Spider Solitaire with one suit: the version the research skipped

Spider uses two full packs, 104 cards, spread across ten columns. You build long runs from King down to Ace, and every complete run leaves the table. In the one-suit version played on this page, every card is a spade, so any card can go on any card one rank higher and every run is automatically in suit.

Most published work on Spider looks at the full four-suit game. The St Andrews study by Blake and Gent, for instance, reports that 98.487% of four-suit deals can be won by a solver that sees every card. Nobody, as far as we could find, has published a comparable figure for one suit. So for this page we did the next best thing: we measured how a careful computer player does on this site's one-suit Spider, and set that beside what real players report elsewhere.

The result: our player, which sees only face-up cards and never undoes a move, won 52.8% of 4,000 deals.

What exists to compare with

Read together, these point one way. If a perfect player can win almost every four-suit deal, one-suit deals, where far more moves are allowed, are unlikely to be the obstacle. Our simple player won just over half. solitairex.io puts games in the same group at 70% to 84%, though without saying how that was measured. The gap between a no-undo computer player and the best-case figure is where planning and Undo come in.

The rules on this site

• 104 cards dealt into ten columns: the first four hold six cards, the other six hold five. Each column shows only its last card.
• The remaining 50 cards form the stock, dealt ten at a time, one onto each column. There are five such deals and no redeal.
• Any card may go on a card one rank higher.
• A run of cards may be moved together if it is in sequence and in suit; with one suit, every sequence qualifies.
• Any card or run may go into an empty column.
• A complete run from King to Ace is removed automatically. Remove all eight to win.
• If a column is empty when you try to deal, the site shows a warning.

What the numbers suggest for your game

An early empty column nearly settles it

Our player emptied a column before the first stock deal in only 1.3% of games, but when it did, it won 51 of those 52 games. An empty column is the most valuable thing on the Spider table. Work towards one from the start.

Turn over hidden cards before tidying runs

Our player's scoring strongly penalised face-down cards. Uncovering them early is what turns a tangle into a plan, because you cannot use cards you cannot see.

Plan for about 120 moves

Games our player won took 117.5 moves on average, including the five stock deals. A win is a long, steady clear-up rather than a sprint.

Deal only when you have run out of useful moves

Each stock deal drops a card on every column, burying whatever you were working on. Our player filled empty columns and made every improving move before dealing.

Undo is where you beat the bot

Our player never takes a move back. You can. If a line leads nowhere, step back and try the other build. That alone should lift you well above 52.8%.

Spider on solitaire-free.uk

Play free in any modern browser on computer, tablet or phone; nothing to install, no sign-up and no email. You get large cards and a zoom control, Hint, unlimited Undo, Restart, Auto Play, drag and drop or tap to move, and a tally of wins, time and moves for the current game. Sound, fullscreen and table colour are adjustable.

Ready for more? Try Spider with two suits or four suits, or compare with FreeCell.

One-suit mistakes worth avoiding

• Leaving short columns alone. A column of a few cards is your best chance of an empty column. Clear it.
• Dealing with a column still empty. The game warns you for good reason. Put something in the space first, ideally a long run.
• Building runs that start low. A run headed by a 5 has nowhere to go but onto a 6. Runs headed by Kings are the ones that can finish.

How we ran the research, step by step

This study was carried out by the solitaire-free.uk team in September 2026. Because there is no published one-suit figure to check against, we paid particular attention to making the method transparent.

1. Rules from our code. 104 cards, ten columns of six and five, a 50-card stock dealt one card per column, same-suit runs move together, any card into an empty column, completed suits removed. One rule we had to assume: dealing is not permitted while any column is empty. The site shows a warning, but we did not test its exact behaviour.
2. A program version. We rebuilt Spider in Python and wrote a computer player for it.
3. Fair deals. We tuned the player's scoring on a handful of deals first (seeds 1 to 30), then measured it on 4,000 completely different deals, seeds 100,001 to 104,000, so the tuning could not flatter the result.
4. The player. At each turn it looks ahead through up to 250 positions using only the face-up cards, treating any card it would uncover as unknown. It never undoes a move. That makes its score a floor for a patient human, not a ceiling.
5. Running it. 4,000 deals at one suit.
6. Checking against published work. There is no published one-suit result. The nearest, the St Andrews four-suit figure of 98.487%, tells us deals are rarely the limiting factor in Spider.
7. The margin of error. A 95% range is where the true figure very probably lies. For our player, 52.8% means between 51.2% and 54.3%.
8. What the figure can't tell you. It is not how many one-suit deals can be won; it is how often a no-undo, no-peeking player wins. A strong human should do better.

Behind the numbers: the code

We built the simulation in Python 3.11 without any add-on libraries. Spider has its own module, spider.py, which covers one, two and four suits, and a shared runner spread the deals across 19 worker processes on a 20-core machine. The player is a best-first look-ahead: from the current position it explores up to 250 possible continuations using only visible cards, scores each one (rewarding same-suit links, empty columns and newly uncovered cards, penalising face-down cards), and plays the sequence that reaches the best score. For one suit we used spades throughout. The 4,000 games took about twenty-one minutes. Percentage ranges use the Wilson score interval; averages use a normal 95% interval.

We used an AI assistant, Claude by Anthropic, to help write and review the simulation and to draft this page. No figure on this page came from the AI. They were all produced by running the code, and we set them against the published research before publishing.

Frequently asked questions

Can every one-suit Spider game be won?

We don't know of a published figure. What we know is that a solver with full information wins about 98.5% of the much harder four-suit game, which suggests very few one-suit deals are impossible.

Why did your computer only win about half?

Because it cannot see face-down cards and never undoes. It is deliberately a floor, not a model of a skilled player.

Is one suit a good way to learn Spider?

Yes. It teaches the value of empty columns and uncovering cards, which matter even more when you add suits.

How many moves does a win need?

Our player's wins averaged about 118 moves. Most people will take somewhat more, because they explore with Undo.

Which game should I try after one suit?

Two suits is the natural next step, and our data shows it is a big jump. The blog has more on how the three versions compare.

By Admin, solitaire-free.uk. Study date: 18 September 2026.