Blackjack offers decisions that respond to shoe composition; baccarat does not. The house edge is built into the rules and payouts either way.
Can past results predict future ones in a dependent game?
Only in the limited sense that the remaining pool is known to have changed. In blackjack, a run of low cards means the remaining shoe is richer in high cards. In an independent game like roulette, past spins carry no information at all. Neither case allows a player to predict the next outcome with certainty.
If you or someone you know is struggling with gambling, help is available. Contact the National Gambling Helpline at the helpline printed in the footer below for free, confidential support, around the clock, every day of the week.
Play responsibly
Gambling should be treated as paid entertainment, never as a way to earn income or recover losses. 1-800-MY-RESET (1-800-697-3738)
18+ or 21+ depending on where you are; follow the minimum age that applies to you.
This article is general information about how these mechanics work. It is not legal advice and not a recommendation to gamble or to use any particular operator. Availability and legality differ by jurisdiction — check the rules that apply where you are.