Is Solitaire Always Winnable? Solvability Explained

Short answer: No universal solitaire guarantee exists. A deal is winnable only if at least one legal path reaches the goal under a precisely defined variant, redeal policy, information model, and starting arrangement. That mathematical property is different from a player’s win rate and from the percentage of deals a particular solver manages to complete. PlaySoli does not state that every Klondike, Spider, or FreeCell deal is winnable SRC-001.

Questions about “the solitaire win rate” sound numerical, but the first task is definitional. Draw 1 with unlimited passes is not the same problem as Draw 3 with limited passes. Ordinary Klondike with hidden cards is not the same information problem as Thoughtful Klondike, where every card identity is known. Spider One Suit and Spider Four Suits share a layout but not the same suit constraints.

This guide explains what researchers, software, and players can legitimately mean by winnable—and how to recognize a percentage that has been applied to the wrong game.

Contents

• Five different questions hidden inside “winnable”
• The rules that change solvability
• What research on Thoughtful Klondike does and does not show
• Klondike, Spider, and FreeCell need separate answers
• Daily deals, random deals, and curated deals
• How to audit a solvability claim
• In brief
• Frequently asked questions

Five different questions hidden inside “winnable”

1. Is this particular deal solvable?

A specific deal is solvable if at least one sequence of legal moves reaches the winning condition. The answer is a property of that deal plus the exact rules.

A failed attempt does not prove unsolvability. It proves only that the attempted path did not reach a win. To prove solvability, one complete legal solution is enough. To prove unsolvability, every relevant legal path must be eliminated or covered by a correct mathematical argument.

2. What proportion of all deals are solvable?

This is a population question. It asks how frequently a deal sampled from a defined distribution is solvable.

The population must be specified. “All shuffled 52-card Klondike deals under Draw 1, unlimited redeals, PlaySoli move rules, and a stated information model” is meaningful. “Solitaire is 80% winnable” is not.

A reported estimate may include a confidence interval, which expresses sampling uncertainty around the estimate. A narrow interval does not repair a mismatched ruleset.

3. What is a player’s win rate?

A player’s win rate is completed games divided by counted attempts under stated conditions. It depends on skill, familiarity, use of undo or hints, whether abandoned deals count, and which deals were offered.

A strong player can have a lower measured rate in Spider Four Suits than a beginner has in Spider One Suit without any contradiction. They are playing different games.

4. What is a solver’s success rate?

A solver may fail because a deal is impossible, because its search limit expires, because its heuristics miss a path, or because the software models a different rule. Solver success is a result of the algorithm and experiment, not automatically the true solvable proportion.

A solver that completes 90% of a sample establishes at least that those completed deals are solvable, assuming its produced moves are legal. It does not establish that the remaining 10% are impossible.

5. What is the chance a player can force a win without seeing hidden cards?

This is a policy question under uncertainty. In ordinary Klondike, a player chooses without knowing every face-down card. A deal can be solvable with perfect knowledge while no strategy guarantees discovery of the solution from the visible position.

Two hidden arrangements can look identical at a decision point yet require opposite choices. A perfect-information solver may choose correctly because it knows the concealed cards; a human playing normally does not have that information.

The rules that change solvability

A percentage belongs to a ruleset. These variables are especially important.

Draw rule

Klondike Draw 1 exposes each stock card directly. Draw 3 exposes up to three cards and makes only the top waste card playable. That changes which stock cards can be accessed and in what order SRC-001 SRC-002 SRC-003.

A result for Draw 1 cannot be transferred to Draw 3 merely because the tableau layout is the same.

Redeal or pass limit

Unlimited stock passes, one pass, and three passes create different state spaces. PlaySoli permits unlimited passes in both Draw 1 and Draw 3 SRC-001. A historical or experimental result with limited passes describes another variant.

Unlimited passes do not make a deal automatically solvable. They prevent stock access from expiring after a fixed number of traversals.

Hidden versus known cards

Ordinary Klondike conceals card identities. Thoughtful Klondike keeps the same physical restrictions but gives the decision-maker knowledge of every card. That informational change can increase the set of deals a policy or solver can complete and changes what “optimal play” means.

