View Single Post
Old 02-02-2012, 11:47 AM   #409

Posts: n/a

Originally Posted by TonyR View Post
"Terrible" because you don't understand it (you clearly don't) or because it doesn't give the answers you want it to? Or both?

Like any model Sagarin has its limitations. But it's a good model, as evidenced, for example, by the fact that the two Super Bowl teams are ranked 1 and 5. The major limitiations I see are that it ranks the whole body of work as opposed to how a team is playing currently, and it rewards/punishes for margin of victory/loss more than perhaps it should. For example, Denver probably gets punished for having a -81 differential as opposed to, say, Carolina's -23.
Terrible because it comes to many self-conflicting and irrational conclusions. I don't have much patience for models that don't reconcile with reality at the end of the day. The last one I looked at still shows the Giants as the 4th best team in the NFC. Many people knew better and predicted a Giants run.

Some things just can't be expressed in a formula.
  Reply With Quote