I don't get the lack of computer love for BYU BB

[jay santos]

Originally Posted by pelagius

I would only be speculating, but I would guess that your methodology is being affected less by extreme observations. Since, Sagarin is basically a least squares methodology (at least the MOV (predictor) component is), Lamar could end up being a very influential observation. I would think a rating system that took a quantile regression approach would rank BYU higher.

I like that approach better. Most of what I do for work is very rudimentary analysis, but if I need a simple correlation tool, I'll take least abs values over least squares.

My model, like I said before, is not a true statistical model. It's a formula based, iterative model, but it's usually pretty close to Sagarin and other well known models.