Before. That is, it works the same way as the interface for a full shoe. For example, entering "1" for the number of decks (a full single deck) yields exactly the same per-hand EVs as entering "0 4 4 4 4 4 4 4 4 4 16".
Items 2 through 5 will take some work. I may have some time to take a look at this over the holiday, will see what I can do, but I don't promise anything
.
Item 6 requires some clarification. As you may find looking elsewhere, CAs generally don't include insurance in EV calculations, because insurance is "separable" in the sense that we can simply add the EV for any insurance wager; the EV for the initial wager is unaffected by whether you take insurance or not. (If you want variances or probability distributions, on the other hand, then things get more interesting.)
Also, insurance EV depends on your strategy for when to *take* insurance. For example, if you use a perfect insurance side count (i.e., if you take insurance exactly/only when it's advantageous to do so), then it's easy: the overall, pre-deal additional EV from taking insurance given an initial shoe of n total cards, of which a are aces and t are tens, is a(3t-n+1)/(2n(n-1)), which is only positive if 3t>=n-1 (exactly when you *would* take insurance with a perfect count).
But if you're using any other imperfect counting strategy to determine when to take insurance, then the resulting EV depends on that strategy, which raises the question of how to *specify* that strategy. (The counting analyzer count_pdf.cpp allows specification of this strategy as a linear true count whose tags may or may not be identical to a count used for playing strategy during the round.)
Eric
Bookmarks