Greedy vs. Planned Play: Solitaire Strategy by Game

By Admin, for the spidersolitaireonline.us team.

The short answer

Solitaire is both luck and skill, but the skill part is much bigger than most people think. The deal decides what is possible. How far ahead you look decides how much of that possibility you actually collect.

To put a number on it, we ran our own simulations in September 2026 on the exact rules used on this site. In every game we tested, a player who thinks one move ahead finishes far below what the same deals allow. In Alaska, the difference was between 1.5% and 75.6% (71.6–79.2%) of deals won. That is not a small edge. That is almost the whole game.

Below is what planning is worth, game by game, and the habits that move you up.

How we measured it (and what the numbers are not)

We wrote two kinds of computer players in Python, with help from an AI assistant (Claude, by Anthropic) for the code and drafts, and ran them on freshly shuffled deals. Every figure below comes from running that code on this site's own rules.

The program (the floor). A simple greedy player in Golf and Pyramid, and a look-ahead player in Klondike, Yukon and Alaska. It plays face down, never undoes, never restarts, and never changes its mind.
The solver (the ceiling). A search program that can see every hidden card before it moves and can explore many lines. This is the "thoughtful" version of the game that researchers use.

Read the program results as a floor, not as how often people win. Read the solver results as a ceiling, because no human can see face-down cards. A real player sits somewhere in between, and how far up depends almost entirely on planning. One more caution: our programs are not equally strong from game to game (greedy in some, look-ahead in others), so compare a floor with its own ceiling, not with another game's floor.

The gap, game by game

Game	Program (floor)	Solver that sees all cards (ceiling)
Golf, King–Ace wrap (10,000 deals)	Greedy 22.4% (21.6–23.2%)	92.7% (92.2–93.2%)
Pyramid, 3 passes (4,000 deals)	Greedy 25.2% (23.9–26.5%)	62.2% proven winnable (60.6–63.6%), 23.9% undecided
Klondike Turn 1	Look-ahead 42.3% (10,000 deals)	76.6% (1,000 deals, 73.9–79.1%)
Klondike Turn 3	Look-ahead 10.6% (10,000 deals)	66.2% (1,000 deals, 63.2–69.1%)
Yukon	Look-ahead 4.5% (10,000 deals)	67.0% (500 deals, 62.8–71.0%)
Alaska	Look-ahead 1.5% (10,000 deals)	75.6% (500 deals, 71.6–79.2%)

All ranges above are 95% confidence ranges from our runs. Where we did not record a range, we have not printed one.

1. Golf: the biggest lesson in the shortest game

A greedy player in Golf Solitaire takes any card that is one rank up or down from the waste pile. It cleared 22.4% (21.6–23.2%) of 10,000 deals. The solver cleared 92.7% (92.2–93.2%) of the same kind of deals. Planning is worth roughly seventy points of win rate here.

Why so large? Golf usually offers you a choice of two or three cards. Greedy takes the first one. A planner asks which choice keeps the longer run alive and which one unlocks a buried column. With the King–Ace wrap turned on, chains can loop around the corner, so the card you skip now is often the bridge you need in four moves.

2. Pyramid: the pass you waste is the game you lose

In Pyramid Solitaire with three passes through the stock, the greedy player — take any pair totaling 13 as soon as you see it — cleared 25.2% (23.9–26.5%) of 4,000 deals. The solver proved 62.2% (60.6–63.6%) winnable, and left 23.9% undecided within its search limit. So the true ceiling is somewhere at or above 62.2%.

The planning lesson is exact: before you make a pair, check whether one of those two cards is the only partner for a card still trapped in the pyramid. Pyramid punishes the fast hand more than any other game we tested, because a pair spent early can never be un-spent.

3. Klondike Turn 1: the friendliest gap

On Klondike Turn 1, our look-ahead program won 42.3% of 10,000 deals. The solver won 76.6% of 1,000 deals (73.9–79.1%). That is about thirty-four points of headroom — the smallest gap in our table, and the reason Turn 1 is the best place to practice planning. You can see the whole stock one card at a time, so the information is there if you are willing to use it.

4. Klondike Turn 3: where planning matters most

Switch to Klondike Turn 3 and the program falls to 10.6% over 10,000 deals, while the solver still reaches 66.2% (63.2–69.1%) over 1,000 deals. Same 52 cards, same layout — only the draw rule changed.

Turn 3 rewards memory and counting. Every time you cycle the stock, the same cards come up in the same groups of three, so a card you pass now will return at a predictable point. A player who tracks that wins games the greedy approach never sees. Prefer the variety? Solitaire Draw 3 is right here too.

5. Yukon: no stock, no excuses

Yukon gives you every card in the tableau and lets you move groups regardless of order. Our program won 4.5% of 10,000 deals; the solver won 67.0% of 500 deals (62.8–71.0%). Since there is no stock to blame, the entire gap is decision quality. Digging out one specific face-down card, in the right order, is the whole game.

