Spider Solitaire 4 Suits: nearly always winnable, rarely won

By Admin, free-solitaire.nz · Figures measured 18 September 2026

Four-suit Spider is the full game: two packs, all four suits, 104 cards. It is the version serious patience players measure themselves against, and it has earned its hard reputation. What surprises most people is that the deals themselves are hardly ever the problem.

Three numbers tell the story, and together they explain this game better than any rulebook:

Almost every deal can be won. Almost nobody wins it by accident. The gap between those two facts is where the skill lives, and this page is about closing it.

The rules

The layout

• 104 cards: two complete packs, so every card appears twice.
• Ten columns dealt 6-6-6-6-5-5-5-5-5-5, with only the bottom card of each turned up.
• A 50-card stock. Each deal puts one card face up on every column, five deals in total, with no redeal.

How cards move

• Any card goes on a card one rank higher, of any suit.
• Only a run that is in order and all one suit can move as a block.
• An empty column takes any card or same-suit run.
• A complete King-to-Ace run in one suit leaves the table by itself. Clear all eight to win.

The game warns you if you try to deal from the stock while a column is empty.

What our simulation found, and what it does not mean

We played 2,000 four-suit deals with a computer player that looks ahead before each move but sees only face-up cards and never takes a move back.

• It won 4 games out of 2,000, or 0.2% (95% range 0.08–0.51%).
• It cleared an average of 0.05 suits per game. In the great majority of games it never completed a single run.
• It emptied a column before the first deal in only 6 games (0.3%), and won 2 of those.

Across our three Spider tests, the same bot's results fell steeply as suits were added:

Read that 0.2% for what it is: a measure of how hard four-suit Spider is when you cannot look ahead properly or take a move back. It is not a measure of how many deals can be won. That answer, from the peer-reviewed study, is about 98.5%. Our bot is a weak player at four suits, and we say so plainly. What its failure shows is that simple, sensible-looking play gets you almost nowhere in this game. You have to plan.

Why four suits is so much harder

Every card can go on any card one rank higher, but only single-suit runs move. With four suits, the chance that the card you want to build on is also the right suit drops to one in four. Most builds early on are mixed. Each mixed build locks cards together, and with four suits you can make a lot of those before you realise the table has seized up.

The only way to pull mixed piles apart is space: an empty column, or a same-suit home for the top card. Four-suit Spider is really a game about creating space and spending it carefully.

Playing on this site

The game is free, with no download, no sign-up and no email. It works in any modern browser on a computer, tablet or phone.

• Unlimited Undo and Restart. At four suits, these are how good players win, not a crutch.
• Hint for a nudge when you cannot see a way on.
• Large, clear cards and a zoom control, which help when ten columns hold four suits.
• Auto Play, drag and drop or tap, and wins, time and moves tracked for the current game.
• Fullscreen, a choice of table colours, and sound you can switch off.

How strong players approach four suits

Play the long game from move one

Before each move, ask what it does to your chances of emptying a column, not just whether it is legal. The legal-but-aimless move is what sank our bot in 1,996 games out of 2,000.

Prefer same-suit builds, even when they are less convenient

A same-suit build keeps a pile movable. A mixed build freezes it. When you have both options, the same-suit one is nearly always right, even if the mixed one uncovers a card slightly sooner.

Uncover cards in the columns you can empty

Spread out and you will turn cards over everywhere without clearing anything. Focus on the one or two shortest columns and work them down to nothing. One empty column is worth far more than a dozen scattered face-up cards.

Use empty columns to sort, not to store

An empty column is your workbench. Use it to break mixed piles into single-suit runs, then free it again. Dropping a King there too early is the most common four-suit mistake.

Plan each deal

You get five deals from the stock, and each one covers every column. Before you deal, make every move that improves your position, and fill empty columns with cards you are happy to see buried.

Explore with Undo and Restart

The published 98.5% comes from a solver that tries many lines. You can do a human version of that. When a position has several paths, try one, see what it uncovers, and back out if it goes nowhere. If a deal is going badly, Restart and play it again knowing where the key cards are. Nobody here is keeping score but you.

