I have come to the conclusion that true count is not necessarily the best metric to analyze blackjack.
I think that at some point Eric Farmer ran some sims that showed basic strategy outperformed TC indices in a significant portion of cases for many counting systems, although TC was a bit better statistically overall.
I would instead opt for running count and pen as the parameters by plotting a relevant RC versus pen graph. The position of the current (pen,RC) point relative to the graph would determine strategy. This approach would be more consistent than the TC approach which can more often be erratic.
Also I think that side counts may be more trouble than their worth.
A simple example of this approach: 6 decks, S17, NDAS, 1 allowed split all ranks, 1 card to split aces
Full shoe EV = -.579% (Using comp dependent basic)
Full shoe EV = -.582% (Using total dependent basic)
Some sample data points (using HiLo) to determine when appropriate to increase bet using only basic strategy
Code:
Cards remaining RC Estimated overall EV using basic strategy (composition dependent)
51 0 -.249%
51 +1 +.473%
50 0 +.011%
26 -1 -.347%
26 0 +.484%
26 +1 +1.442%
It may be unlikely to find a such a deeply dealt game, but the point is that there is an advantage at an RC of 0 that would not be identified using TC.
I get this data by using combinatorial analysis to get statistical probability of each rank using HiLo, then use these values to compute estimates.
Basic concept is simple.
k_c
Bookmarks