|
BGonline.org Forums
Question about match strategy for Grandmasters
Posted By: Rick Janowski In Response To: Question about match strategy for Grandmasters (Maik Stiebler)
Date: Thursday, 11 August 2016, at 8:16 a.m.
A word of caution. The Jacobs-Trice Fish METs are fundamentally flawed because the methodology adopted assumed the end result would be identical to that predicted by the Elo formula Random Walk theorem. The various parameters in Walter Trice's program were adjusted without any rational basis; they were merely changed step-wise to tally with presumed outcomes. The main reason that the results are flawed is because the relative values of 1-point matches (and other short match lengths) derived from simplistic random walk theory are unrealistic, favouring the stronger player too much. A far superior skew adjusted MET was included in the work by Norman Zadeh back in the mid 1970s, which included the first modern backgammon MET (pre-dating others by 2 or 3 years). Zadeh seems to have spot on with his predictions here because his normal MET was remarkably close to the METs derived by David Rockwell and Neil Kazaross, using the might and knowledge of the modern Bots supported and driven by ultra powerful computers.
|
BGonline.org Forums is maintained by Stick with WebBBS 5.12.