What this study measures
This page is a data study, not a strategy guide. Everyone knows a 9 is better than a 1 on the total. What nobody seems to have counted is how much better in play, once formations, claims and the opponent get involved. To find out, the site's AI played 400 games of the base game against itself, with the same strength on both sides and a fixed seed. Every game ended with a winner.
At the end of each game we looked at every clan card on the table, 19,898 cards in all. For each one we noted whether its side had won its stone, lost it or left it undecided, which formation that side held, and which player had laid it. We also noted how early in the game each value first came down.
These are the same 400 games behind our Schotten Totten tips, our Schotten Totten stone statistics and our Schotten Totten formation statistics. Two copies of one AI are not two people. They read the board the same way and never bluff, so read the figures as what careful, consistent play produces, not as a law of human play.
How often each value wins its stone
| Value | Won | Lost | Open | Win rate | By the winner |
|---|---|---|---|---|---|
| 1 | 819 | 919 | 271 | 47.1% | 50.8% |
| 2 | 875 | 946 | 319 | 48.1% | 50.9% |
| 3 | 936 | 892 | 380 | 51.2% | 48.7% |
| 4 | 981 | 892 | 365 | 52.4% | 52.4% |
| 5 | 955 | 900 | 404 | 51.5% | 50.2% |
| 6 | 984 | 888 | 388 | 52.6% | 50.9% |
| 7 | 1,018 | 794 | 435 | 56.2% | 50.5% |
| 8 | 1,044 | 772 | 453 | 57.5% | 50.7% |
| 9 | 1,097 | 767 | 404 | 58.9% | 49.6% |
Cards on a won, lost or still open stone, by value. "Win rate" counts the cards on decided stones only: won divided by won plus lost. "By the winner" is the share of the cards the game's winner laid. Card values and formations: Schotten Totten rulebook, transcribed by RulesPal; figures: the site's game engine, base game, seed 20261011, the same AI on both sides, its search set to 200 iterations; in the base game it plays its heuristic move without running the search, so that setting has no effect.
The higher the value, the more often its stone is won, from 47% for a 1 to 59% for a 9. Each value has about 1,800 cards on decided stones, so chance moves its percentage by about two points either way, a little more since cards on one stone share its result. The gap between the ends of the scale, twelve points, is real. The order inside the middle group is not: 3, 4, 5 and 6 sit between 51% and 53%, and a rerun could reshuffle them.
The gap is also smaller than the total suggests. A 9 is worth nine times a 1 on the total, yet it wins its stone only a quarter more often. Only about one won card in ten sat in a plain sum, as the next sections show.
No value picks the winner
The last column of the table, "By the winner", is the surprise. The player who won the game laid between 49% and 52% of the cards of every value, and 50.5% of all cards. With 2,000 to 2,300 cards per value, a gap of one or two points is what chance produces. In other words, the winner did not get the 9s.
Both players draw from the same deck, so over a game they see roughly the same cards. A 9 helps the stone it lands on, but the opponent gets their own 9s too. The study does not show what separates the winner, only that it is not the number of high cards laid. Where the cards go is the likely answer, but these games do not measure it.
Which formation each value wins in
| Value | Three of a kind | Colour run | Colour | Run | Sum |
|---|---|---|---|---|---|
| 1 | 405 | 128 | 182 | 68 | 36 |
| 2 | 318 | 216 | 171 | 107 | 63 |
| 3 | 276 | 306 | 171 | 136 | 47 |
| 4 | 285 | 284 | 206 | 121 | 85 |
| 5 | 291 | 281 | 169 | 121 | 93 |
| 6 | 279 | 273 | 172 | 139 | 121 |
| 7 | 255 | 304 | 176 | 161 | 122 |
| 8 | 384 | 219 | 172 | 122 | 147 |
| 9 | 516 | 137 | 186 | 75 | 183 |
Cards on won stones, by the formation their side held. Same games and source as above.
