Learn the probability of drawing important cards in solitaire and how luck affects your chances of winning.
Every card drawn from an unknown position in solitaire — from the stock, from a face-down tableau stack, or from a deal trigger in Spider — is a sample from a specific probability distribution: the set of cards that could occupy that position, weighted equally, given everything the player currently knows. This distribution changes with every piece of new information: each revealed card removes itself from the unknown population and updates the probabilities of all remaining unknown positions. A player who understands this updating process is doing conditional probability in practice — not as a formal calculation, but as an informed habit of tracking which key cards remain unseen and therefore remain possible in unknown positions.
Every card drawn from an unknown position in solitaire — from the stock, from a face-down tableau stack, or from a deal trigger in Spider — is a sample from a specific probability distribution: the set of cards that could occupy that position, weighted equally, given everything the player currently knows. This distribution changes with every piece of new information: each revealed card removes itself from the unknown population and updates the probabilities of all remaining unknown positions. A player who understands this updating process is doing conditional probability in practice — not as a formal calculation, but as an informed habit of tracking which key cards remain unseen and therefore remain possible in unknown positions.
Key cards in solitaire are the cards whose appearance at a specific moment most directly advances the win condition. In Klondike, key cards are the four Aces (which open the foundations) and the cards needed to continue specific uncovering chains. In Golf and TriPeaks, key cards are the rank-adjacent cards that extend the current chain before the stock must be drawn. In Spider, key cards are the highest-rank same-suit cards needed to begin suit consolidation. In Pyramid, key cards are the rank pairs that clear the most pyramid exposure. Each variant has a different set of key cards, but in every variant the probability of drawing a key card at a specific moment follows the same mathematical structure: it is the number of key cards remaining in the unknown population divided by the total number of unknown cards remaining.
This article covers the probability calculations for key card draws in each major variant, how the probability changes as the game progresses, which strategic habits correctly respond to key card probability, and the common mistakes players make by misreading or ignoring key card probability signals.
The basic key card draw probability formula is P(key card on next draw) = k / n, where k is the number of key cards remaining in the unknown population and n is the total number of unknown cards remaining. At the start of a Klondike Turn 1 game, 24 cards are in the stock (unknown) and 28 cards are in the tableau (21 face-down, 7 face-up). If no Aces are visible in the 7 face-up tableau cards, all 4 Aces remain in the unknown population of 24 stock cards plus 21 face-down tableau cards, giving n = 45 unknown cards and k = 4, so P(Ace on next stock draw) = 4/45 ≈ 8.9%. If 1 Ace is face-up in the tableau, k drops to 3 and P(Ace on next stock draw) = 3/45 ≈ 6.7%.
This basic calculation has a compounding property: as known cards accumulate, the unknown population shrinks and the conditional probability of each remaining key card in that population increases. After 15 stock cards have been drawn and neither Ace has appeared in those 15 draws, the probability that the next stock draw is an Ace rises — assuming no Aces appeared in the 15 draws, k stays at 4 (or however many Aces remain unaccounted for) while n decreases from 45 to 30, raising P(Ace on next draw) from 4/45 to 4/30 ≈ 13.3%. This updating — the probability of drawing a key card increases as the unknown population shrinks without that card appearing — is the Bayesian update that experienced players perform intuitively as the game progresses.
The conditional probability framework produces a specific strategic implication that many players miss: the longer a key card has not appeared, the more likely it is to appear in the remaining unknown population, all else being equal. This is not the gambler's fallacy — it does not mean that a key card is "due." It is a mathematically correct statement about conditional probability given a uniformly random deal: if 30 cards have been revealed and the Ace of Spades is not among them, the probability that the Ace of Spades is in the remaining 22 unknown positions is higher than it was at the start of the game, because it has been confirmed absent from 30 positions and must occupy one of the 22 remaining. The strategy consequence: late in a game, when the unknown population is small, each remaining face-down card has a higher probability of being any specific key card — which makes late uncovering moves more valuable in expectation than early uncovering moves on the same face-down card count.
