The general method I use: Start by estimating the probabilities of the ace landing on my first card, dealers first card, my second card, dealers second card, and none of the above. I've come up with a corresponding set of EORs for each of these 5 situation (identical for player card 1 & 2; standard betting EORs for "none of the above"). From there I can produce "net EORs." For instance, in the case where you play three hands, and are certain that one of them will get the target ace, you're net EOR would be:

.333 * EORs given ace in hand + .666 * Standard EOR

In this situation, Hi-Opt II with no side count does very well at 97.7% correlation (quite a bit better than it handles any of the five situations individually, interestingly). But in the shoe games where the above formula might make sense, maybe you have such a high edge when the ace is coming that you're always heavily underbetting. In that case a negative count wouldn't make a difference anyway.

But the same method yields a convenient 99% correlation with Zen on a single deck game, where underbetting isn't really an issue.

So, for practical purposes, a count derived from the above method is probably the most useful application of the data.

> ?..not beyond a motivated and disciplined intellect.
> :-)