Well now we can get rid of the pen difference between Eric's thread and Gronbog's work. I errantly used win rate in the last T Count thread when T Count results only report c-SCORE but that wouldn't I have changed anything by much in relative terms. Lack of headers in Tarzan's posts caused some confusion on my part.
https://www.blackjacktheforum.com/sh...itivity-to-ROR
The things that aren't the same in each sim for my extrapolation is spread, Eric used 1-16 while Gornbog's sim used 1-12, the number of split's as Eric didn't allow respiting because it is too complicated for CDZ. The latter is certainly insignificant. So that just leaves the spread as an assumption that may cause some slop.
From Tarzan's results in the OP:
Tarzan Expert Count gets a 4.694% increase in c-SCORE over Hiopt2/ASC.
Tarzan Expert+KC=0.10 gets a 6.387% increase in c-SCORE over Hiopt2/ASC.
From Eric's sim in the link:
Hiopt2/ASC: c-SCORE 26.90
Optimal play and betting: c-SCORE 35.38
Adjusting Eric's c-SCORE numbers for Hiopt2/ASC by the appropriate percentage increases from Tarzan's T Count data:
Tarzan Expert Count: c-SCORE 28.16.
Tarzan Expert+KC=0.10: c-SCORE 28.62.
As I did before, if we make Hiopt2/ASC 0 and perfect play 100, T count would be:
(TcountSCORE - Hiopt2wASCcountSCORE)/(perfectplaySCORE - Hiopt2wASCcountSCORE):
Tarzan Expert Count: 1.26/8.48 = 14.86 on a scale of 0 to 100.
Tarzan Expert+KC=0.10: 1.72/8.48 = 20.28 on a scale of 0 to 100.
So Don's number of 20 seems to be very good for T Count's top performing count.
Bookmarks