6. Alaska: the widest gap we found

Alaska is Yukon with same-suit building that can run up or down. Our program won 1.5% of 10,000 deals. The solver won 75.6% of 500 deals (71.6–79.2%). Alaska is arguably easier than Yukon when played well, and nearly hopeless when played fast. If you want proof that solitaire is a skill game, this row is it.

7 and 8. Spider and FreeCell: we don't have the pair yet

We play and love Spider Solitaire and FreeCell, but our September 2026 runs did not produce a matched program-and-solver pair on the same deals for either game, so we are not going to print numbers for them. Two-deck games are far heavier to solve. We would rather say that plainly than guess. The strategy lesson still transfers: in Spider, the first empty column is worth more than three quick pile-ups, and in FreeCell, a free cell you spend now is a move you cannot make later.

What the published research says

Our figures are ours. Two outside sources are worth knowing, and they point the same direction.

Charlie Blake and Ian Gent, writing in the Journal of Artificial Intelligence Research (vol. 85, February 2026), report the winnability of Klondike as 81.945% ± 0.084% in the "thoughtful" variant, where the player knows the rank and suit of every card. The journal notes this is a thirty-fold reduction in the confidence interval compared with the best previous result. That is the ceiling under their rule set — and it says most deals are not lost by the shuffle.

For what real people actually score, the team at SolitaireX published site data in Data Insights and Strategies to Improve Your Win Rate: in one month, 913,106 games of Classic Solitaire were played on their site, with 409,374 wins — about a 45% win rate. Put those two together and the picture is clear: roughly four out of five Klondike deals can be won, and human players collect a bit over half of them. The rest is planning.

Five habits that move you up from the floor

Before your first move, look at the whole board. Ten seconds of scanning beats ten fast clicks. Find the deepest face-down pile and ask what would uncover it.
Don't send cards to the foundation just because you can. A 5 sitting in the tableau may be the landing spot for a 4 you need next. In Klondike, this single habit is the most common difference between a 20% player and a 40% player.
Value empty columns above everything. In Yukon, Alaska, FreeCell and Spider, an empty column is a workspace. Creating one is almost always worth delaying a "nice" move.
In Turn 3, count the stock. Note where the card you need sits in the cycle, then plan the tableau move that lets you use it when it comes back around.
Use Undo as a thinking tool. Try a line, see where it dead-ends, undo, and take the better branch. Our programs never get to do that — you do. That is exactly the advantage the floor numbers leave on the table.

Common mistakes we see (and make)

Emptying the stock early "to see what's there" in Pyramid, then having no passes left when it matters.
Taking the first matching card in Golf instead of the one that keeps a longer run going.
Moving a King into an empty column out of habit, when a smaller card would have unlocked more.
Playing when tired and calling the result bad luck. Our Alaska floor of 1.5% is what pure reflex looks like.

How to play these games here

Everything on spidersolitaireonline.us is completely free, with no ads of any kind, no pop-ups and no paywalls. There is nothing to download or install, no sign-up, no account and no email address. We built the site simply for enjoyment: no gambling, no in-app purchases, and no timer pushing you to hurry — which matters a lot when the whole point is to slow down and plan.

It runs in any modern browser on desktop, tablet and phone. The playing cards are large and clear and the text is easy to read, which helps older players and anyone with tired eyes, and there is a zoom control to make the board bigger. Every feature is available to everyone: Hint, unlimited Undo, Restart, Auto Play to finish a solved game, and your choice of drag and drop or tap to move cards. Klondike offers both Turn 1 and Turn 3. The game tracks wins, time and moves for your current game, sound can be switched off, and you can go fullscreen or change the table color.

Browse the full list on the games page, or read more strategy on the blog.

Frequently asked questions

Is solitaire luck or skill?

Both, but skill dominates. The shuffle sets a ceiling — Blake and Gent report 81.945% ± 0.084% for thoughtful Klondike — and your planning decides how close you get to it. Our own runs show a program that never plans winning 1.5% of Alaska deals where a solver wins 75.6% (71.6–79.2%).

How do I get better at solitaire fastest?

Pick one game and stay with it for a few weeks. Start with Klondike Turn 1, where our floor-to-ceiling gap was narrowest (42.3% vs. 76.6%, 73.9–79.1%), so improvement shows up quickly. Then use Undo deliberately to explore alternate lines before committing.

Do I have to sign up or download anything?

No. Click a game and play. No download, no install, no account, no email — and it is free, with no ads.

Is using Hint or Undo cheating?

Not here. Every feature is free for everyone, and unlimited Undo is how you learn to see two or three moves ahead. Think of it as practice, not a shortcut.

Which game is hardest?

Of the games we measured, Klondike Turn 3 had the lowest ceiling (66.2%, 63.2–69.1%), and Alaska had the widest gap between careless and careful play. Spider with four suits is widely considered the toughest two-deck game, but we do not have matched figures for it, so we are not printing a number.

Why can a solver win so many more games than I do?

Because it sees the face-down cards and can search thousands of lines. Treat solver figures as a ceiling to aim at, never as a fair score to compare yourself against.

What Thinking Ahead Is Worth: Greedy Play vs. Planning in 8 Solitaire Games