What the first cards tell you
Before the first move, a solitaire deal has already shown part of its hand. We grouped the deals our solver played by what they held at the start and compared how each group did. In some games the first cards say a lot; in others, almost nothing. Here are the findings side by side, and the reading habits that follow from them.
The numbers come from the tables on each game page. Most are the share of a group of deals that our solver won. FreeCell counts how often a deal was hard for the solver, meaning it looked at more than about 2,000 positions to find the win; Eight Off and TriPeaks count the deals nobody can win.
| Game, sign and group | Result |
|---|---|
| Klondike Aces buried at the bottom of columns four to seven, which you cannot see; share won (first 5,000 deals) | |
| None (2,483 deals) | 86.9% |
| Two (479 deals) | 80.8% |
| Klondike Turn 3 Aces face up; share won (first 20,000 deals) | |
| Three Aces (111 deals) | 87.4% |
| No Ace (11,051 deals) | 71.2% |
| Spider Kings in the stock, which you cannot see; share won (all 10,000 deals) | |
| Two or fewer (1,543 deals) | 97.0% |
| Six or more (1,101 deals) | 91.0% |
| Spider 2 Suits Kings in the stock, also unseen; share won (all 10,000 deals) | |
| Two or fewer (1,543 deals) | 91.1% |
| Six or more (1,101 deals) | 77.2% |
| Spider 4 Suits Face-up cards that can go on their own suit; share won (all 5,000 deals) | |
| Three or more (919 deals) | 51.0% |
| None (938 deals) | 41.5% |
| FreeCell Cards lying on the four Aces; share hard for the solver (deals 1 to 3,000) | |
| 8 or fewer (766 deals) | 6.9% |
| 17 or more (225 deals) | 23.6% |
| Baker's Game Cards on top of the four Aces; share won (all 100,000 deals) | |
| 8 or fewer (25,029 deals) | 86.7% |
| 15 or more (18,214 deals) | 60.4% |
| Seahaven Towers The two cards dealt into the cells; share won (all 20,000 deals) | |
| Both can leave at once (1,599 deals) | 98.2% |
| Neither can (10,399 deals) | 82.3% |
| Eight Off Aces at the bottom of a column; deals that cannot be won (all 50,000) | |
| No Ace there | 10 of 24,989 |
| Three or four Aces | 9 of 481 |
| Golf Aces, Queens and Kings in the columns; share that can be won (all 20,000 deals) | |
| 6 or fewer (2,701 deals) | 47.5% |
| 10 or more (3,261 deals) | 10.2% |
| Pyramid Pairs that lie in one line; share won (all 20,000 deals) | |
| At most 9 (1,957 deals) | 94.0% |
| 19 or more (675 deals) | 11.6% |
| TriPeaks Face-up pairs one rank apart; share that cannot be won (all 100,000 deals) | |
| Eight or more (40,866 deals) | 1.6% |
| Three or fewer (4,363 deals) | 10.9% |
Why the opening matters more in some games
Klondike: little turning one, a lot turning three
In Klondike turning one card, the seven face-up cards predict a little: 83.6% of deals with no Ace showing were won, 89.1% with two. Every stock card can be reached on every pass, so a poor start can usually be worked around, and a row all of one colour, where the first moves have nowhere to go, costs only about three points.
Turning three cards, two stock cards in three are out of reach on each pass, and the same seven cards tell you far more. Each Ace showing adds about five points. Three Kings face up are the worst sign: 59.3% won, against 73.6% with none, since Kings block their columns and each needs a space of its own.
Spider: the Kings you cannot see
With one or two suits, the face-up row in Spider says little; Aces showing move the result by about a point. What counts is how many of the eight Kings are still in the stock. A King that arrives with a deal covers a column until an empty column takes it or a full run built on it leaves. Between few Kings in the stock and many, the gap is six points with one suit and about fourteen with two.
With four suits the face-up cards finally matter, because their moves are the cheap ones, made before the first deal. Face-up Aces, which feel like good news, make no difference: an Ace is the last card of a run.
The FreeCell family: buried Aces and ready moves
In FreeCell, Baker's Game, Seahaven Towers and Eight Off every card is face up, and the sign that keeps coming back is how deep the Aces lie. In Baker's Game it is worth 26 points. In Eight Off, three or four Aces at the bottom of columns make a deal nearly fifty times as likely to be impossible as none, though 98% of those deals can still be won.
The second sign is moves ready at the start. In Baker's Game, deals where three top cards fit on another top card were won 89.5% of the time, deals where none did 66.3%. In Seahaven Towers the clearest sign is the two cards dealt into the cells, and in more than half the deals neither can leave at once.
Golf: Aces, Queens and Kings in the columns
In Golf nothing goes on a King and Aces and Kings do not join up, so an Ace goes only on a 2, a King only on a Queen and a Queen only on a Jack. Every other card has two places. Each of these three ranks dealt into the columns rather than the stock costs roughly eight points, and with all four Kings there, 15.8% of deals can be won.