The key card probability calculation assumes a uniformly random deal — which, as established in the shuffle randomness guide, is what the Fisher-Yates algorithm produces. This assumption is generally correct for practical solitaire play, but the deal structure itself creates positional probability differences that affect key card accessibility independently of the random shuffle.
In Klondike, the deal structure distributes cards across 7 columns of depths 1 through 7, with only the top card of each column face-up at the start. The depth distribution means that Aces can be buried at depths 1 through 7 in the face-down stacks — and an Ace buried at depth 7 (the bottom of the deepest column) requires 6 specific uncovering moves before it can be accessed, regardless of its draw probability from the stock. The key card accessibility calculation therefore has two components: the probability that the key card is in an accessible position (stock or shallowly buried) and the conditional probability of reaching it given its depth if buried. A Klondike opening with all four Aces face-down at depths 4 through 7 across the longest columns has much lower expected Ace accessibility than a deal where one Ace is in the stock and the others are at depths 1 through 3 — even if the raw probability of drawing an Ace from the unknown population is identical in both cases, because the depth distribution determines which probability is realised into an actual draw.
In Golf Solitaire, card distribution probability operates differently because all tableau cards are face-up and the key card at any moment is rank-adjacent to the current chain card. The probability calculation is: how many rank-adjacent cards are visible in the current tableau, and what is the probability that the top stock card is rank-adjacent if none of the visible tableau cards extend the chain? In a Golf game where the current chain card is a 7, the key cards are all 6s and 8s. With 4 sixes and 4 eights in a 52-card deck, there are 8 potential chain-continuing cards. If 3 are already played or visible in unplayable tableau positions, 5 remain in the unknown stock population. With 24 stock cards remaining, P(chain continues on next draw) = 5/24 ≈ 20.8%. This precise calculation — doable in real time for Golf because all tableau information is visible — is the foundation of the chain probability evaluation described in the probability strategy guide.
Yukon Solitaire provides the most complete key card probability picture because its complete-information format (all cards face-up from the start) makes the key card population fully enumerable before the first move. In Yukon, the player knows exactly which key cards are accessible (face-up and legally movable) and which are blocked (face-up but under other face-up cards that must be moved first). The key card probability in Yukon is therefore not a draw probability but an accessibility chain probability: given the current board state, what sequence of moves is required to make each key card accessible, and which sequence produces the first key card access in the fewest moves? This accessibility chain analysis is the complete-information version of the key card probability tracking that Klondike and Spider require under hidden information conditions.
The hypergeometric distribution: exact key card draw probabilities without replacement. Drawing cards from a shuffled deck without replacement — the standard solitaire draw — follows the hypergeometric distribution, not the binomial distribution that applies to draws with replacement. The hypergeometric distribution gives the probability of drawing exactly k key cards from a draw of n cards from a population of N total cards containing K key cards total. For solitaire, the most useful application is the cumulative hypergeometric probability: what is the probability of drawing at least one key card within the next n stock draws? This calculation tells the player how likely it is that continuing to draw from the stock will produce a needed key card within a specific number of draws — which directly informs the decision of whether to continue drawing or to evaluate whether a blocking pattern (key card buried in tableau rather than stock) is the actual problem. For example: in a Klondike game where 1 Ace remains unseen, 20 stock cards remain unknown, and 5 of those 20 draws are available before the pass ends, the probability of drawing the Ace at least once in those 5 draws is 1 − (19/20 × 18/19 × 17/18 × 16/17 × 15/16) = 1 − 15/20 = 1 − 0.75 = 25%. The player who computes or estimates this probability correctly knows there is a 75% chance the Ace is NOT in the next 5 draws — probably buried in the tableau — and should prioritise uncovering moves rather than stock draws as the Ace recovery strategy.
