Yukon Solitaire: a high ceiling hidden behind face-down cards

Yukon deals every card onto the table at the start. There is no stock and no waste pile, so nothing new will arrive to rescue you. In exchange, Yukon hands you an unusually powerful move: you can pick up any face-up card along with everything stacked on top of it, whether those cards are in order or not.

How often does that power turn into a win? We asked a solver that can see the face-down cards to play 500 deals. It won 67.0% of them and proved only 3.4% impossible. The other 29.6% were still open when it hit its search limit. So somewhere between two-thirds and 96.6% of Yukon deals can be won with perfect play.

That is a high ceiling for a game many players find tough. The difficulty is not the deals; it is the 21 face-down cards you cannot see. You can play Yukon here for free, in any modern browser.

Yukon's ceiling in one table

There is no published figure for Yukon in the Blake & Gent (2026) study, so our range is, as far as we know, the only one available for these rules.

A wide gap, and what it tells you

Our no-peeking player won only 4.5% of 10,000 deals. The solver, seeing every card, won at least 67%. We want to be clear about why. Our player is simple: it looks 300 positions ahead using only the face-up cards, and it never undoes a move. Yukon's big group moves create so many options that a short look-ahead often picks a move that buries something important.

A thoughtful person with Undo will do much better than 4.5%. The honest summary is: most Yukon deals can be won, but you have to plan several moves ahead, and the face-down cards make that hard.

One surprise: whether Aces started face down made almost no difference to our player. It won 4.4% of deals where two or more Aces were hidden and 4.5% where at most one was. In Yukon, what matters is how quickly you can open up columns, not where the Aces happen to fall.

Yukon rules on this site

• One 52-card deck in seven columns. The first column has a single card. The other six have 6, 7, 8, 9, 10 and 11 cards, with the last five of each face up.
• No stock pile: the whole deck is dealt out at once.
• You may move any face-up card together with all the cards on top of it.
• The card you move must go on a card one rank higher and of the opposite color.
• Only a King, with whatever sits on it, may fill an empty column.
• The foundations are filled one suit at a time, Ace up to King.

Climbing toward the ceiling

Uncover the tall columns first

The right-hand columns hide up to six cards each. Every card you flip gives you information the solver already had.

Use group moves to dig, not to decorate

Moving a big group just to tidy a column rarely helps. Move groups when it exposes a face-down card or frees a card for the foundations.

Plan for the Kings

Only Kings fill empty columns. A King buried halfway down a column is trouble; getting it onto an empty column early can open everything else.

Keep track of what is blocking each Ace

You can move a buried Ace's whole covering group in one go, provided the bottom card of that group has somewhere to land.

Undo to look ahead

Our player's weakness is that it never backs up. When a group move leaves you stuck, step back and try a different card as the base of the move.

Yukon on the ceiling ladder

How does Yukon compare with the other games our solver played with every card visible?

• Klondike Turn 1: at least 76.6% solved, 2.1% proven impossible.
• Alaska: at least 75.6% solved, 7.2% proven impossible.
• Yukon: at least 67.0% solved, 3.4% proven impossible.
• Klondike Turn 3: at least 66.2% solved, 9.5% proven impossible.

Among these four games, where the solver saw every card, only Klondike Turn 1 had fewer proven-impossible deals than Yukon. Its undecided share, 29.6%, is the highest of the four, most likely because group moves create an enormous number of positions to check. Once the undecided deals are settled by a longer search, Yukon's true ceiling could well be among the highest in the study.

Inside the test, one step at a time

The solitaire-online.us team ran this study in September 2026 to estimate how many Yukon deals on this site can be won.

1. Reading the rules from our game code. The deal (1, 6, 7, 8, 9, 10 and 11 cards, last five face up), no stock, group moves of any face-up card, alternating-color building and Kings-only empty columns all come from this page's game files.
2. Rebuilding the game. We coded Yukon with these rules; Alaska Solitaire shares the same program with a different building rule.
3. Dealing the cards. Seeds 1 to 500 for the solver and 1 to 10,000 for the no-peeking player.
4. Choosing the players. A solver that sees all face-down cards gives the ceiling. A look-ahead player on visible cards only, with no Undo, gives a floor.
5. Running it. The solver had up to 50,000 positions per deal. A deal counted as proven impossible only when the search ran out of positions before hitting that limit.
6. Checking against published research. Yukon is not covered by Blake & Gent (2026), so there is no published figure to compare with.
7. Working out the margin of error. In 19 repeats out of 20, the true share solved would fall inside our 95% range, 62.8–71.0%.
8. What the numbers cannot tell you. 500 deals and a 50,000-position limit leave almost 30% undecided, so the ceiling is a range. Our no-peeking player is weak and should not be read as a typical human result.

The code behind the ceiling

We built the simulation in Python 3.11 using only the standard library. Yukon and Alaska share the module yukon.py, which contains both the solver and the no-peeking player; runner.py spread the deals across 19 worker processes on a 20-core machine. Each shuffle is recreated from its seed.

The solver is a depth-first search that sees every card. It tries moves in turn, backs up from dead ends, and uses a safe automatic rule for sending cards to the foundations when they can no longer be needed in the columns. The no-peeking player searches 300 positions each turn, scoring each by cards on the foundations, face-down cards remaining and how many cards sit in correct order. The 500-deal solver run took about 8 minutes, and the 10,000-deal player run about 2 minutes. Ranges use the Wilson score interval.

Parts of the code and a first draft of this page were written with Claude, an AI assistant from Anthropic, and then reviewed by our team. It did not supply any results; each figure is the output of running the code.

Because each shuffle comes from a numbered seed, anyone with the same code gets exactly the same deals and the same verdicts.

By Admin, solitaire-online.us. Study date: September 18, 2026.

Ceiling questions from players

How many Yukon games can be won?

With every card visible, between 67.0% and 96.6% of our 500 deals. Only 3.4% were proven impossible.

Why does Yukon feel so hard, then?

Because 21 cards start face down and the group moves create a huge number of choices. The ceiling is high; finding the route is the hard part.

Is there a stock pile in Yukon?

No. Every card is dealt at the start.

How is Alaska different?

Alaska Solitaire uses the same deal but builds in the same suit, up or down. Our solver proved more of its deals winnable: 75.6%.

Where should a Yukon player go next?

For a game with no hidden cards, try FreeCell. All our games are on the games page.