I've posted a couple of threads recently trying to come up with a model to illustrate this.
The most recent is this:
http://forumserver.twoplustwo.com/sh...3452&page=
In a nutshell, my theory is that your expectation is based not just on your stack size, but your ability to grow it over time, which has some limiting factors constraining it.
For example, you can never win more in any one hand than the lower of your stack and your opponent's stack. And you can never win more than all the chips in the tourney.
More concretely, if you have more chips than anyone else at your table, you can't "invest" all of your chips in a good hand, so not all of them are "working." Thus your potential growth rate decreases as you get farther and farther ahead of the field, no matter how good you are.
Thus I think the key to estimating your expectation is finding a weighted expected growth rate based on your relative skill, and the constraints of the tourney format.
To estimate your growth rate, I took a formula from biostatistics that's used to measure growth rates of species in environments with limited resources. The parameters are quite analogous to a poker tourney.
I need data to test it, but I have high hopes for it.
I have a lot more to say about this, but I'm at work, so I'll have to add more later.