FLASH1296
Well-Known Member
FROM my favorite web site:
(Dead link: http://chance.dartmouth.edu/chancewiki/index.php/Chance_News_46)
Check it out
[This is a transcription of the FAX sent to Don after Dr. Griffin was notified of our results which indicated an error in his basic strategy table. -srj]
DON SCHLESINGER:
I have gotten on the computer and checked the figures. My recommendation is wrong!
About 1980, when I first had my "exact" program, I ran it for stand on soft 17, very carefully. I then developed a modification of it that, instead of recomputing the dealer's probabilities for every possible hand the player could hold, computed these figures only for 10 sub decks, then produced a table of "effects of removals" for dealer probabilities, from which, by linear formula, one inferred the actual probabilities, thus dealer hit soft 17, it gave
saving much computer time. I call this method "Quasi": for (6,2) vs 5, Hit = +.129242 Double = +.128437
Apparently I did not judge this to be close enough to recompute with the exact program, which recomputation I have just done, in double precision:
Hit = +.129947 Double = +.130021, and that's the truth!
I am sending you in the mail my Montreal conference paper which partially explains some of these approximations, including a new, interactive, one which would have put me on the right side of the matter, it gives:
Hit = +.129943 Double = +.130016
I have no idea how many of these hands I have misplayed in the past 17 years in downtown Las Vegas - about 1 out of ~1000 off the top of the deck. I'm mortified.
(signed) Peter Griffin
[The following was hand written on the bottom of the FAX. -srj]
Actually I've known this for many years. I've been on a penny a day retainer from the Horseshoe to keep it quiet. Now the cat's out of the bag!
(Dead link: http://chance.dartmouth.edu/chancewiki/index.php/Chance_News_46)
Check it out
************************************************************
[This is a transcription of the FAX sent to Don after Dr. Griffin was notified of our results which indicated an error in his basic strategy table. -srj]
California State University, Sacramento
------------------------------------------------------------
Department of Mathematics & Statistics 17 JUNE 97
------------------------------------------------------------
Department of Mathematics & Statistics 17 JUNE 97
DON SCHLESINGER:
I have gotten on the computer and checked the figures. My recommendation is wrong!
About 1980, when I first had my "exact" program, I ran it for stand on soft 17, very carefully. I then developed a modification of it that, instead of recomputing the dealer's probabilities for every possible hand the player could hold, computed these figures only for 10 sub decks, then produced a table of "effects of removals" for dealer probabilities, from which, by linear formula, one inferred the actual probabilities, thus dealer hit soft 17, it gave
saving much computer time. I call this method "Quasi": for (6,2) vs 5, Hit = +.129242 Double = +.128437
Apparently I did not judge this to be close enough to recompute with the exact program, which recomputation I have just done, in double precision:
Hit = +.129947 Double = +.130021, and that's the truth!
I am sending you in the mail my Montreal conference paper which partially explains some of these approximations, including a new, interactive, one which would have put me on the right side of the matter, it gives:
Hit = +.129943 Double = +.130016
I have no idea how many of these hands I have misplayed in the past 17 years in downtown Las Vegas - about 1 out of ~1000 off the top of the deck. I'm mortified.
(signed) Peter Griffin
[The following was hand written on the bottom of the FAX. -srj]
Actually I've known this for many years. I've been on a penny a day retainer from the Horseshoe to keep it quiet. Now the cat's out of the bag!
