MonkeyFocker,
I really wish you'd read it before slamming it. The best measure of a model is: if your model suggests a bet on the opener, does the line movement agree?
If your model suggests betting an opener of -105 and it closes at -115, that is a 10-point victory. If that same game closed at +105 on that side, that is a 10-point failure. You look at openers and closers to analyze a model. If your model correctly predicts the direction of the line movement more than 60% of the time at whatever threshold you use, you are probably going to win.