| |
BGonline.org Forums
XG's Post-Crawford Take Point oddity - fixed but help needed
Posted By: Maik Stiebler In Response To: XG's Post-Crawford Take Point oddity - fixed but help needed (Rick Janowski)
Date: Saturday, 14 February 2015, at 8:41 a.m.
it is preferable to make consistent use of the computer generated values rather than mixing with the original 25 x 25 values
Years ago, I used Claes Thornberg's ME calculator to extend Zorba's G11 table. Thornberg's code calculates the match equities iteratively, starting at 1a:1a, so that whenever an entry for a new score is calculated, all values for scores that can be reached from that score have already been computed. I modified the code to start calculating at 1a:12a and refer to the G11 values if needed, thereby smoothing the transition between the original and the extended part, so that a procedure like the one you described would not be needed. I would recommend the same for XG. Extended tables could probably be calculated on-the-fly in a fraction of a second.
I also found that reducing the gammon rate all the way down to 18% gave better results, i.e. an even smoother transition, than using 26%. It seems to be needed to balance the big "ideally efficient cubes" assumption in the ME calculator.
By the way, I liked Timothy's suggestion of using whatever XG already has to estimate MWC at the start of the game (although I think XGR++ would not be needed, as it would make on-the-fly calculations impractical). I suspect that what XG does is at least as sophisticated as Thornberg's code.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.