Solitaire Turn 3: fewer wins, more laps of the stock

Turn 3 is Klondike with the stock turned three cards at a time. Only the top card of each group of three can be played, so a card you need can sit hidden in the middle of a group for lap after lap. The layout, the columns and the goal are the same as the classic game. What changes is how much of the stock you can actually use.

To measure that, the solitaire-online.nz team gave 10,000 games of Turn 3 to the same computer player we used for Turn 1. It won 10.6% of them. When it did win, it needed 3.8 trips through the stock on average, against 2.3 at Turn 1, and 111.1 moves.

That last number surprises people. Winning games at Turn 3 were slightly shorter than at Turn 1, not longer. The reason is which games get won: at Turn 3 only the friendlier deals come out, and friendly deals finish quickly.

Turn 3 at a glance

Put simply: expect to go round the stock three or four times in a game you win, and expect to lose more often than you win. Turn 3 suits a longer sitting, when you have time to go back with Undo and try a different order.

The rules at Turn 3

• Deal. Seven columns holding 1 to 7 cards, the bottom card of each face up, and 24 cards in the stock.
• Stock. Each tap turns three cards onto the waste. You may only use the card showing on top. Once it is played, the one beneath becomes available.
• Redeals. When the stock is empty, flip the waste back and start again. There is no limit on how many times.
• Columns. Build down in alternating colours. Move any face-up run whose top card fits. When a column is cleared, only a King can go there.
• Foundations. Build each suit up from Ace to King. One card may come back down to the columns if you need it.

Why the laps add up

At Turn 1 every card in the stock comes up once per lap. At Turn 3 you only see every third card, and which cards those are changes only when you play one from the waste. Play nothing, and the next lap shows you exactly the same cards. That is why our player needed more laps, and why so many of its games stalled.

Every card you play from the waste shifts the groups of three that follow it. One well-chosen play from the waste can open up a card you have been waiting on for three laps.

Getting more from each lap

Count the laps

If a lap brings nothing new, the next one will not either unless something changes. Before you turn the waste over again, check the columns for a move that plays a card from the waste or frees one up.

Play from the waste when it helps, not always

Taking the top waste card changes which cards come up next time. Sometimes that is exactly what you need. Sometimes it buries a card you were about to reach. Undo lets you try both.

Uncover hidden cards before anything else

Our player's first priority after foundation moves was turning over face-down cards. The more of the columns you can see, the more places a card from the stock can go.

Do not rush cards up to the foundations

A 4 sent up too early might have been the only place for a 3 of the other colour. At Turn 3 you cannot count on the stock to rescue you, so keep a few low cards in play until they are clearly spare.

Turn 3 on any screen

The game works in any modern browser on a phone, tablet or computer. There is no download, no account and no email to hand over. Tap or drag to move cards, zoom in if they look small, and use Hint, Undo and Restart as often as you like. Auto Play sends the last cards home, sound can be muted, and there is a fullscreen mode and a choice of table colours. The counter shows your moves and time for the current game.

What a Turn 3 session looks like

A Turn 3 game usually splits into two halves. In the first, you are clearing the columns and learning which stock cards you can reach. In the second, you are going round the stock again and again, trying to shift the groups of three so that the card you need finally comes up on top.

That second half is where Turn 3 takes its time. Our computer player averaged 3.8 laps in its wins, and in the games it lost it typically went round until nothing changed at all. For a person, the practical rule is to pause at the end of each lap and ask what will be different next time. If the answer is nothing, change something in the columns first.

Because only about one game in ten came out for our player, many people keep Turn 3 for when they have time to experiment with Undo. On a quick break, Turn 1 is the more forgiving choice.

How we did the research, step by step

The solitaire-online.nz team ran this in September 2026 to find out what a Turn 3 game on this site looks like in practice: how many laps, how many moves, and how often a careful player wins.

1. We took the rules from this site's game code. Three cards per turn, only the top one playable, unlimited redeals, alternating colours in the columns, Kings only into empty columns, one card back from a foundation allowed.
2. We rebuilt Turn 3 as a program and checked 1,000 Turn 3 games (and 1,000 Turn 1 games) move by move to make sure every move was legal.
3. We dealt numbered games. Every shuffle comes from a numbered seed, so anyone with the same program gets the same 10,000 deals.
4. We used two players. A computer player that sees only what you see and never uses Undo, which shows what steady play achieves. And a solver that sees every hidden card, which shows how many deals could be won at all.
5. We ran it. The computer player played 10,000 deals. The solver worked through 1,000 deals with a limit of 150,000 positions each.
6. We compared with published work. The solver proved 66.2% of deals winnable and 9.5% impossible, with 24.3% undecided, so between 66.2% and 90.5% can be won. The published figure of 81.945% from Blake & Gent, 2026 falls inside that.
7. We put error bars on it. A 95% range is where the answer would land 19 times out of 20 if we repeated the test with fresh deals. For laps through the stock it is 3.7 to 3.9.
8. We noted what it cannot tell you. The move and lap figures only come from won games, which were the easier deals. A person with Undo will win more often than our player and may take more laps doing it.

The technical side

The code is Python 3.11 with only the standard library. Turn 3 lives in the same klondike.py module as Turn 1. The computer player works through a fixed list of preferences and only draws from the stock when nothing better is on offer, and it gives up after a full lap with no progress. For the solver, Turn 3 needed extra care: it models exactly which stock cards can be reached given the groups of three, rather than treating the whole stock as available. A shared runner.py spread the games across 19 worker processes. The 10,000 computer games took about 8 seconds, the 1,000 solver games about 17 minutes.

The error bars are Wilson score intervals. An AI assistant, Claude by Anthropic, helped us write and review the code and draft this page, but it did not supply any of the figures. They all come from running the program, and we checked them against the published study before publishing.

By Admin, solitaire-online.nz. Study date: 18 September 2026.

Short answers before your next deal

How many times should I go through the stock?

In our player's winning games it averaged 3.8 laps. If you are well past that with no progress, look for a change in the columns rather than another lap.

Why are won games shorter at Turn 3?

Because only the easier deals get won, and easy deals finish quickly. The average across winning games at Turn 3 was 111.1 moves, against 126.1 at Turn 1.

Are some Turn 3 games impossible?

Yes. Even seeing every card, our solver proved 9.5% of deals impossible, against 2.1% at Turn 1.

Should I switch back to Turn 1?

Both are real, and both are played widely. If you want the version you are more likely to finish, try Turn 1.

What else suits a longer session?

FreeCell rewards careful thinking, and Diplomat is a long two-deck game. Browse them all on the games page.