FreeCell Probability & Mathematics

Every FreeCell deal is generated from a seed number using a pseudorandom number generator (PRNG). The original Microsoft FreeCell used a simple 32-bit linear congruential generator: given the same seed, you always get the same card layout. Deal #1 is always the same. Deal #30000 is always the same. This determinism is what makes FreeCell a shared puzzle — two players on opposite sides of the world can play the exact same deal by entering the same number.

A standard 52-card deck has 52! (52 factorial) possible arrangements — approximately 8.07 × 10⁶⁷. That is a number with 68 digits. The original Microsoft PRNG, with its 32-bit seed, could only generate about 2.1 billion (2³¹) distinct deals, which is a vanishingly small fraction of all possible deals. Modern implementations use larger seed spaces, but the fundamental principle remains: every seed maps to exactly one layout, and every layout is fully determined before the first move.

FreeCell's solvability rate is extraordinary. In the original Microsoft set of 32,000 numbered deals, only one — deal #11982 — has been proven unsolvable through exhaustive computer search. Every other deal in the set can be won with the right sequence of moves.

When researchers extended the analysis to the first one million deal numbers, they found a total of eight unsolvable deals. That gives an insolvability rate of approximately 0.0008%, or equivalently, a solvability rate of roughly 99.999%.

Why is solvability so high? Three factors combine. First, all 52 cards are visible, eliminating the information barriers that make many Klondike deals unsolvable. Second, four free cells provide enough temporary storage to rearrange almost any configuration. Third, the ability to place any card in an empty cascade gives the player enormous flexibility. Together, these mechanics create a game where the solution space is vast enough to accommodate almost any starting arrangement.

Deal #11982 holds a unique place in FreeCell history. It is the only deal in the original Microsoft set that no human and no computer algorithm has ever solved — not because nobody has tried, but because exhaustive search has proven no solution exists. Every legal sequence of moves from the starting position terminates in a dead end.

The Internet FreeCell Project, a distributed community effort in the late 1990s, confirmed solutions for 31,999 of the original 32,000 deals. Only #11982 remained. When solver programs later verified the result algorithmically, the deal's status shifted from "unsolved" to "proven unsolvable."

If 99.999% of deals are solvable, why do most players win only 75–85% of their games? The gap between theoretical solvability and practical win rate is one of the most interesting aspects of FreeCell's mathematics.

A deal being "solvable" means there exists at least one sequence of legal moves that leads to a win. It does not mean that sequence is easy to find. Many solvable deals have a very narrow winning path — perhaps only a few valid opening moves out of dozens of options lead to a solution. One wrong choice early in the game can close off the winning line entirely.

The gap narrows dramatically with undo. A player who uses undo to explore multiple lines of play can approach the theoretical limit because they are effectively testing many possible paths rather than committing to a single one. This is why FreeCell's undo feature is so important — it transforms the game from a single-shot attempt into a systematic search.

The four free cells and the supermove mechanic are the mathematical heart of FreeCell. The maximum number of cards you can move in a single sequence is governed by a precise formula:

Notice the asymmetry. Empty free cells contribute linearly (addition), while empty cascades contribute exponentially (doubling). This mathematical structure is why experienced players treat empty cascades as far more valuable than free cells.

With the maximum configuration (4 empty free cells and 4 empty cascades), you could theoretically move (1 + 4) × 2⁴ = 80 cards at once. In practice, you never have 4 empty cascades because the cards have to be somewhere. But even one empty cascade makes a dramatic difference: it doubles your capacity compared to having none.

In game theory, FreeCell is classified as a perfect information game. All 52 cards are visible to the player from the first move. There are no hidden cards, no random events during play, and no opponent making secret decisions. This puts FreeCell in the same mathematical category as chess and Go, rather than with games like poker or Klondike that involve hidden information.

Perfect information means that, in theory, you could calculate the optimal move at every decision point by analyzing the entire game tree. In practice, the game tree is far too large for a human to analyze completely (the number of possible game states is astronomical), but the principle matters: every piece of information you need to make the best possible decision is right in front of you.

This is the fundamental reason FreeCell feels so different from other solitaire games. When you lose a FreeCell game, it is almost always because you made a suboptimal decision somewhere — not because the cards were against you. That accountability is what makes FreeCell endlessly compelling for players who enjoy strategic depth.