Solitaire and Mathematics: Win Rates, Probability, and Computational Complexity

Quick Answer: Klondike solitaire has a theoretical win rate of ~79% (if all cards were face-up), but only ~15–25% with standard hidden cards due to incomplete information. FreeCell is solvable in ~99.999% of deals (only a handful of known impossible deals). Solitaire is PSPACE-hard in the general case, meaning no polynomial-time algorithm is known to determine if an arbitrary deal is winnable.

Solitaire looks simple on the surface, but it harbors fascinating mathematical depth. Questions about win probability, computational complexity, optimal strategy, and the existence of unsolvable deals have attracted mathematicians, computer scientists, and statisticians for decades. Here is what the math actually says.

Win Rate Mathematics

Klondike Solitaire Win Rates

Klondike's win rate is one of the most studied questions in recreational mathematics. The answer depends critically on what you assume:

With perfect information (all cards face-up): Computer simulations by Solitaire Laboratory and others estimate the theoretical win rate at approximately 79–82%. This assumes optimal play with complete knowledge of all card positions.

With standard hidden cards (practical win rate): Estimates range from 15–25% for skilled players using Turn-1 (draw one card at a time). The hidden face-down cards introduce irreducible uncertainty — you cannot know whether a move is optimal because key information is hidden.

Why the gap? Hidden information forces players to make probabilistic decisions without knowing actual card positions. A player who can see all cards will take paths that, in the standard game, require "guessing" which hidden card to reveal first.

Turn-3 win rate: Approximately 8–15%, significantly lower than Turn-1 because the restricted stock access makes it much harder to reach needed cards.

FreeCell Win Rates

FreeCell presents a dramatically different mathematical picture:

• All 52 cards are visible from the start — no hidden information
• Standard FreeCell is dealt from 2^32 possible starting positions (in Microsoft's implementation)
• Of these, approximately 8 deals are known to be unsolvable (deals 11982, 146692, 186216, 455889, 495505, 512118, 517776, and 781948 in Microsoft's numbering)
• All other deals are solvable with correct play — theoretical win rate approaches 100%

The practical win rate for a skilled FreeCell player is approximately 80–85%, with games occasionally abandoned incorrectly (deals that were actually winnable).

Spider Solitaire Win Rates

Spider's win rate varies dramatically by configuration:

• 1 suit: ~98% winnable (trivially easy)
• 2 suits: ~50–70% winnable
• 4 suits: ~30–50% winnable (estimates vary widely — this is computationally hard to determine precisely)

Computational Complexity of Solitaire

Is Solitaire NP-Hard?

A landmark 2002 paper by Rustan Leino and colleagues showed that generalized solitaire (with an arbitrary number of tableau columns and cards) is PSPACE-hard — even harder than NP-hard problems. This means:

1. No polynomial-time algorithm is known that can determine whether an arbitrary solitaire deal is winnable
2. The problem is at least as hard as any problem solvable with polynomial space
3. Solving solitaire optimally, in general, requires exponential time in the worst case

What this means practically: For the fixed 52-card game, computers can brute-force solutions (finite search space). But for general solitaire configurations, the problem is computationally intractable.

FreeCell Computational Results

FreeCell, interestingly, is more amenable to computer analysis:

• Because all cards are visible, FreeCell is a perfect-information game
• It can be modeled as a directed graph search problem
• A computer can exhaustively solve any FreeCell deal, though it may take exponential time in the worst case
• The specific unsolvable deals were found through exhaustive computer analysis

Probability in Solitaire Play

Card Draw Probability

When drawing from a well-shuffled stock, the probability of drawing a specific needed card depends on how many remain:

Example (Turn-1 Klondike):

• If you need a red 7 and there are 15 undrawn stock cards, with 1 red 7 remaining:
• P(drawing it next) = 1/15 ≈ 6.7%
• P(drawing it somewhere in stock) depends on how many cards remain accessible

Cascade Combinatorics

The number of distinct Klondike deals is 52! ≈ 8 × 10^67 — a staggeringly large number. Even at a billion deals per second, it would take longer than the age of the universe to enumerate all possible Klondike starting positions.

Expected Value of Undo

In scored digital solitaire, the expected value of using undo is often negative (net score decrease) even when it prevents losing, because many scoring systems penalize undo use. This creates a mathematically interesting decision: undo is beneficial for winning probability but harmful for score maximization.

The Unsolvable Deal Problem

FreeCell Deal #11982

FreeCell deal #11982 in Microsoft's numbering is the most famous unsolvable solitaire deal. It was identified through exhaustive computer search. The deal:

4C 2C 9C 8C QS 4S 2H
5H QH 3C AC 3H 4H QD
QC 9S 6H 9H 3S KS 3D
5D 2S JC 5C JH 6D AS
2D KD TH TC TD 8D
7H JS KH TS KC 7C
6S 7S 9D 8S 8H 7D
6C 5S AD 4D JD 6H

No sequence of moves can solve this deal. This was proven by exhaustive computer search.

Klondike Unsolvable Deals

For Klondike, the fraction of unsolvable deals is harder to compute precisely because of hidden information. Estimates suggest roughly 20–25% of Klondike deals are mathematically unsolvable regardless of play, with another 55–60% being solvable but only with perfect play.

Statistical Patterns in Solitaire

The Importance of Aces

Statistically, the position of Aces in the initial deal is the single strongest predictor of Klondike win probability. Deals with Aces buried deep under face-down cards have dramatically lower win rates than deals with Aces accessible early.

Suit Distribution Effects

In Spider 4-suit, the initial distribution of suits across columns significantly affects solvability. Uneven suit distribution (many cards of one suit clustered in the same columns) reduces win rate.

Optimal Strategy Mathematics

FreeCell Optimal Play

For FreeCell, optimal play can be defined precisely because all information is known. The mathematically optimal strategy:

1. Always maximize the number of empty cells and columns at each decision point
2. Never make a move that reduces future flexibility unless it advances a foundation
3. Use depth-first search mentally to verify that a sequence of moves doesn't reach a dead end

Klondike Strategy Under Incomplete Information

With hidden cards, Klondike strategy is inherently probabilistic. The mathematically grounded approach:

• Prefer moves that reveal face-down cards (expected information gain)
• Defer moves that don't reveal cards until necessary
• Use Bayesian updating: as cards are revealed, update your estimate of remaining buried cards

For practical strategy applications, see our Klondike beginner tips and advanced FreeCell techniques.

Frequently Asked Questions

What is the mathematical win rate for Klondike solitaire?

With perfect information (all cards visible), Klondike is winnable approximately 79–82% of the time. With standard hidden face-down cards, practical win rates for skilled players are 15–25% for Turn-1 and 8–15% for Turn-3. The gap exists because hidden information forces suboptimal decisions.

Is every FreeCell game winnable?

Nearly all FreeCell games in the standard Microsoft deck are winnable. Only 8 specific deal numbers (out of millions) are known to be unsolvable. These were identified through exhaustive computer analysis. FreeCell deal #11982 is the most famous unsolvable game.

Is solitaire NP-hard?

Generalized solitaire is PSPACE-hard — a complexity class harder than NP. This means no polynomial-time algorithm is known to determine whether an arbitrary solitaire configuration is winnable. For fixed 52-card games, computers can solve specific deals through exhaustive search, but the general problem is computationally intractable.

How many possible Klondike deals are there?

There are 52! (52 factorial) possible ways to arrange a standard 52-card deck, approximately 8 × 10^67. This is such a vast number that computers cannot enumerate all possible Klondike starting positions — each game is effectively unique.

For more on solitaire variants and their unique challenge structures, see our solitaire game variations overview.