Is there an Ill 18(22) generated for these?
Yes, at the request of Mr. Lewis, I did the work for SD. The indexes were derived by algebraic approximation. For full tables see Wong's PBJ.
If so, what would be the % payoff in playing these as opposed to the standard Ill 18?
A very difficult question to answer with an ultimate precision. As with many BJ related questions an extensive simulation may be necessary to achieve a satisfactory answer.
According to Griffin, to compute the gain from learning cd indexes as opposed to the generic ones, you will need to compute the correlation for each two card hand, so as to obtain your average mean for the particular play and compare and see how much you have increased your correlation, in relation to the other one, the non composition dependence hand.
An example: Sd, s17, ndas, rsp = 4
Hand Hi-lo corr. Prob. of occurrence
16 vs T .558358
T,6 vs T .589865 0.014479638
9,7 vs T .522364 0.0038612368
Here your weighted correlation for T, 6 and 9, 7 is:
Wc = (Sum Ci*Pi)/2 = .575654
So your total gain in efficiency using both indexes has increased by:
.575654/.558358 ~3, 1%
From a practical point of view, I?ll bet that the complexity and memory efforts added are not fair compensated in a substantial amount of EV increase. I could simple be plain wrong, a possibility, which should not be disregarded.
Sincerely
Z
Bookmarks