In my 7 mutant cube skill experiments based on fartoffski "game stages" formulae, the results of a total of 101,000 games were as follows:
EXPERIMENT 11:
Mutant won 51.51% of games, 45.73% of points in 20,000 games vs GnuBG 2-ply
EXPERIMENT 12:
Mutant won 52.50% of games, 47.51% of points in 20,000 games vs GnuBG 2-ply
EXPERIMENT 14-a:
Mutant won 49.38% of games, 47.46% of points in 20,000 games vs GnuBG 3-ply
EXPERIMENT 14-b:
Mutant won 49.40% of games, 47.80% of points in 1,000 games vs GnuBG 4-ply
EXPERIMENT 14-c1:
Mutant won 49.94% of games, 48.19% of points in 10,000 games vs GnuBG 3-ply
EXPERIMENT 14-c2:
Mutant won 49.81% of games, 47.49% of points in 10,000 games vs GnuBG 3-ply
EXPERIMENT 14-d:
Mutant won 49.57% of games, 47.30% of points in 20,000 games vs GnuBG 2-ply
My mutant is not better then GnuBG (yet?) but good enough for me to enjoy mocking at the gamblegammon world's dogmatized jackoffski cube formulae. :)
You can see the descriptions, Python scripts, game logs and my analyses at
my web site:
https://montanaonline.net/backgammon/py2.php
While at it, feel free to look at my older Murat vs bots experiments from the main page:
https://montanaonline.net/backgammon/index.php
and to the other mutant vs GnuBG experiment from the "Deforming the cube skill dogmatism" page:
https://montanaonline.net/backgammon/py.php
MK
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)