Conditional probability updating: the Bayesian framework for key card tracking. Each new card revealed updates the conditional probability of all remaining unknown positions. The formal Bayesian update: P(key card in position X | new information) = P(new information | key card in position X) × P(key card in position X) / P(new information). In practice, this reduces to the simpler counting update described above: remove revealed cards from the unknown population, update k (key cards remaining) and n (total unknown remaining), recalculate k/n. The value of tracking this update explicitly is that it changes the strategic priority of uncovering moves as the game progresses. In the early game, uncovering the leftmost (shortest) face-down column has a moderate key card probability — one of 21 face-down tableau cards, with a few key cards distributed across all 21. In the late game, with 5 face-down cards remaining and 2 key cards still unaccounted for, each uncovering move has a 2/5 = 40% probability of revealing a key card — a dramatically higher expected value that justifies prioritising uncovering over almost all other move types.
Expected value of stock draw timing: when to draw versus when to uncover. The comparison between stock draw probability and uncovering probability allows a formal expected value comparison of the two main information-gaining actions in Klondike. If P(key card in next stock draw) = k_stock / n_stock and P(key card revealed by next uncovering move) = k_tableau / n_tableau, the player should prefer uncovering when k_tableau / n_tableau > k_stock / n_stock, and prefer drawing when the inequality reverses. Early in the game, n_tableau is large (many face-down cards) and k_tableau is small (a few key cards spread across many positions), so the per-move uncovering probability may be lower than the stock draw probability. Late in the game, n_tableau shrinks faster than k_tableau (as key cards that appeared in the stock reduce k_stock faster than uncovering moves reveal key cards from k_tableau, or vice versa), and the comparison shifts. The stock-last principle — draw only after exhausting tableau moves — is a heuristic that approximates this expected value comparison without requiring explicit calculation: in most game states, tableau uncovering moves have higher key card expected value than stock draws, making the heuristic correct in the majority of positions.
In Klondike: prioritise the uncovering chain with highest Ace density. When multiple face-down columns are available to uncover, the column selection decision has key card probability content. If one column's face-down stack is known to be in a region of the deck that was dealt before a known Ace position, and another column's stack was dealt after, the first has higher conditional probability of containing the Ace. In practice, this level of tracking is generally too demanding for real-time play — but the simpler version applies: the column with the fewest face-down cards remaining has the highest probability of Ace accessibility per uncovering move (because each face-down card in a short stack has a higher probability of being the key card than each face-down card in a long stack, given the same key card count in the full face-down population). Always uncover the shortest available face-down column first when two uncovering moves are equally available is the practical approximation of the conditional probability optimum.
In Golf and TriPeaks: calculate chain continuation probability before each stock draw. Before drawing from the stock in Golf or TriPeaks, the player should evaluate: how many rank-adjacent cards are visible in the current tableau? If two or more are visible, the chain can be extended without a stock draw — and the stock draw should be deferred until all visible chain extensions are exhausted. If no visible chain extensions exist, the stock draw is the only continuation option. The probability calculation: count the visible rank-adjacent cards, estimate the total remaining in the unknown stock, and use this to assess the probability that the stock draw continues versus breaks the current chain. This calculation does not require precise arithmetic — the directional assessment ("several rank-adjacent cards visible = chain probably continues; none visible = chain likely breaks on stock draw") is sufficient for correct strategy in most positions. For the complete Golf chain strategy framework, see our algorithms guide on heuristic search and our FreeCell statistics guide for the complete information contrast.
In Spider: track suit completion probability as key card probability. In Spider 2-Suit and 4-Suit, the key card at any moment is the card needed to complete a same-suit sequence from King to Ace so that the completed sequence can be removed to the foundation. The probability of the needed suit-completion card being in the next stock deal (which adds one card to each column) can be estimated from the visible information: how many cards of the needed suit and rank have been seen, and how many remain unseen? In Spider 4-Suit with 104 cards total, tracking the distribution of 13 cards per suit across 10 columns and the undealt stock is complex — but the strategic habit of always knowing which suits are closest to completion (and therefore which key cards have the highest immediate impact) enables the correct prioritisation of same-suit build moves that builds toward the highest-probability suit completion first.