How we did the research, step by step

The free-solitaire.nz team ran this study in September 2026. Here is exactly what we did, in order, so you can judge the numbers for yourself.

1. Read the rules from our own game. We took the rules from this site's game code, not from a rule book: 104 cards in four suits, ten columns dealt 6-6-6-6-5-5-5-5-5-5 with the last card face up, a 50-card stock dealt one card to each column, any card on one rank higher, only same-suit runs moving as a block, any card into an empty column, and full King-to-Ace suits removed automatically.
2. Wrote down one assumption. The game warns you when you try to deal with a column empty. We assumed the usual rule that you cannot deal onto an empty column. We did not test this in a browser.
3. Rebuilt the game in code. We wrote a small program that plays four-suit Spider by those rules and checks every move.
4. Built a player that only sees what you see. Before each move our bot looks up to 250 positions ahead using the face-up cards only. It treats any card it would uncover as unknown and never uses Undo.
5. Kept the test deals separate. We tuned the bot's scoring on Spider deals 1–30, then tested it on 2,000 different deals (numbers 100001–102000). None of the test deals played any part in the tuning.
6. Played and counted. The bot won 4 of the 2,000 deals (0.2%, 95% range 0.08–0.5%) and completed 0.05 suits per game on average.
7. Checked the ceiling. We did not build a solver that can see the hidden cards for Spider, so we cannot say from our own work how many deals are winnable. For that we use the published figure: Blake and Gent, Journal of Artificial Intelligence Research, February 2026, put four-suit Spider at 98.5% (±1.5 points) when every card is known.
8. Put it next to real players. solitairex.io places four-suit Spider in its hardest group, at 11% or less.

What this means: our 0.2% is a floor from a deliberately simple, no-Undo player, not a measure of how winnable the game is. The gap between 0.2% and 98.5% is mostly skill, planning and Undo. Ranges are 95% confidence intervals. Deals are generated from fixed seeds, so anyone can replay them.

The technical side

The simulation is Python 3.11, standard library only, in spider.py set to four suits. A shared runner.py uses Python's multiprocessing to run 19 deals at once, and each deal is shuffled from a numbered seed starting at 100001.

• The bot runs a best-first look-ahead of up to 250 positions each turn on the visible cards only, with a record of positions already explored. At four suits a look-ahead that short is simply too short-sighted, which is why its result is so low.
• Scale. The 2,000 four-suit deals took about 3½ minutes, far less than at one or two suits, because the bot usually ran out of useful moves early.
• No solver. We did not build a solver that sees every card for Spider, so the upper figure on this page is the published 98.5%, not ours.
• The 95% range is a Wilson score interval, which stays honest even for a result as close to zero as this one.

As on the other Spider pages, the bot was tuned on seeds 1 to 30 and measured on different deals. We used an AI assistant (Claude, by Anthropic) to help write and review the code and draft this page. Every number comes from running the code, not from the AI, and the team checked the results against published research.

Frequently asked questions

Can every four-suit Spider game be won?

Nearly. A peer-reviewed study found about 98.5% of deals can be won when every card is known in advance. In real play you cannot see the hidden cards, so your own rate will be much lower.

What is a good win rate at four suits?

Real-player data from solitairex.io puts four-suit Spider at 11% or less. Winning one game in ten is respectable. Players who use Undo and plan carefully can do better.

Why did your bot win only 0.2%?

It is deliberately limited: it sees only face-up cards, never takes a move back and only looks a short way ahead. Those limits hurt far more with four suits than with one. That is why we present it as a measure of difficulty, not of how many deals can be won.

Is using Undo cheating?

No. There is no leaderboard here and nothing to buy. At four suits, Undo is simply how people learn the game.

How long does a game take?

Often thirty minutes or more. It is a good one for a long, wet Waikato weekend.

What should I play to build up to it?

Start with one-suit Spider, move up to two suits, and come back here once two suits feels comfortable. For a different kind of challenge, try Yukon or FreeCell. Every game is on the games page, and more tips are on the blog.