Spider Solitaire, one suit: how much of the game is really in your hands?

One-suit Spider is where most people meet the game. It uses 104 cards, all of them spades, spread over ten columns. Your job is to build eight complete runs from King down to Ace; each finished run lifts off the table on its own.

This site's research is about ceilings: how many deals could be won with perfect play. For Spider we have to be upfront. Spider's search space is enormous, and we did not build a solver that could settle it. What we can offer is a published ceiling for the hardest version, and our own measured floor for this one. Between the two, the picture is clear enough to be useful.

The floor: a plain computer player that sees only face-up cards and never takes a move back won 52.8% of 4,000 one-suit deals. The published ceiling for four-suit Spider, with every card visible, is about 98.5%. One suit is the most forgiving version, so there is a lot of room between what a simple player achieves and what is possible.

Floor and ceiling side by side

The published figure is from Blake & Gent (2026), Journal of Artificial Intelligence Research. The player tier is from the solitairex.io homepage, which groups games by the win rates its players reach but does not publish its method.

What sits between 52.8% and the top

Our computer player is deliberately simple. It never uses Undo, it cannot see what is under a face-down card, and it scores positions with a handful of fixed rules. Even so, it won more than half its games and completed nearly five of the eight suits on average.

A person who plans ahead and uses Undo does better, and real players on solitairex.io sit in the 70–84% band for one-suit Spider. The published study shows that even four-suit deals are almost all winnable when every card is known. So when you lose a one-suit game, the deal was very probably not the problem.

How this site deals Spider

• 104 cards, all one suit, in ten columns: four of six cards and six of five. The top card of each is face up.
• The remaining 50 cards stay in the stock. Each click deals one card onto every column.
• Any card can go on a card one rank higher.
• A run can be moved together only when it is all one suit, which in this version is always.
• Any card may go into an empty column.
• Finished King-to-Ace runs clear themselves off the table.
• You get a warning if any column is empty when you deal.

Closing the gap

Empty a column before you deal, if you can

Our player managed to clear a column before its first deal in only 1.3% of games, but when it did, it won 51 of those 52 games. Emptying a column early is rare, and powerful.

Uncover face-down cards first

Every hidden card is information the player lacks. When two moves are equal, take the one that flips a card.

Build long runs on high cards

A run starting from a King is safe; a run starting from a 5 will need moving again. Put your sequences where they can grow.

Deal only when you've run out of useful moves

Each deal buries every column under one more card. Before you click, check that no move would uncover a card or tidy a run.

Use Undo as a look-ahead

The ceiling figures come from solvers that know everything. Undo lets you test a line and step back, which is the closest a person can get.

One-suit Spider on the floor ladder

Since we have no Spider ceiling of our own, it helps to compare floors. With the same idea behind each player, seeing only face-up cards and never undoing, one-suit Spider comes out on top:

• Spider, one suit: 52.8%.
• Diplomat: 47.9%.
• Klondike Turn 1: 42.3%.
• Klondike Turn 3: 10.6%.

A high floor like this usually signals a forgiving game. In one-suit Spider, every ordered run can be moved as a block, and finished suits leave the table on their own, so mistakes are easier to recover from than in most solitaires.

Inside the test, one step at a time

The solitaire-online.us team ran this study in September 2026. For Spider, our aim was to measure the floor on this site's rules and to set it against the best published ceiling.

1. Reading the rules from our game code. Ten columns, 50 cards in the stock dealt ten at a time, same-suit runs, any card into an empty column, and automatic removal of finished runs, all from this site's game files. One rule we assumed: that you cannot deal while a column is empty. The site warns you, but we did not test it in a browser.
2. Rebuilding the game. We wrote Spider as a program with those rules, using spades for the single suit.
3. Dealing the cards. Seeds 100,001 to 104,000. We tuned the computer player on seeds 1 to 30 and then measured it on seeds it had never seen.
4. Choosing the player. A look-ahead player that searches 250 positions each turn using only the visible cards. Because it can never take a move back, what it wins marks the bottom of the range, not the top. We did not build a Spider solver, so we have no ceiling of our own.
5. Running it. 4,000 deals, each played to a win or until the player had nothing useful left.
6. Checking against published research. There is no published figure for one-suit Spider on these rules. Blake & Gent (2026) give 98.487% for four suits with every card visible, which we use as the reference ceiling.
7. Working out the margin of error. A 95% range is the band where the true value would land 19 times in 20. For our player's win rate it is 51.2–54.3%.
8. What the numbers cannot tell you. Our figure is what a simple, never-undoing player does. It is not how winnable the deals are, and it is not what a careful person achieves.

The code behind the ceiling

All the code is Python 3.11 using the standard library only. Spider has its own module, spider.py, and runner.py spreads the 4,000 deals across 19 worker processes on a 20-core machine. Each deal is rebuilt from a numbered seed.

On each turn the player runs a best-first search of up to 250 positions over the cards it can see, treating any newly uncovered card as unknown. It scores each position with fixed points: a bonus for every same-suit link and every empty column, a smaller bonus for any-suit links and freshly turned cards, and a penalty for every card still face down. It then plays toward the best-scoring position, and deals only when nothing improves the score. The one-suit run took about 21 minutes. Ranges use the Wilson score interval.

We used an AI assistant, Claude from Anthropic, to help write and check this code and to draft the page. The AI did not produce any figure here; every number comes from running the program or from the published sources linked above.

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

What share of one-suit deals is winnable?

Nobody has published a figure for one suit on these rules, and we did not measure it. The published study found about 98.5% of four-suit deals winnable with every card visible, and one suit is the easiest version.

Why did your computer player win only about half?

Because it is simple on purpose: it never undoes and cannot see hidden cards. It is a floor, not a ceiling.

How many moves did the one-suit wins take?

Our player's wins averaged about 118 moves, counting each deal from the stock as a move.

Is it time to try two suits?

When you win most one-suit games. The step is bigger than it looks; see Spider 2 Suits and Spider 4 Suits.

What other games have a high ceiling?

FreeCell had no provably impossible deal in our tests. See all games on the games page.