Have any of you math types looked at floating advantage as if applies to shuffle tracking? Presumably you would have the advantage in "known-count" slugs where you have a good idea of the marriage (let's, just for the sake of argument, postulate that there could be such a slug), but would you get it in an NRSed slug where you only know the count on half the cards?
Floating advantage applies to any count-based strategy in any card game which can be counted. However, assessing its importance in the type of situation you envisage is rather complex and unlikely to yield significant extra return...
Floating advantage-strong version advocated by DS and ML: that the advantage in the true count evaluated remainder of a pack of cards [BTW this can be within a well bounded tracking segment as well] is the same as if the number of cards left were the starting number of cards, in all respects.
My definition of Floating Advantage: a perception of the strong version created by a combination of rounding errors, in accounting of actually possible true counts, a bow effect, where the predictions of your true count distort as per standard rules involving the Graham-Stokes equation, and a statistically insignificant tendency for basic strategy expectation to rise within a pack of cards as it depletes.
Bow-effect: A well known effect, little applied to blackjack, until recently [and somewhat poorly applied at that BTW, IMHO], where the differences between a perfect measure of player edge within a depleted pack, and an inperfect measure of player edge, with comulative errors in measuring player edge, and where the two measures cover the same range of predictions of player edge and the same mean of their measures, has a predictable difference in measured edges, where the lesser measure shows increases above actual player edge in more extreme ranges of its measure, and decreases above actual player edge in more middle ranges.
That said, what you are always going to find is that the increase in player edge from being able to play according to the composition of subsets that are far smaller than the full pack of cards, is always going to far outweigh the differences that occur within a tracked segment due to either major view of floating advantage.
You are thus straining at gnats and allowing free passage for camels. Forgetaboutit!
Just compare the gains even from the strong version of the FA, in BJA, with the losses from how well you estimate the boundaries of a tracking segment.
