| |
BGonline.org Forums
XG cube analysis -- Methodology question
Posted By: Rick Janowski In Response To: XG cube analysis -- Methodology question (Henrik Bukkjaer)
Date: Friday, 7 December 2012, at 2:06 a.m.
Sconyers database gives exact equities for all 15 checker vs 15 checker positions in the both home boards covering races up to 90 pips long and all equities for the three cube positions plus cubeless equity. Longer races can also be checked using the single checker vs single checker race model which can cover races up to 375 pips long again with all equities provided.
The full range of these position will effectively cover x1 and x2 values (back-calculated cube efficiency values for both sides) from zero to about 0.9 in extreme cases). The only positions in reality which have a single x (ie x1 = x2) are generally those where the roller has something close to 50% winning chances). The deeper ply evaluation levels (XG+, XG++ and 5-ply and above) produce very similar values (from back-calculation of the various equities) as those provided by the exact Sconyers database. Morover, several fully-rolled out contact positions have back-calculated x1 and x2 values which are very similar to x1 and x2 values back-calculated from the deeper ply evaluations.
This may appear to be counter-intuitive but the results I have seen back up this effect of rapid refinement of equity and cube efficiency estimates at the deeper ply levels.
To clarify, by back calculation of x1 and x2 levels, I mean comparing a known equity (obtained from Sconyers Database or XG rollout using recommended parameters) with the theoretical equity equation to derive the implicit x1 or x2 value.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.