Here is gammon save data for the four positions:
- 31P-62 s[S R7] /15 (Stick), s[R=S $51 U53] ~5
42P-62 s[$ S14 R14] "<=20 (Tom), s[$ S14 R15] <5 (David)
53P-62 s[$ S21 R32] /5 (Stick)
64P-62 s[$ S56] /5 (Stick), s[$ S25 R39 D42] ~46
Key: P = Point (make the 5pt, 4pt, 3pt or 2pt), $ = Slot (13/5), S = Split (24/18 13/11), R = Run (24/16), U = Up (24/22 24/18), D = Down (13/11 13/7). Margins are in thousandths. Before brackets, s is save. After brackets is a bot symbol (~ is Snowie, / is GnuBG, < is XG1 3-ply, "<= is XG2 3-ply with Roller cube) and the number of trials in thousands (rounded down). For example, s[$ S14 R14] "<20= means that at (gammon) save, $ is best, S is -.014 and R is -.014 in an XG2 3-ply checker, Roller cube, rollout of 20000+ trials.
For 42P-62-s, there is also a GnuBG rollout of s[S R29 U59] /5, rolled out by one of Stick's rollateers, but $ wasn't even tried as a candidate. Also, the wide S/R margin and the version number 0.14.3-devel makes me suspect corruption.
For review, the four positions are diagrammed below, each captioned with the best single (gammon save) rollout result currently available. The deeper White's point, the less willing Blue is to separate his back checkers.
|
| ||||||||||||||||
|
| ||||||||||||||||