I posted this over on Purple Row, but I think it fits here as well.
I've been working on a different method for ranking teams. The idea is basically that you can't tell how good a team is just by their wins and losses. You need to account for their opponents as well as the margins of victory or defeat. Of course, to know how good the opponents are, you need to know about their opponents. In other words, you have to solve for the strengths of all the teams at the same time.
The method I've come up with does this. It calculates the probability that a team will win against an average (.500) opponent. One advantage is that it will work with unbalanced schedules. For example, if you have one division with a bunch of really good teams, they're going to beat up on each other and it will appear that they are all close to .500. Even Pythagorean standings won't be able to tell that the teams are better than average. There are strength of schedule calculations out there, but I don't think they are adequate.
Anyway, if you're interested I wrote up a description of my method here. I'm interested in any feedback people might have.