Pyramid: pairs in one line, and the top card
In Pyramid every card except a King needs a partner that makes 13. Two partners where one lies in the triangle below the other can never go together, and these pairs in one line make the widest gap on this page; a typical deal has 13. The top card matters too. It goes last, so unless it is a King it needs a partner from the stock: with only one there, 60.3% of deals were won; with two or three, 75.2%.
TriPeaks: how the ten face-up cards link
In TriPeaks Aces and Kings join up, so every rank has two neighbours, and only 3.0% of deals cannot be won. What tells them apart is how well the ten face-up cards link: count the pairs one rank apart, such as a 6 and a 7 or a King and an Ace. The rank of the first pile card makes no difference.
The common threads
Different as the games are, four things keep turning up.
- Buried Aces. An Ace under many cards holds back its whole suit. It shows in all four FreeCell games, and in Klondike, where two Aces at the bottom of columns four to seven cost about six points turning one card and eight turning three.
- Blockers with one home. Building by suit gives a card a single place to go. In Baker's Game a 9 of clubs lying over the 4 of clubs must move first, and only the 10 of clubs takes it; the average deal has about 17 such pairs. Golf's Aces, Queens and Kings play the same part.
- Cards with no partner. In Pyramid, 15% of deals hold a card with no possible partner and are lost before the first move. In TriPeaks, when two face-down peak tops share a rank, the end of the game needs the same neighbours twice, and the share that cannot be won rises from 2.5% to 5.4%.
- Free opening moves. Ready builds, cards that go straight up and cards that fit the first pile card helped in every game where we counted them.
Kings matter less than people expect where an empty column takes only them. In Seahaven Towers, deals with all four Kings on top of other cards were won 88.7% of the time, deals with one or none there 89.0%.
What you cannot see
Klondike starts with 21 cards face down, Spider with 44 and TriPeaks with 18, and in those three, Golf and Pyramid the order of the stock stays unknown until you turn it. That limits any reading. The strongest sign in one- and two-suit Spider, the Kings in the stock, cannot be counted at the start, and Klondike's buried Aces lie under face-down cards. The solver that checked each deal saw every card.
It has limits of its own. In Klondike and Spider it gives up after a set number of positions, and giving up proves nothing, so there "won by the solver" is not the same as "winnable". It won 85.2% of Klondike deals turning one card and 74.1% turning three, where a published study with a longer search found 90.5% and 81.9%. With four suits it won 46.5% of Spider deals, while research suggests almost every one can be won. In those games a weak group may hold deals that are hard to search, not only hard to win. Golf, TriPeaks, Baker's Game and Eight Off have exact lists, so their numbers describe the deals themselves.
These are also correlations, not causes. In two-suit Spider a face-up King is worth about four points, partly because it is one King fewer waiting in the stock. And group sizes matter: between groups of a few hundred deals, a gap of a point or two can be chance alone. The worst FreeCell group holds 225 deals; the 26 points in Baker's Game rest on more than 43,000.
How to read a deal in 30 seconds
Every game dealt here has a known way through, so a bad sign means a narrower path, not a lost game, and no reason to ask for another. Reading the deal tells you where the work will be.
Klondike
- Turning one, glance and play: note the Aces and whether the row is all one colour.
- Turning three, count the Aces and Kings. Few Aces means a slow stock; three Kings means you need spaces early.
- Either way, go through the stock once before you commit, and dig into columns five to seven.
Spider
- With one or two suits, count the Kings you can see: each is one that will not fall out of the stock. After every deal, work towards an empty column again.
- With four suits, count the face-up cards that can move at the start, and how many of those moves are in suit.
FreeCell, Baker's Game, Seahaven Towers and Eight Off
- Count the cards lying on each Ace: that is your digging order.
- Count the top cards that fit on another top card: each is a move ready to make.
- Building by suit, find each high card lying over a low one of its suit, and where its one home is.
- In Seahaven Towers, check whether the cell cards can leave at once; in Eight Off, look for Aces at the bottom.
Golf, Pyramid and TriPeaks
- Golf: count the Aces, Queens and Kings in the columns; eight is the most common number. Then trace the longest run from the first pile card.
- Pyramid: count the top card's partners in the pyramid, then look for partners lying in one line.
- TriPeaks: count the face-up pairs one rank apart. Three or fewer means the stock will do more of the work.
Then play. Undo and Hint cover the guesses that hidden cards force on you, and how many games can be won has the overall numbers for every game. Before you make that first move, which first moves still win shows the games where that move alone can lose the deal.
Sources
The layout figures are our own, from the solver runs described on each game page. The published Klondike and Spider figures come from Charlie Blake and Ian Gent, The Winnability of Klondike Solitaire and Many Other Patience Games.