Yukon: no published figure, so we produced one
Yukon looks like Klondike at first glance, but it has no stock at all. All 52 cards are dealt into seven columns at the start, most of them face up, and the defining rule is generous: you may move any face-up card, together with every card on top of it, whether or not those cards are in order. That makes for big, dramatic moves and a very different sort of puzzle.
When we set out to compare each of our games with the published research, Yukon had nothing to compare against. The major 2026 study from the University of St Andrews, by Blake and Gent, covers dozens of patience games but not Yukon. So this page reports our own measurements of this site's Yukon, with the method in full so you can judge them.
The headline: a solver that could see every hidden card proved 67.0% of 500 deals winnable and 3.4% impossible, and left 29.6% undecided. So between 67.0% and 96.6% of Yukon deals can be won with perfect play.
Our figures, and what little else there is
| Measure | Our study | Published |
|---|---|---|
| Deals proven winnable, every card visible | 67.0% (95% range 62.8–71.0%, 500 deals) | No published Yukon figure found |
| Deals proven impossible | 3.4% | — |
| Deals the solver could not decide | 29.6% | — |
| Deals won by a player that sees only face-up cards and never undoes | 4.5% (95% range 4.1–4.9%, 10,000 deals) | — |
| Cards that same player got home, on average | 7.4 of 52 | — |
The gap between those two players is enormous, and we want to be straight about why. Most of it is the weakness of our simple player, not the difficulty of the deals. Yukon's move-anything rule creates a huge number of options, and a player that only looks a few hundred positions ahead, without Undo, picks badly among them. A thoughtful human will do much better than 4.5%.
Yukon rules as played here
- Seven columns. The first holds one card; the others hold 6, 7, 8, 9, 10 and 11 cards.
- In every column except the first, the last five cards are face up and the rest face down.
- There is no stock and no waste pile. All 52 cards are dealt out before you make a move.
- A card may be placed on a card one rank higher of the opposite colour.
- You may move any face-up card and everything on top of it, even if the cards above it are not in sequence.
- Only a King, with whatever sits on it, may be moved into an empty column.
- Build the foundations up in suit from Ace to King.
How to play Yukon better
Uncover before you arrange
About 1.6 of the four Aces begin face down, on average. Your first priority is turning over hidden cards, because every one you reveal is one fewer surprise. Big group moves are most useful when they expose a face-down card.
Don't let buried Aces worry you
Our player won 4.4% of deals with two or more Aces hidden and 4.5% with fewer. Where the Aces start made almost no difference. What matters is how well you use the group moves.
Think in stacks, not single cards
Because you can move any face-up card with everything above it, the question is always "which card do I want this pile to land on?" Look for moves that put a disordered stack on a card where it will not block anything you need.
Use Kings to open space
An empty column only accepts a King, but a King with a long stack on top of it can clear a great deal of clutter in one move.
Borrow the solver's foresight with Undo
Yukon wins take planning. Our player's wins averaged about 97 moves. Use Undo to explore a sequence, and back out if it buries more than it frees.
Yukon on solitaire-free.uk
It's free, with nothing to install, no account and no email address, and it plays in any current browser, on desktop, tablet or mobile. Enjoy large cards with zoom, Hint, unlimited Undo, Restart, Auto Play, drag and drop or tap to move, and wins, time and moves for the game in hand. Sound, fullscreen and table colours are up to you.
If you like Yukon's style, Alaska uses the same deal with a twist in the building rule.
Yukon mistakes to watch for
- Moving a stack just because you can. The move-anything rule tempts you to shuffle piles about. Make a group move only if it reveals a face-down card or frees one you need.
- Burying low cards under big stacks. A 2 at the bottom of a twelve-card pile is hard to reach. Look at what you are covering before you move.
- Clearing a column with no King to fill it. Only a King can use the space.
- Losing heart early. Our solver proved only 3.4% of deals impossible. A stuck game is usually a sign to Undo and try a different order.
How we ran the research, step by step
The solitaire-free.uk team did this work in September 2026. With no published Yukon result to compare against, we treated transparency as the priority.
- Rules from our code. Columns of 1, 6, 7, 8, 9, 10 and 11 cards with the last five face up, no stock, alternating-colour building, any face-up card moves with everything on it, Kings only in empty columns.
- A program version. We rebuilt Yukon in Python, with both a solver and a simple player.
- Numbered deals. Deals 1 to 10,000 for the simple player; the first 500 of those for the solver. Each is shuffled from its own seed.
- Two players. The solver sees every face-down card, which no human can, so it gives a ceiling. It could search up to 50,000 positions per deal. The simple player sees only face-up cards, looks up to 300 positions ahead and never undoes, so it gives a floor.
- Running it. 500 deals for the solver and 10,000 for the simple player.
- Published comparison. We found none for Yukon, including in the St Andrews study.
- The margin of error. A 95% range is where the true value very probably lies. The solver's proven 67.0% has a range of 62.8% to 71.0%; its undecided deals widen the answer to 67.0–96.6%.
- What the figures can't tell you. Our simple player is weak at Yukon, so 4.5% says little about how often people win. The solver's undecided deals are unknown, not lost.
Behind the numbers: the code
Written in Python 3.11 using just the standard library, Yukon lives in yukon.py, which also handles Alaska. A shared runner spread the work over 19 processes on a 20-core machine. The solver is a depth-first search, limited to 50,000 positions per deal, that automatically plays cards to the foundations only where that is provably safe; if it exhausts every option, the deal is proven impossible. The simple player runs a look-ahead of up to 300 positions on the visible cards, scoring positions by cards home, hidden cards remaining and how many cards are correctly linked. The 10,000 simple-player games took about two minutes and the 500 solver deals about eight. Ranges use the Wilson score interval.
The code and this page were written with help from Claude, an AI assistant made by Anthropic, which we used for writing and reviewing the Python and for drafting. Every number is the output of the code, not of the AI, and we reviewed the results before publishing.
Frequently asked questions
What proportion of Yukon deals can be won?
With every card known, at least 67.0% and at most 96.6% in our study of 500 deals. We know of no published figure.
Why is your computer's win rate so low?
Yukon offers a huge choice of moves, and our simple player, which cannot see hidden cards or undo, chooses poorly. It is a floor, not a typical result.
Is Yukon a tougher game than Klondike?
For our computer players, yes: it won 42.3% of Klondike Turn 1 games but 4.5% of Yukon. Many people find Yukon rewarding once the group-move rule clicks.
Can I move cards that are not in order?
Yes. Any face-up card can be moved with everything on top of it, in any order. Only the card at the bottom of the moved stack has to fit where it lands.
What is similar?
Alaska is its closest relative. For another all-cards-dealt game, try FreeCell. More on our research is on the blog.
By Admin, solitaire-free.uk. Study date: 18 September 2026.