Fair Solitaire Every game can be won. We checked. Play FreeCell

Which first moves still win

Every deal here can be won from the start. Can it still be won after your first move? We took the first deals of each game's list, made every different move the game allows at the start, and let our solver play on from each one. In FreeCell a typical deal offers ten first moves, and all ten still win. In Golf a typical deal offers two, and in one deal in three where you have a choice at all, only one of them wins.

What was measured

A first move is anything the game accepts before you have played: a card into a free cell, onto another column or up to its foundation, a turn of the stock, a new row in Spider, a pair or a King in Pyramid. Klondike and the FreeCell games send safe cards to the foundations by themselves. Those automatic moves happen before the first move, as on the table, and again after it.

Moves were counted by where they lead, not by how you make them. A card dropped into the first free cell or the third leaves the same position, and so does a King moved to either of two empty columns. Each counts once. Counting them separately would make a game with four free cells look four times as forgiving as it is. Taking a card back off a foundation was left out, because at the start it only undoes a move the game made for you.

From every new position, the same solver that runs Hint tried to win, with Hint's own search limits. How it checks a win, and how a search proves a loss, are covered in winnable is not the same as easy and the impossible deals. In Golf, Pyramid and TriPeaks it can try every position, so a first move it cannot win is proven to lose the deal, except for six Pyramid moves where it reached its limit first. In the other nine games it gives up after a set number of positions, so a lost move there only means it found no win.

Twelve games side by side

Every first move of the first deals on each game's list
Game First moves in a typical deal Share that still win Deals where only one wins
Klondike 1,500 deals 3 1 to 7 99.2% 0.6% 9 deals
Klondike Turn 3 3,000 deals 3 1 to 7 98.8% 0.6% 17 deals
Spider 1,000 deals 7 1 to 17 97.7% 0.1% 1 deal
Spider 2 Suits 600 deals 7 1 to 16 93.0% 0.2% 1 deal
Spider 4 Suits 120 deals 7 1 to 13 68.4% 1.7% 2 deals
FreeCell 400 deals 10 8 to 16 98.9% 0.3% 1 deal
Baker's Game 4,000 deals 9 8 to 13 83.8% 0.7% 28 deals
Seahaven Towers 1,500 deals 12 9 to 17 87.1% 0.3% 5 deals
Eight Off 5,000 deals 10 8 to 14 99.6% 0% 0 deals
Golf 3,000 deals 2 1 to 6 82.9% 23.9% 717 deals
Pyramid 3,000 deals 4 1 to 9 98.3% 0.2% 5 deals
TriPeaks 10,000 deals 3 1 to 7 99.1% 0.8% 84 deals

First moves: the median number of different first moves, with the fewest and most in any deal. Share that still win: the share of a deal's first moves that keep it winnable, averaged over the deals. Only one wins: deals with two or more first moves where exactly one wins. * The solver gives up after a set number of positions, so in these games the share that still win can only be higher, and the deals where only one wins only fewer, than the table says.

The games that forgive a careless start

Klondike gives you little to choose from: a typical deal offers three first moves, and one of them is always a turn of the stock. Turning one card at a time, that move can always be worked back. Go round the stock once and you are back where you began, perhaps with an Ace already home. The solver won after it in 1,424 of 1,432 deals, and seven of the other eight came good with ten times the search. Turning three, a turn of the stock won in 2,992 of 2,996 deals.

FreeCell gives every card two places in the columns, the two cards of the other colour one rank higher, and four free cells to wait in. A card parked early usually finds its way out. In 385 of 400 deals, every first move still won. Eight Off builds by suit, which is strict, but it has eight free cells and four of them are empty at the start: 99.5% of moves into a free cell kept the deal winnable.

TriPeaks lets Aces and Kings join up, so every card has two neighbours, and the 23 stock cards give the chain plenty of chances to start again. Turning the stock first won in 9,861 of 10,000 deals. Pyramid sends the stock round three times, so drawing is never a waste: it won in all 3,000 deals, and so did every King removed at the start. In one-suit Spider every build is in suit, and 97.7% of builds kept the deal winnable.

The games that do not

Golf is the strictest. A card on the pile never comes back, nothing goes on a King, and the stock goes through once. A typical deal offers just two first moves, and 840 of the 3,000 deals offer only one. Of the 2,160 where you have a choice, 717 allow exactly one good start. Turning the stock to see what comes, a natural way to wait, loses the deal in 909 of 3,000. All of these losses are proven.

