| |
BGonline.org Forums
Long Strong XG RO of 41$11
Posted By: Nack Ballard In Response To: Long Strong XG RO of 41$11 (Mislav Radica)
Date: Tuesday, 24 November 2009, at 3:16 p.m.
According [to Neil’s] 36k long 3/4ply XG rollout hitting play wins 0.33% more gammons and 0.21% more backgammon so why this play is not better at GG?
The backgammons are overkill. The relevant numbers for the H play are 52.92% wins of which 16.73% are gammons (or better). A shortcut for determining relative value is to add these together, yielding 69.65%. Then compare that to the N play’s 53.50 + 16.40 = 69.90%. N wins more wins + gammons than H does, so N looks better at the GG score.
The numbers from Snowie’s money rollout come to a similar conclusion: N is 53.5 + 16.2 = 69.7%, and H is 52.9 + 16.4 = 69.3%, indicating that N is better at GG.
Near play wins 0.58% more plain wins so it could be better at DMP, as [you] stated.
Right; this is even easier. For DMP, compare XG’s straight win percentages: N wins 53.50% and H wins only 52.92%. (Snowie’s numbers are identical to the third decimal place: 53.5% to 52.9%.)
At GS 24/20* should be better bcuz is gives 2.3% less gammons for Black.
True, but to get the whole picture don’t forget about wins. At GS, the better play H gets (52.92 – 12.51%) = 40.41%, while N gets only 53.50 – 14.81) = 38.69%.
Can I think about this problematic on that way?
Yes, assuming that you remember to factor in wins (as outlined above). Indeed, this stratification is the method Snowie applies to its money rollouts to show the supposedly best plays at DMP, GS and GG (accessible by clicking on the 1-1, 1-2 and 2-1 buttons on the right of its output display).
This method is a decent indicator, though it is crude. To obtain a more reliable answer, the position should be rolled out with the checker plays at GG instead of for money. Snowie-4 won’t do that; GNU and XG will.
The GNU (2-ply, or what I called 3-ply in the table) results rolled at GG and DMP confirm the crude methods’ conclusions that N is better at those scores. However, GNU’s money result appears to have a strong bias towards N that likely (at least partially) carries over into its AtS (At the Score) results. Gnu’s 2-ply N bias is the reason I encouraged Neil (or someone else) to roll out 41$-11 at DMP (and if N holds up at that score then it surely will at GG as well) either on XG or on Gnu 3-ply-move.
Nack
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.