[eastbay]

If you're looking for something to do to advance the theory, my suggestion would be to pick up the idea I detailed in "ICM with a skill factor" (or something like that) and try applying it along with the usual ICM to some standard SnG scenarios and see if the results change, and how. Try to figure out if it gives better answers than the original formulation.

The second thing which is more complex is the issue that any stack ratio based valuation model is probably going to be insufficient for true expert play at Giga/DN's level because it misses a lot of the strategic subtleties. Relative stack positions (who's to the left and right of who), blind size, and position (and other things) are all things which clearly affect the value of your chips, and yet ICM ignores all of them outright. So, how to start generalizing? I've proposed ways to consider position with a simple discounting approach. But I think you could quickly get lost in complexity where it wasn't clear if you were making progress or just making things foggier and less clear if they made any useful sense.

There is another approach altogether that I hope to return to sometime soon, but I'm not getting into what it is right now.

eastbay