Baker's Game and Seahaven Towers build by suit, so every card has one home in the columns: the 7 of hearts goes only on the 8 of hearts. A card put into a free cell can leave only onto that one card, into an empty column (in Seahaven Towers, only if it is a King), or up to its foundation once its suit has climbed to it. Which card you park is a real decision. In Baker's Game, 82.4% of moves into a free cell kept the deal winnable, against 97.2% of moves onto another column. In Seahaven Towers, which starts with two of its four cells full, the figures are 85.4% and 96.9%. In more than half of all Baker's Game deals, at least one first move led to no win.

Spider gets stricter with every suit. Dealing a new row before anything else covers all ten columns at once: the solver found a win after it in 96.5% of one-suit deals, 86.0% with two suits and 40.8% with four. With four suits, a first build onto its own suit kept the deal winnable 96.3% of the time, onto another suit 64.9%. Read the four-suit figures with care. This is the game where our solver is weakest, and a lost move there often means only that it ran out of search.

The narrowest deals

These deals had the fewest winning first moves in each sample. Every link opens that exact deal.

Outside Golf, Pyramid and TriPeaks, every losing move above was tried again with ten times the search, and still no win turned up. FreeCell, Eight Off and Spider with one or two suits had no deal in their samples where only one first move won after that second look. The narrowest Eight Off deals, such as #422, still leave three winning moves out of eight.

Golf #12, move by move

Golf #12 starts with the 10 of hearts on the pile, and three cards fit it: the 9 of hearts on column 4, the 9 of spades on column 5 and the Jack of diamonds on column 6. Only the 9 of hearts wins. Under it lies the 8 of diamonds, which goes next, then the 7 of hearts from column 3.

The other two cards are needed later. In the solver's winning game, the 8 of hearts, the 9 of spades and the 10, Jack, Queen and King of diamonds go onto the pile in one run of six. Spend the 9 of spades or the Jack of diamonds on the first move and that run is broken: the best anyone could do from there is to leave three cards on the table. Turn the stock first and the best is to leave one.

Baker's Game #143, move by move

In Baker's Game #143 the Ace and 2 of hearts go up before you move. No top card fits on another, so all eight first moves put a card into a free cell. Only one wins: the 6 of spades, on top of column 3. Below it lie the 8 of clubs, the 8 of hearts, the 2 of diamonds, the 7 of spades, the Jack of diamonds and the 3 of hearts.

In the solver's game, the 8 of clubs, the 8 of hearts and the 2 of diamonds follow the 6 into the cells, filling all four. The 7 of spades goes onto the 8 of spades, the Jack of diamonds onto the Queen of diamonds, and the 3 and 4 of hearts go home by themselves. After six moves of yours, column 3 is empty. The 8 of diamonds moves into it, the Ace of spades goes home, and the 6 of spades comes out of its cell onto the 7.

That plan needs all four cells at once. Put any other card into a cell first and only three are left. After each of those seven moves, the solver found no way to win, even with ten times its usual search.

What to take to the table

Slow down on move one in Golf, Baker's Game, Seahaven Towers and four-suit Spider. In Golf, follow each card that fits down its column before you play any of them. Treat the stock as a move like any other: turning first loses about three deals in ten, yet in #11153 it is the only way in. In the by-suit FreeCell games, before you park a card, look at what it is sitting on. A cell you will need to clear a whole column is worth more than any single card. In four-suit Spider, build in suit if you can, and do not deal first.

Relax in Klondike, FreeCell, Eight Off, TriPeaks, Pyramid and one-suit Spider. Almost any first move keeps the deal winnable there. The hard choices come later.

Use Hint at the very start. Before your first move, Hint gives you a move on a winning line, so it can never pick one of the losing first moves above. If you have already moved and doubt it, ask Hint: when it finds no way to win from where you are, it says so and offers to take you back, which after a single move means a fresh start. Undo does the same in one tap, and the start can always be won.

What these numbers do not say

Where the numbers come from

Every figure on this page comes from one script, scripts/first-moves-stats.ts, run on 13 November 2026. It tried all 176,064 different first moves of 33,120 deals, the first on each game's list: 1,500 Klondike deals turning one card and 3,000 turning three; 1,000 Spider deals with one suit, 600 with two and 120 with four; the first 400 FreeCell deals, 4,000 of Baker's Game, 1,500 of Seahaven Towers and 5,000 of Eight Off; 3,000 each of Golf and Pyramid; and 10,000 of TriPeaks. The solver searched with Hint's limits: 200,000 positions in the FreeCell games, 150,000 in Klondike, 60,000 in Spider, 1,500,000 in Pyramid and 3,000,000 in Golf and TriPeaks.

Read more: what the first cards tell you, and how many games can be won, game by game.