First, a link to predictor:
http://www.usatoday.com/sports/sagarin/fbt05.htm
[ QUOTE ]
I think the big weakness it has is that it doesn't take into account how teams are playing recently.
[/ QUOTE ]
That is certainly true. However, I think most people overadjust for how teams are playing recently. A lot of times, "hot" teams are just experiencing some positive variance, so they aren't really any better, just luckier. So I don't think this is actually a big problem.
To me, its biggest flaw is that it cannot take injuries into account. That is something you need to consider carefully.
[ QUOTE ]
For example, Michigan State is #18 on it still.
[/ QUOTE ]
People cite Michigan State a lot. [img]/images/graemlins/smile.gif[/img] I admit that seems strange, but it might not be so wrong. Consider that against four excellent teams, they won AT Notre Dame and played fairly close to Michigan, Penn State, and at Ohio State. The thing about predictor is, you get credit for playing close games with good teams, even if you lose. Which makes sense.
[ QUOTE ]
That being said, I think this would have some good applications, especially in late season games when there is a lot more data.
[/ QUOTE ]
Yes, it's definitely stronger the more data you have.
[ QUOTE ]
How do you account for home field advantage with the formula? I am pretty sure Sagarin uses a universal constant for home field, which would not be good in many cases.
[/ QUOTE ]
He lists it in the rankings on the left hand side (every ten teams). Right now it's 2.62. It is indeed a constant for all teams. This turns out not to be such a bad assumption. Just using 3 works pretty well.
[ QUOTE ]
Also, this may be good for bowl matchups to pick teams that may be better than their record that have played a tough schedule.
[/ QUOTE ]
Yes, it's very good for seeing what teams are better/worse than their record. This year, Big Ten teams tend to be underrated and good SEC teams overrated.
Predictor is almost always very close to the Vegas line. If it's off by more than a couple of points, it's worth thinking about why, and often worth betting if you can't figure out a good reason.