[Aisthesis]

I think exactly the kind of situation you're talking about is where this consideration is most relevant. Basically, you're sufficiently shortstacked in a tournament situation that you're not going to be able to play the hand out, so you're all-in or fold.

Just to get the further consideration started, let's assume the following situation: The blinds have escalated to 100/200 with antes of 25(I like Cloutier's suggestion of viewing the ante stage as a time to move, since the antes really do noticeably increase total pot-size at this point).

So, the first question is: At what stack-size "x" (expressed as number of big blinds in your stack) is it enough for the all-in to have a 50% probability of having the best hand? With a larger stack, you'll presumably want the probability of having the best hand to be higher.

For simplicity's sake, it's probably also best to assume that it's folded to you in the given position. UTG, then, you're automatically talking about the ranking for 9-players, and it goes down from there. One will also need to assume that the "bunching factor" is irrelevant if it's folded to you in LP or in SB.

The number of callers expected is also going to be a relevant factor in choosing the hand-ranking system. But I think as first assumption, 1 caller ought to work and seems fairly typical--at least if my rough guess as to stack-size is accurate.

Then there's the further question of the probability that the all-in will fold the remainder of the field. Actually, here, the assumption made in the discussion leading to the Sklansky-Karlsson rankings seems like a good start: Just assume that only hands superior to yours will call (I think that was the way it was done in the relevant article).

I'm not sure exactly how to do this, but with the given assumptions (which seem fairly reasonable), one can assign a 50% probability to the remainder of the field folding. Then, you have to figure out what the equity is on your "median hand" vs. the superior hands. I think it should be possible to come up with an actual maximum stack-size for which it would be the best play to go all-in with the given hand (although this stack-size may vary from one position to the next).

While these assumptions restrict the scope of the question a little more than necessary (e.g., how about if you have callers before you?), I think working through the stack-size issue under these restrictive assumption would be the next step in the analysis.