[fraac]

[ QUOTE ]
In large databases you'd probably find the rankings for such a list would vary from player to player (even among successful players). A wild guess makes me think that the very best and worst hands would generally be similar, but the middle hands would vary the most.

[/ QUOTE ]

Depending on how each player plays the different hand classes, yes. So I'm not sure that averaging results from more than one player would yield anything meaningful. However, finding a few great players, getting their starting hand EVs combined with their hand class call/raise ratios (when I say this does it make sense? I'm guessing that the way someone plays low pairs or suited aces, for example, is pretty fixed over a broad set of game conditions (such as those specified in the OP)), and doing what the neurolinguistic crowd call 'modelling' them may be useful indeed.