| |
BGonline.org Forums
Example of play prediction method - Bumps
Posted By: Rick Janowski In Response To: Example of play prediction method (Dmitriy Obukhov)
Date: Saturday, 18 May 2013, at 9:59 a.m.
The equities for a particular checker play over the full range of pre-Crawford scores may be imagined as 3-dimensional surface with the x and y axes representing the points needed by the player on roll and his opponent, and the z-axis, the height, representing equities. When there are two competing plays, such as GG+ and GS+, where the two surfaces intersect represents the series of switch points (i.e., where GG+ = GS+). The play prediction method aims to determine the approximate position of this intersection, as a simple linear function, usually with negligible error. This generally works very well when the two surfaces have slopes at clearly different gradients, such as when both the GG+ margin and GS+ margin are significant. However, where the two plays are generally fairly close at either GG+ and GS+, or both, localised bumps, typically caused by cube leverage issues, may cause the general lower surface to penetrate through the generally higher surface at isolated points or clusters. This appears to happen to some extent with the opening 21 where the player on roll is leading, and for 64 where the player on roll is leading and the main option is between 64R and 64S. I suspect some of the 2nd roll and 3rd roll positions where small doublets are thrown may be particularly prone to this localised bumping. However, relative errors will generally be fairly small, in my view.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.