[wahooriver]

I would like to thank the many responders. I apologize for not making my point more clearly.

Statistics (and here I assume that ICM is a statistical model) depend entirely on assumptions. If one changes the underlying assumptions then our results can change dramatically.

Models like ICM are most useful for analyzing simulated situations.

As I read the responses, the responders like ICM because you can adjust the assumptions. I believe you have made my point. The most important skill in poker (including SNGs) is to understand your opponents and how they probably view you. The examples supporting ICM make that point. You must estimate the likelihood of you opponents making a play (etc.).

ICM represents and interesting exploratory model, but its supporters point out that reading your opponents is more important. That is my point. We must learn to read our opponents more than relying on mathematical techniques.