Over the years I had come to prefer the Ward method, just based on general impressions and a handful of test cases. In preparing to write this article, however, I decided to do a more rigorous test, comparing the Thorp and Ward formulas.
   I have a database of bearoff equities on my computer for about 180 million positions.
All the positions in which both sides have 11 or fewer checkers in their home boards are there, plus a few million 12-checker positions. So I wrote a little program to apply the Thorp and Ward formulas to all of them and tabulate the right and wrong answers.
   The result was that Ward got 163 million right answers. Thorp only got about 153 million right. A goodly proportion of the extra Thorp errors are rather serious ones, such as evaluating drops as no-doubles and vice-versa. Ward makes very few of these major blunders.

-- Excerpts from "Adjusted Pip Count Methods" / Walter Trice