|
Hi,
I am not sure if you refer to the Kleinman formula to estimate cubeless
winning chances in race based on normal distribution theory. If it is the
case, here is a brief description. It comes from the help section of Snowie
Professional version 1.1 (Under preparation, mainly a patch of V1.0)
André Nicoulin, Oasya.
The Kleinman Rule
In match play, the score will strongly affect the value of the take points.
Hence you need in this case a more precise formula, which will provide you
a probability of winning the race. In "Vision Laugh at Counting" (1992),
Dany Kleinman develops such a formula. His formula is based on a concept
very familiar to statisticians named normal distributions. The rule is
expressed has follows.
Compute the player pipcount, and decrease it by 4 for taking into account
the fact that he is on roll. This leads to the corrected player pipcount P.
Compute the opponent pipcount as usual, O.
Compute the difference D equal to O minus P. It represents the lead in the
race of P over O.
Compute the sum S equal to O + P. It represents the total length of the
race.
Compute the Kleinman metric D * D / S, i.e. D square over S. You then have
to compare the Kleinman metric with reference figures in order to know the
winning chances of the opponent:
Winning chances
of the opponent 17% 20% 21% 22% 24% 25% 30%
Kleinman metric 1.8 1.4 1.3 1.2 1.0 0.9 0.55
A more detailed table can be found in in "Fascinating Backgammon, Antonio
Ortega, second edition,1994. A description of how you reach this formula
can be found in "Vision Laugh ...", Dany Kleinman.
|