What is the best strategy for improving key card draw probability in solitaire?The key card draw probability itself is fixed by the shuffle — the player cannot change the probability that a specific card occupies a specific unknown position. What the player can do is maximise the expected value of the actions that reveal key cards. Three habits achieve this. First, prioritise short face-down columns for uncovering: shorter stacks have higher key card probability per move because the same number of key cards is distributed across fewer positions. Second, exhaust visible chain extensions before stock draws in Golf and TriPeaks: each visible extension consumes a known non-key-card reveal, which increases the conditional probability that the next stock draw is a key card (by updating the unknown population). Third, use the Bayesian update mentally: after 20 stock draws without seeing the needed key card, recognise that its conditional probability of being in the remaining face-down tableau population has increased, and shift uncovering priority accordingly. These three habits collectively ensure that the player's action selection correctly tracks the conditional key card probability as it updates throughout the game.Which solitaire game has the most predictable key card draw probability?Golf Solitaire has the most predictable key card draw probability because its complete-information tableau makes the current key card population (rank-adjacent cards visible in the tableau) fully enumerable at every moment, and the stock draw probability is calculable precisely from the known remaining stock size. A Golf player who counts visible rank-adjacent cards before each draw knows exactly how many chain-continuation candidates are in the tableau and can estimate the stock probability from the remaining unknown population. Yukon Solitaire has no draw probability at all — all cards are visible from the start — making key card accessibility a pure sequencing calculation rather than a probabilistic one. At the other extreme, Klondike Turn 3 has the least predictable key card draw probability because the three-card draw groups make individual card draw probabilities dependent on the grouping structure of the remaining stock, which is partially observable but not fully trackable without significant mental effort.Can every solitaire game be solved by optimising key card draw probability?No. Optimising key card draw probability — always taking the action with the highest expected key card revelation value — is a component of correct solitaire strategy, but it is neither sufficient alone nor universally applicable. In unwinnable deals, the key cards are inaccessible not because draw probability is low but because their positions create structural blockages (circular dependencies, key card burials beyond accessible depth) that no sequence of draws or uncovering moves can resolve regardless of probability optimisation. In narrow-solution-count winnable deals, the winning path may require a move that has low key card probability in the short term but enables a high key card probability position three to five moves later — and greedy probability optimisation (always take the highest-probability move now) may foreclose the winning path by missing this longer-horizon structure. The complete strategy framework integrates key card probability tracking with sequencing, resource management, and positional assessment — each applying to a different aspect of the decision problem that probability alone cannot fully resolve.
In solitaire, the basic principles of probability revolve around the likelihood of drawing specific cards from a shuffled deck. Each card has an equal chance of being drawn, which is influenced by the number of cards remaining in the deck and the cards already revealed. For example, if you're playing with a standard 52-card deck and 10 cards are already face-up, the probability of drawing a specific card is 1 in 42 (since 42 cards remain). Understanding these principles helps players make informed decisions about when to draw from the stock or play from the tableau.
To adjust your strategy based on the odds of drawing key cards, first, keep track of the cards that have already been played or revealed. This knowledge allows you to calculate the probability of drawing a needed card. For instance, if you need a specific ace and you know two have already been revealed, the odds of drawing the remaining two aces are lower. You might choose to delay drawing from the stock if you believe a more favorable card could be drawn from the tableau instead. Additionally, consider the potential outcomes of each move, weighing the risks against the probability of success.
Mathematical models such as combinatorial analysis and probability trees can help you understand key card probability in solitaire. Combinatorial analysis allows you to calculate the number of ways cards can be arranged, helping you determine the likelihood of drawing specific cards based on the current game state. Probability trees visually represent the possible outcomes of each move, showing how the game progresses with each card drawn. By applying these models, you can better predict the chances of drawing key cards and make more strategic decisions throughout your game.