FreeCell starts with all 52 cards visible SRC-001 SRC-007. Spider begins with hidden cards but has different movement and dealing structures. Visibility cannot be compared without the rest of the rules.

Empty-column rule

Klondike allows only a King-led move into an empty column. FreeCell accepts one card as a primitive move. Spider accepts a legal single card or same-suit sequence SRC-001. Changing an empty-column rule changes mobility and therefore solvability.

Group-movement rule

Spider restricts group movement to descending same-suit sequences. FreeCell sequence movement depends on available cells and empty columns. A solver that permits a larger group move than the implementation has analyzed a different game.

Automatic actions

Automatic card turning or Spider run removal normally implements an inevitable consequence. Automatic foundation moves can be more delicate: if software advances a card that could legally remain on the tableau, the policy may remove a strategic choice unless the card can be brought back.

An experiment should state whether such moves are forced, optional, or reversible.

Deal distribution

A uniformly shuffled deck, a finite list of numbered deals, a curated daily challenge, and a generator that excludes known failures are different populations. Even perfectly computed percentages cannot be compared unless their deal distributions match.

What research on Thoughtful Klondike does and does not show

A peer-reviewed 2026 study by Charlie Blake and Ian P. Gent used the general-purpose Solvitaire program to estimate the winnability of many precisely defined patience variants SRC-030. For Thoughtful Klondike, in which the player knows the rank and suit of every card, it reports 81.945%, with a 95% confidence interval of ±0.084 percentage points.

That result is valuable because it is precise and its game model is defined. It is also frequently at risk of being oversimplified.

What the figure supports

• It is an estimate for the paper’s defined Thoughtful Klondike variant.
• “Thoughtful” means all card identities are known to the decision-maker.
• The uncertainty statement is part of the result, not decoration.
• The study demonstrates how a common solver can analyze many patience variants under explicit rules.

What the figure does not support

• It is not a measured PlaySoli player win rate.
• It is not automatically the solvable percentage of Draw 3.
• It is not the chance that a normal player can win without seeing hidden cards.
• It is not a universal percentage for every implementation called Klondike.
• It says nothing directly about Spider or FreeCell unless the paper separately defines and reports those variants.

Earlier research has also framed Klondike results as lower bounds, using planning methods that solve some sampled deals without claiming every unsolved deal is impossible SRC-031. The distinction remains essential: demonstrating many wins can raise a lower bound, but proving the exact population value requires stronger coverage.

A correct way to cite the result

A 2026 JAIR study estimates Thoughtful Klondike—where all card identities are known—at 81.945% winnable, with a 95% confidence interval of ±0.084 percentage points under its defined rules SRC-030. This figure should not be presented as the ordinary PlaySoli win rate.

An incorrect way to cite the result

Solitaire is 81.945% winnable.

The second sentence removes the variant, information model, uncertainty, and distinction between deal solvability and player performance.

Klondike, Spider, and FreeCell need separate answers

Klondike

PlaySoli Klondike has hidden tableau cards and unlimited stock passes. Draw 1 and Draw 3 differ in stock exposure SRC-001 SRC-002 SRC-003. These facts immediately rule out a single unqualified PlaySoli “Klondike percentage.”

For a particular loss, several explanations remain possible:

• the deal was unsolvable under the selected rules;
• the deal was solvable, but an earlier choice blocked the path;
• the deal was solvable with complete card knowledge but not forceable from the information available at each decision;
• the player abandoned before finding a viable line.

A replay with the same cards can demonstrate a solution, but a second loss still cannot demonstrate impossibility.

Spider

Spider uses two decks and requires eight complete same-suit King-to-Ace runs. One-, two-, and four-suit versions change how often rank order also creates same-suit mobility SRC-001 SRC-004 SRC-005 SRC-006.

The suit level is therefore not a cosmetic difficulty setting. It changes the structure of legal group movement. Any winnability claim must specify:

• suit count;
• opening layout;
• whether the player knows face-down cards;
• the “no row while a column is empty” rule;
• whether completed runs are removed automatically;
• the deal distribution and solver method.

PlaySoli makes no universal promise that every Spider deal is solvable SRC-001.

