Originally Posted by tthree
I would listen to these guys. I don't use HILO (HIOPT II) and play mostly shoe S17 games. I will say if the reference is right (I am not doubting it) I am surprised by it. That hand relies so much on neutral cards I can't imagine those are risk averse indices. The EOR table has 6, 7 and 8 as the only cards valuable in surplus. 2, 3 and 4 are pretty valuable in deficit as the 6 in surplus but not near as important as the 7 and 8 are in surplus. A pretty weak correlation to most counts without the side counts.
One reason why I had this question was that when I simulated this index using CVData and a toy high-level counting system, I got conflict results. When I used the original counting system, I got an index to split when TC > -4.1 (this number has been converted so that you could somehow think of it as the index for a hi-lo system). In a different simulation run where I divided each tag value by 2 and used the "half point support" option (so that supposedly the resulting indices should be consistent, i.e., also approximately half of the indices generated from the 1st set). This turned out to be indeed the case for essentially all indices except for this one: 3,3 vs 8. The second simulation run suggested to split 3,3 vs 8 when TC > -1.8 (also converted and comparable, with the half tag values already taken into account; otherwise the index was -0.9). Both simulation runs were based on the beat-to-death method, i.e., with a HUGE number of simulations. I think these inconsistent results basically agreed with what tthree suggested, i.e., the counting of the non-neutral cards has little correlation with a correct play for this (essentially only relevant cards would be 7 and 8).
My above indices also disagreed with the basic strategy for 2D from Wizard of Odds: http://wizardofodds.com/blackjack/strategy/2deck.html
. That is, my indices suggest to split as a basic stragegy (when TC = 0). But I noticed that all others' indicess mentioned above (i.e., split when TC > some positive number, despite for different # of decks) were essentially consistent with the basic strategy.
Maybe without side counting 7 or 8, one (reference) play would be using the H17 or S17 index and ignoring DAS here (especially for hi-lo system which treats both 7 and 8 as neutral). For certain counting system assigning a positive tag value to 7, there then is a very remote correlation of the TC with the splitting decision.