Golf Solitaire: what the first card turned up tells you
Golf is the quickest game on klondike.nz. Thirty-five cards sit face up in seven columns of five, one card goes to the foundation, and you clear the table by playing cards one rank up or down. Everything is visible except the order of the 16-card stock. That makes the opening easy to read — and one moment matters more than any other: the very first foundation card.
We dealt 10,000 Golf games under this site's rules. In 27.4% of them, the first foundation card had no match anywhere on the table, so you had to turn a stock card straight away. Did that doom the game? Not at all. A solver that knew the stock order proved 92.7% of all deals winnable. A simple greedy player that couldn't see the stock won 22.4%.
Much of the difference comes from one rule this site uses: a King can go on an Ace and an Ace on a King. Without that wrap, our solver found only 44.0% of deals winnable.
Golf deals, by the numbers
| What we looked at | With K↔A wrap (this site) | Without wrap | How we got it |
|---|---|---|---|
| First foundation card has no play on the table | 27.4% (95% range 26.5–28.3%) | 31.2% | Inspecting the deal |
| Deals that can be won (knowing the stock order) | 92.7% (92.2–93.2%) | 44.0% (42.2–45.8%) | Exhaustive search, every deal settled |
| Deals proven impossible | 7.3% | 56.0% | Same search |
| Greedy player's wins, not knowing the stock | 22.4% | 3.2% | Plays the longest visible chain, draws when stuck |
| Stock cards left over in the first solution found | about 1.1 | about 0.9 | Exhaustive search |
| Deals | 10,000 | 3,000 |
The published figure for Golf without the wrap, from Blake & Gent (2026), is 45.109%. Our 44.0% (range 42.2–45.8%) agrees.
How Golf works on klondike.nz
Thirty-five cards in view
Seven columns of five cards, all face up. One card starts the foundation pile. Sixteen cards form the stock.
Clearing the course
Clear all 35 cards from the columns onto the foundation.
Moves
- Play the exposed bottom card of any column onto the foundation if it is one rank higher or lower than the top card, regardless of suit.
- On this site, Kings and Aces connect: a King can follow an Ace and an Ace can follow a King.
- When you're stuck, turn the next stock card onto the foundation. There is no redeal.
Reading a Golf deal in ten seconds
Check the first card first
About one deal in four starts with nothing to play. Don't read too much into it: the solver still won 92.7% of all deals. It just means your first move is a stock card.
Look for long chains, not single moves
The greedy player we tested picked the move that set up the longest visible run. Even so, it won only 22.4%, because it couldn't see the stock. Before you play, trace a chain — 6, 5, 4, 5, 6 — as far as it goes, and prefer moves that keep options open in both directions.
Save your stock
The first solutions our solver found left only about one stock card unused. Every card you turn when you had a table move available is one you may miss at the end.
Use the wrap deliberately
Kings and Aces are usually dead ends in Golf. Here they aren't. A King on the foundation can take an Ace, and vice versa, so don't panic when a King turns up.
Dig where it's deep
All columns start at five cards. Keep them roughly even; a column you ignore will trap its bottom cards when you run out of stock.
What the numbers mean for you
With the wrap, more than nine in ten Golf deals can be won — but only by someone who plays like they know what's coming. The gap between 92.7% and the greedy player's 22.4% is almost entirely decisions, which is good news: Golf rewards practice and Undo.
A ten-second checklist for Golf
Golf games are short, and reading the deal takes seconds. Here is a routine that follows from our results.
- Check the first foundation card. In about one deal in four (27.4%) nothing on the table fits. If so, turn a stock card and carry on; it isn't a bad sign.
- Trace the longest chain. Follow the exposed cards up and down in rank as far as you can go, and compare two or three chains before choosing.
- Look for turning points. A chain that goes 7, 8, 7 keeps two directions open. Those are worth more than straight runs.
- Remember the wrap. Here a King connects to an Ace, so a chain can run Q, K, A, 2.
- Keep the columns level. A column that never gets played will be the one left at the end.
- Count your stock. There are 16 cards and no redeal. The solver's winning lines left only about one of them unused, so every card you turn early costs you.
Nine deals in ten can be won with the wrap. The checklist is how you get closer to that than a greedy player's 22.4%.
Play Golf on klondike.nz
Free in any modern browser on computer, tablet or phone, with no download, no sign-up and no email needed. Big cards with zoom; Hint, unlimited Undo, Restart and Auto Play; tap or drag; wins, time and moves for the current game; sound, fullscreen and colour options. Perfect for a five-minute break between jobs.
Where these numbers come from
The klondike.nz team ran this study in September 2026.
- Rules from our code. Seven columns of five face up, a 16-card stock with no redeal, plays one rank up or down, and the King–Ace wrap this site uses.
- A program version. Built to those rules, with a switch to turn the wrap off.
- Numbered deals. Seeds 1 to 10,000 with the wrap and 1 to 3,000 without.
- Looking at each deal. We checked whether the first foundation card had a play.
- Two players. An exhaustive search that knows the stock order and settles every deal, and a greedy player that doesn't know the stock.
- Comparison. Without the wrap, our 44.0% matches the published 45.109%. There is no published figure for the wrap version.
- Error bars. 92.2–93.2% is the 95% range around 92.7%.
- Limits. The exhaustive search knows the stock order, which you don't, so 92.7% is a ceiling.
Under the bonnet
We coded it in Python 3.11 with no outside libraries, in golf.py. A shared runner hands deals to 19 worker processes via multiprocessing; the 10,000-deal run with the wrap took about 12 minutes and the no-wrap run about 7. The solver is a depth-first search over column heights, stock position and foundation rank that remembers dead-end positions, so it can settle every deal exactly. The greedy player plays the table card with the longest visible chain and draws only when stuck. We use Wilson score 95% intervals. An AI assistant — Claude, made by Anthropic — helped us write and review the code and draft this page. The AI didn't produce any of the figures; they are the code's results, checked against the published study.
Questions players ask
Can a King go on an Ace here?
Yes, and an Ace on a King. That rule lifts winnable deals from 44.0% to 92.7% in our test.
Is a deal with no first move a bad deal?
No. It happens in 27.4% of deals, and most deals can still be won.
How many Golf deals are impossible?
7.3% on this site's rules.
Why do I lose more often than 92.7%?
Because that figure assumes you know the stock order. Our greedy player, which doesn't, won 22.4%.
What should I play next?
Try Pyramid, another quick game, or read more on our blog.
By Admin, klondike.nz. Study date: 18 September 2026.