Four-suit Spider: nearly every deal can be won, almost none by reacting
Four-suit Spider is the purest test of planning on this site. Two full decks, every suit in play, and eight single-suit runs to complete. The cards are almost never the problem. The plan is.
Two numbers make the point. A published solver that sees every card found 98.5% of four-suit deals winnable (±1.5%). Our look-ahead player, which sees only face-up cards and never undoes, won 0.2% of 2,000 deals (95% range 0.1–0.5%), just 4 games. Everything between those two figures is what better strategy, more information and Undo can buy.
This page explains that gap, the rules as played on spidersolitaireonline.us, and what careful players do differently.
At a glance: the gap planning has to close
| Measure | Result | 95% range | Games |
|---|---|---|---|
| Deals winnable with every card known (Blake and Gent, 2026) | 98.5% | ±1.5% | published |
| Our look-ahead player's win rate | 0.2% | 0.1–0.5% | 2,000 |
| Win rate after an empty column before the first deal | 33.3% | 9.7–70.0% | 6 |
| Games where that early empty column happened | 0.3% | 0.1–0.7% | 2,000 |
| Runs completed per game, of 8 | 0.05 | 0.03–0.06 | 2,000 |
| Games with at least 4 runs completed | 0.2% | 0.1–0.5% | 2,000 |
We did not run a greedy Spider player or our own four-suit solver, so the closest pair we have is our look-ahead floor against the published full-information ceiling. The early-empty-column figure again points toward space as the prize, but with just 6 such games out of 2,000 it tells you little on its own. With one and two suits, the same player won 98.1% and 71.4% of games after an early empty column.
Why is the floor so low? Our player searches up to 250 positions ahead, but only through the cards it can see, and in four-suit Spider most cards start face down while almost every build is mixed-suit. It completed one run in about every twenty games.
The rules as played on spidersolitaireonline.us
- Cards: two decks, all four suits, 104 cards.
- Layout: ten columns, the first four with six cards and the rest with five. Only the bottom card is face up.
- Stock: 50 cards, dealt ten at a time, one on each column. Five deals, no reuse.
- Building: any card on a card one rank higher, of any suit.
- Moving runs: only a single-suit run moves as a group.
- Empty columns: accept any card or movable run.
- Completing runs: a King-to-Ace run in one suit is removed automatically. Eight runs wins.
The game warns you if you try to deal while a column is empty.
How careful players close the gap
Think of each deal as a puzzle, not a race
Our player's four wins averaged 209.3 moves (193.9–224.6), far more than one-suit wins at 117.5, though four games make a shaky average. Four-suit Spider rewards long, slow thinking.
Uncover before you build
The solver behind the 98.5% knows where every card is; you do not. The nearest you can get is to turn cards over. Prefer moves that reveal a hidden card over moves that merely tidy a run.
Every mixed build is a debt
Mixed builds are legal, but they cannot move as a unit. When you must make one, use high cards that will not bury the low cards you need soon, and know how you will pay it back.
Protect empty columns
A space lets you break a mixed run apart and rebuild it in suit. Use it, then clear it again, rather than parking a card there for good.
Undo is your window onto hidden cards
Try a move, see what turns up, and take it back if needed. Our player never did this, which is a large part of why it won so rarely.
Playing here
Four-suit Spider on spidersolitaireonline.us is free in any modern browser on computer, tablet or phone, with no download, no account and no email. Large cards with zoom, Hint, unlimited Undo, Restart, Auto Play, drag-and-drop or tap controls, a counter for wins, time and moves, sound on or off, fullscreen and a choice of table colors are all included.
How we ran the study
The spidersolitaireonline.us team ran this simulation in September 2026 on this site's own Spider rules.
- Read the rules from our game's code, as listed above. We assumed, without confirming in a browser, that you cannot deal while a column is empty.
- Wrote the game in Python 3.11, standard library only (
spider.py), one program for one, two and four suits. - Tuned on deals 1 to 30, measured on deals 100,001 to 102,000, each shuffled from a numbered seed.
- Used a player that sees only face-up cards, searches up to 250 positions per turn, treats hidden cards as unknown and never undoes. Its result is a floor.
- Played 2,000 deals, half as many as for one and two suits, because the result was already clear.
- Took the ceiling from published research: Blake and Gent, Journal of Artificial Intelligence Research, 2026, 98.5% ±1.5% for a solver that knows every card. That measures something different from our floor, so neither confirms the other.
- Added 95% ranges (Wilson score intervals for rates, normal intervals for averages).
- Named the limits. With 4 wins and 6 early-empty-column games, those sub-figures are rough. The 0.2% says nothing about how often skilled people win.
We used Claude, an AI assistant from Anthropic, to help write and review the code and draft this page. Every figure is the output of running the code or comes from the cited study, and the team checked them before publishing.
Questions
Can four-suit deals really be won?
Almost always in principle: 98.5% with every card known. In practice it takes planning, patience and Undo.
Why did your program do so badly?
It looks only a short way ahead, cannot see face-down cards and never undoes. Four suits punish all three.
Should I start here?
Probably not. Learn to create space on one suit, then learn to build in suit on two suits.
Is there a four-suit game with nothing hidden?
Simple Simon uses Spider-style rules with one deck and every card face up. FreeCell is another open-card game. More on the games page and the blog.
By Admin, spidersolitaireonline.us (Solitairen OOD). Research run September 2026.