FreeCell

FreeCell’s fully visible opening makes it more amenable to complete planning and solver verification. It is also associated with famous numbered-deal collections and a strong reputation for solvability SRC-033. Reputation is not a universal proof.

A FreeCell guarantee can fail for at least three reasons:

1. the initial deal may genuinely have no solution;
2. an implementation may use a different sequence-capacity or empty-column convention;
3. the offered deal set may include arrangements outside a historically studied numbered collection.

PlaySoli deals all cards face up into eight columns, uses four cells and four foundations, accepts a single card in an empty column, and calculates group-move length from available workspace SRC-001 SRC-007. It does not state that every deal is solvable.

Daily deals, random deals, and curated deals

A daily deal usually means that players receive the same arrangement for a date. It does not, by itself, reveal how that arrangement was selected.

Possible systems include:

• deterministic generation from the date;
• a randomly sampled deal saved for the day;
• editorial selection;
• solver-verified selection;
• difficulty-banded selection;
• a fixed numbered sequence.

Unless a product documents curation or solver verification, do not infer that “daily” means guaranteed solvable. The useful property of a shared daily deal is comparability: players can discuss the same starting state.

A random deal also needs definition. Software pseudo-randomness may be reproducible from a seed, while physical shuffling creates a practical random process. Neither term establishes solvability.

A curated deal may have been tested for a desired property, but the property must be stated. Curated can mean visually interesting, difficult, beginner-friendly, or solver-confirmed; it is not a synonym for winnable.

For PlaySoli’s dated mode, see Daily Solitaire.

How to audit a solvability claim

Before trusting or comparing a percentage, require answers to these questions.

1. Which named variant?

Record Klondike Draw 1, Klondike Draw 3, Spider Two Suits, FreeCell, or another exact title. “Solitaire” is insufficient.

2. Which complete rules?

State deck count, tableau construction, foundation or run goal, empty-column rules, group movement, stock draw, redeal limit, and automatic actions.

3. What can the player or solver see?

Distinguish hidden-information play from perfect-information or Thoughtful play.

4. What is the population of deals?

All random shuffles, a numbered set, a daily set, or a filtered sample produce different claims.

5. What outcome was measured?

Look for one of these exact labels:

• proven solvable deals;
• proven unsolvable deals;
• solver completions within a resource limit;
• player wins;
• first-attempt wins;
• wins with undo or restarts;
• estimated population winnability.

6. What uncertainty or bound is reported?

A sample estimate should include its sample size and uncertainty. A lower bound should not be rewritten as an exact rate. A solver timeout should not be rewritten as an impossible deal.

7. Can the result be transferred to PlaySoli?

Compare every material rule. PlaySoli’s unlimited Klondike passes, Draw 3 access, Spider group rule, empty-column deal restriction, and FreeCell workspace model must match before a numerical claim is imported SRC-001.

Worked interpretation examples

Example A: “I lost this Draw 1 deal, so it was unwinnable”

The conclusion does not follow. The attempt establishes a loss along one path. A replay solution would establish solvability; a correct exhaustive proof would be needed for impossibility.

Example B: “A solver won 92 of 100 deals, so 92% are winnable”

At most, the experiment shows that those 92 sampled deals were solvable if the solutions are legal. The remaining eight may be impossible or merely unsolved. The sample also needs a defined distribution and uncertainty.

Example C: “Thoughtful Klondike is about 81.945%, so my personal rate should be close”

No. The research result concerns deal winnability with complete card knowledge under a defined variant. A personal rate includes decision skill and ordinary information limitations SRC-030.

Example D: “FreeCell shows every card, so every deal has a solution”

Complete information helps planning and verification. It does not logically guarantee that every initial arrangement has a legal path to the foundations.

In brief

• Solvability is a property of a deal under exact rules; a win rate is a property of players or solvers under an experiment.
• Draw rule, redeal limit, information, empty-column convention, group movement, automation, and deal distribution all matter.
• The 2026 result of 81.945%, with a 95% confidence interval of ±0.084 percentage points, applies to defined Thoughtful Klondike, not to ordinary PlaySoli play SRC-030.
• A solver failure is not automatically an impossibility proof.
• Daily does not automatically mean curated or guaranteed solvable.
• FreeCell’s open information does not justify saying every deal is solvable.
• PlaySoli does not guarantee every Klondike, Spider, or FreeCell deal SRC-001.