The two ends of the scale win most often in three of a kind: 49% of the won 1s and 47% of the won 9s, against 25 to 30% for the 3s to 7s. The middle values split their wins almost evenly between three of a kind and a colour run: the colour run takes 28 to 33% of their wins, against 16% for the 1s and 12% for the 9s. That is the shape the rules predict. A 9 fits only one run, 7-8-9, and a 1 only 1-2-3, while a 5 fits three. Our Schotten Totten tips give that advice without a number, and these games bear it out.
The total is where the high cards pull ahead. 183 of the 9s on won stones sat in a plain sum, against only 36 of the 1s. A 9 on a failed plan still leaves a decent total; a 1 leaves almost nothing.
How each value loses
Every value sat in a losing sum about as often: between 303 and 367 cards of each value, with the 9s at 365. A 9 does not rescue a side that ended up as three unrelated cards. Our formation statistics show the same thing from the other end: more than half of all losing lines are sums.
The 1s have one loss of their own. 162 of the 1s on lost stones sat in three of a kind, more than any other value, because three 1s is the lowest three of a kind there is: any other three of a kind beats it, and so does a colour run. Only 51 of the 9s did.
Low cards stay in hand
The site's AI does not spend its cards in order. In the median game, the first 1 and the first 2 came down on the eighth move of the game (both players counted), the first 3 on the seventh, and the first 4, 8 and 9 on the fifth. The study records when each value first comes down, not why. A likely reading: low cards wait for a stone where they complete three of a kind or the bottom of a run, while high cards open stones that can still become anything.
What it changes at the table
Three practical readings, in order of confidence.
First, do not count your 9s to know who is ahead. The winner of the game played half of them, no more. Count formations instead: which stones can still become three of a kind or a colour run, and which are already sums.
Second, give each value the job it does best. Ends of the scale, 1s, 2s, 8s and 9s, are your three-of-a-kind cards. Middle values are also your colour-run cards, twice as often as a 1 or a 9. A 9 left over still makes a decent total; a 1 left over does not, so decide early where your 1s will go.
Third, a low card is not a dead card. It is held longer, not thrown away, and three 1s still beat any colour or run. Just do not build three of a kind with 1s against an opponent who can still finish a higher one. For how a stone is decided, read how to claim a stone in Schotten Totten, and for the odds of each formation in a deal, Schotten Totten formations ranked by rarity.
The study in five lines
- A 9 sits on the winning side of its stone 59% of the time, a 1 only 47%. The climb from 1 to 9 is broad; the middle values sit within noise of each other.
- The game's winner played about half the cards of every value, between 49% and 52%. Laying more high cards did not mark the winner.
- 1s and 9s win most often in three of a kind, almost half of their wins. The 3s to 7s win in a colour run about twice as often as the 1s and 9s do.
- A 9 wins on the plain total five times as often as a 1 does, but every value loses in a plain sum about as often.
- Low cards stay in hand longer. In the median game the first 1 or 2 comes down on the eighth move, the first 8 or 9 on the fifth.
Frequently asked questions
Which card value is the strongest in Schotten Totten?
The 9, but by less than you might think. In 400 games between two copies of the site's AI, a 9 sat on the side that won its stone 59% of the time, against 47% for a 1. The values in between climb, and 3 to 6 sit close together, around 51 to 53%.
Do you win more games with more high cards?
Not in our games. The game's winner played between 49% and 52% of the cards of each value, which is about what chance gives. Both players draw from the same deck. Who lays the 9s does not pick the winner; what does is beyond this study.
What are low cards good for?
Three of a kind, mostly. Almost half of the 1s on won stones sat in three 1s, against 36% for the 2s and 37% for the 8s. A low card is also the start of the lowest runs. It is rarely worth anything on the total, and in these games it came down later than high cards.
How were these Schotten Totten card statistics measured?
The site's AI played 400 games of the base game against itself, with the same settings on both sides and a fixed seed, so the run can be repeated and gives the same numbers. At the end of each game we looked at every clan card on the table, noted whether its side had won, lost or not yet decided its stone, which formation that side held, and whether the card was played by the game's winner.