| |
|---|---|
| 62S-53 | |
- [P S9 X24 $77] ~31*15 5
[P S10 X20] /20*10
[P S13 X27] <62*15
[P S15 X25] ^20*5
d[P X2 S5] /5
s[S P18] /5
g[P X15] /5
Key: P = Point (8/3 6/3), S = Split (24/21 13/8), X = hit-and-split (24/21 6/1*), $ = Slot (13/5). Margins are in thousandths. Before the brackets, d = dmp, s = save, g = go. After the brackets is a bot symbol (~ Snowie, / GnuBG, < XG 3-ply, ^ XG 4-ply) and the number of trials in thousands, rounded down (an asterisk repeats a number).
For example, "s[S P18] /5" means that at (gammon) save, S is best and P is -.018 in a GnuBG rollout of 5000+ trials.
Miran mentioned the rollout result of [P X22 S22.1] /5 (posted on the site). Due to the combination of (a) the P/S margin being so high (though a mere 5k trials is fickle -- it could well be variance), and (b) the file having been rolled with the old version 16.0 ming, I suspect the file may be corrupt. Until it has been rerolled (with the same seed on version 90.0 ming), we won't know. In the meantime I've averaged that result with David recent contribution of [P S5.6 X17.5] 15*5, which explains the composite [P S10 X20] /20*10 listed above.
Isolating just P and S, here are money aggs (bot aggregates) for three related positions:
- 21S-53 [S P2]a
32S-53 [P S10]a
62S-53 [P S11]a
The duplication of Opp's 2s (13/11 and 6/4*) makes P and S close contenders, and at gammon save S is best in all three positions. (Also, for 21S-53 at dmp, S gets the nod.)
Nack