Common mistakes when discussing win rates

Omitting the variant

A bare “solitaire percentage” is almost never publishable. Add the exact game and rules.

Mixing player performance with mathematical winnability

A person’s record reflects choices, conditions, and counted attempts. It does not directly estimate the proportion of solvable deals.

Ignoring hidden information

Perfect-information results can be useful upper references for a decision problem but cannot be presented as ordinary human performance.

Treating unsolved as unsolvable

Search limits and weak heuristics can leave solvable deals incomplete. Reserve “proven unsolvable” for evidence that justifies it.

Borrowing a percentage from a different redeal rule

Limited and unlimited passes are different games for analytical purposes.

Assuming product labels disclose deal selection

“Daily,” “random,” and “numbered” describe access or identity. They do not by themselves state whether a solver checked the deal.

Frequently asked questions

Is classic solitaire always winnable?

No universal claim is justified. “Classic solitaire” can refer to different Klondike draw and redeal rules, and individual starting arrangements can differ in solvability.

What percentage of Klondike deals are winnable?

A percentage requires an exact variant and information model. A 2026 study reports 81.945% for Thoughtful Klondike, with a 95% confidence interval of ±0.084 percentage points, where all card identities are known SRC-030. That is not a general PlaySoli or human win rate.

Is Draw 1 more winnable than Draw 3?

Draw 1 provides more direct stock access, so the analytical problems differ. A precise comparison still requires matching redeal rules, information, and deal samples. Do not infer a numeric gap without source-matched evidence.

Does unlimited redeal make every Klondike deal solvable?

No. It removes a pass limit, but hidden-card order and tableau dependencies can still block completion.

Are all FreeCell deals solvable?

No universal guarantee should be stated. Many studied deal sets have a very high solvable proportion, but some arrangements and rule variations can be unsolvable, and PlaySoli does not promise every deal SRC-001 SRC-033.

Is Spider One Suit always winnable?

PlaySoli does not make that guarantee. One suit removes suit-conflict difficulty, but the exact deal and available legal order still matter.

Can a computer prove a deal impossible?

Yes, in principle, if it correctly models the rules and exhaustively eliminates all legal winning paths or applies sound proofs. A timeout or failed heuristic search is not enough.

Does replaying with undo change whether the deal is solvable?

The underlying deal remains the same if undo does not alter cards. Undo changes the player’s search process and therefore affects measured performance conditions.

Are Daily Solitaire deals guaranteed to have a solution?

Not unless the product explicitly states and documents solver verification or curation for that property. A daily mode primarily means a shared dated arrangement.

Why do different websites publish different win percentages?

They may use different variants, pass limits, information assumptions, samples, solver capabilities, or definitions of a counted win. Compare methodology before comparing numbers.

Related PlaySoli guides

• Strategy without winning promises
• How draw rules change the game
• FreeCell movement and solver modeling
• How the daily mode works
• Direct answers to common solitaire questions

Sources used

• SRC-001 PlaySoli implementation and editorial specification.
• SRC-002 PlaySoli — Klondike Draw 1.
• SRC-003 PlaySoli — Klondike Draw 3.
• SRC-004 PlaySoli — Spider One Suit.
• SRC-005 PlaySoli — Spider Two Suits.
• SRC-006 PlaySoli — Spider Four Suits.
• SRC-007 PlaySoli — FreeCell.
• SRC-030 Blake and Gent — peer-reviewed analysis of Klondike and other patience variants.
• SRC-031 Bjarnason, Fern, and Tadepalli — lower-bounding Klondike with Monte-Carlo planning.
• SRC-033 Michael Keller — FreeCell technical references and deal research.

Material checked: 2026-07-17.

Disputed or unverified facts: No universal PlaySoli solvability percentage is asserted. Numerical results are retained only with their original variant, information model, and uncertainty. Deal-selection methods for a mode should not be inferred unless publicly documented.

Editorial responsibility: PlaySoli Editorial Team.