While looking over the analysis between Monty and Michael Petch, I
think I spotted a tiny bug in gnubg. It may be of interest to gnubg's
programmers to correct it. On the other hand, sorry if it seems
picky. Judging from the way the equities move after the very first
roll of a game, it appears that the equity-change is computed
wrongly. Suppose for example that A starts with an opening roll of
average value. Then gnubg seems to calculate the equity change as
being close to 0, reflecting the fact that, considering A started, A
has been neither lucky nor unlucky. However, it would seem much more
logical to compare the after-roll equity with 0 rather than to compare
it with a typical A-starts scenario.
Paul Epstein
|
