FreeCell ships with a legend. The original Microsoft version told players it was "believed that every game is winnable," and the internet spent years trying to break that claim. It eventually did. Deal #11982 of the original 32,000 became famous as the one that cannot be won, and later analysis of the first million Microsoft deals found just eight impossible ones. That's roughly one unsolvable deal per 125,000. FreeCell isn't a game where every deal is winnable; it's a game where losing to the deal — rather than to yourself — is a once-in-years event.
Every daily deal on Tapbreak FreeCell is verified winnable by a solver before it ships. But we wanted our own number for the wild, so we ran an experiment.
The experiment: 5,000 random deals
We generated 5,000 uniformly random FreeCell deals, not curated and not filtered, the same raw shuffles any classic FreeCell app would hand you, then fed each one to the solver that powers our daily verification and the in-game hint. It plays the game the way a careful human does, just faster: it explores positions in order of promise, tracks what it has already seen, and either finds a winning line or exhausts its budget.
With the everyday budget we use in production, about a tenth of a second per deal, 4,911 of 5,000 deals (98.2%) were proven solvable outright.
The remaining 89 got the heavyweight treatment: a search budget one hundred times larger, up to twenty million positions per deal, minutes of compute instead of milliseconds. That pass cracked 80 more. Final score: 4,991 of 5,000 proven solvable, zero proven unsolvable, plus 9 deals that refused to answer either way.
The 9 that fought back
Those nine are the most interesting result in the data. After twenty-plus million explored positions each, our solver could neither find a winning line nor prove none exists. Are they unsolvable? Almost certainly not. At the known rate of one impossible deal per 125,000, the expected number of genuinely dead deals in a 5,000-deal sample is about 0.04. The overwhelming likelihood is that all nine are winnable — through lines so narrow that even a computer wading through millions of positions didn't stumble onto them in the time we gave it.
Which is worth sitting with for a second: hard to prove is not the same as impossible. If a deal can humble a solver, it will certainly humble a Tuesday-evening human. Losing a FreeCell game says almost nothing about whether the deal was fair. It nearly always was.
Why FreeCell feels different from other solitaire
The reason FreeCell rewards this kind of analysis is that it's a perfect-information game. All 52 cards are dealt face up; there is no stock to reveal, no hidden card to hope for. Whatever happens after the deal is decided entirely by your choices, which is why the solvability question matters here in a way it doesn't for luck-driven solitaires. When 99.98%+ of deals are winnable and you can see everything, every loss has a lesson in it somewhere.
The practical skill that separates wins from losses is capacity management. The number of cards you can move as a unit is (free cells + 1), doubled for every empty column you hold. Four occupied free cells means you can only move single cards; two open cells and an empty column means six. Strong FreeCell play is mostly the discipline of keeping cells empty and treating an open column like gold — spending that capacity only when the move buys something back.
What this means when you play
On Tapbreak, the guarantee is absolute rather than statistical: the daily deal is solver-verified before it reaches you, so the one-in-125,000 dud can never be your daily. When you're stuck, a winning line exists from the opening position, and the Hint button isn't guessing when it nudges you. It plays out a full winning line internally and shows you that line's first move.
And if you want the full library: every past daily lives in the FreeCell archive, free to replay without touching your records. Five thousand deals say the game is almost never against you. Go win one.