Why 13/11 6/5
Two plays share the top. 13/11 6/5 brings a builder down and slots your 5-point, the most valuable point on the board, hoping to cover it next turn. 24/23 13/11 splits the back checkers instead, so a hit costs less and more rolls make an advanced anchor later.
The top five plays, rolled out
| # | Play | Equity | Behind the best | Wins | Gammons won · lost | 3-ply (rank) |
|---|---|---|---|---|---|---|
| 1 | 13/11 6/5 | +0.036 ±0.015 | — | 50.3% | 14.8% · 14.7% | +0.002 (#2) |
| 2 | 24/23 13/11 | +0.012 ±0.013 | 0.024 ≈ | 50.2% | 14.1% · 13.8% | +0.002 (#1) |
| 3 | 24/21 | −0.018 ±0.014 | 0.054 | 49.3% | 13.1% · 13.4% | −0.023 (#3) |
| 4 | 13/10 | −0.052 ±0.014 | 0.087 † | 48.6% | 14.7% · 15.0% | −0.034 (#4) |
| 5 | 24/23 24/22 | −0.056 ±0.012 | 0.091 | 48.9% | 12.8% · 14.0% | −0.048 (#5) |
Equity is the average result in points per game for the player on roll, with the cube in play: each play rolled out 1,296 times, ± one standard error. Gammons include backgammons. ≈ marks a play behind the best by less than twice the larger standard error, inside the margin of error. † marks a play whose rollout cost and 3-ply cost differ by more than 0.05; treat that number as rough.
The alternatives
The slot finishes ahead in the rollouts, by 0.024, while the 3-ply evaluation rates the two level: play either. The other candidates are clearly worse: running a back checker with 24/21, bringing one checker down with 13/10, or splitting both back checkers with 24/23 24/22.
All 15 legal plays, ranked at 3-ply
- 24/23 13/11 +0.002
- 13/11 6/5 +0.002
- 24/21 −0.023
- 13/10 −0.034
- 24/23 24/22 −0.048
- 24/22 6/5 −0.063
- 24/23 6/4 −0.086
- 13/11 8/7 −0.112
- 8/5 −0.119
- 24/23 8/6 −0.124
- 6/3 −0.131
- 24/22 8/7 −0.146
- 6/5 6/4 −0.150
- 8/7 6/4 −0.186
- 8/7 8/6 −0.320
Questions about the 2-1 opening
What is the best opening move for 2-1 in backgammon?
13/11 6/5 and 24/23 13/11 are effectively tied. In GNU Backgammon rollouts of the top five plays, 13/11 6/5 scores +0.036 and 24/23 13/11 +0.012, a gap inside the margin of error.
Is 24/23 13/11 a mistake with an opening 2-1?
No. It finishes 0.024 behind 13/11 6/5 in GNU Backgammon’s rollouts, inside their margin of error, so the two plays are effectively equal.
How was this 2-1 analysis made?
GNU Backgammon 1.07.001 ranked all 15 legal plays at 3-ply, then rolled out the best five, 1,296 games each, with the cube in play (money game, Jacoby off). GNU Backgammon 1.06.002, the version Boardgammon’s servers run, ranks the same five plays in the same order at 3-ply.
How these numbers were made
GNU Backgammon 1.07.001 ranked every legal play at 3-ply with the cube in play (money game, Jacoby off, no beavers), then rolled out the best five: 1,296 games each, with variance reduction and quasi-random dice. GNU Backgammon 1.06.002, the version Boardgammon’s servers run, ranks the same five plays in the same order at 3-ply. Evaluated on 2026-10-20. The same engine analyses every finished online game on Boardgammon: see how that compares. New to the numbers? Read what equity and a rollout are.
Terms on this page: Slot, Builder, Golden point, Anchor. Next in the strategy guides: cube theory.