Head scratcher

[gelesen]

32 BYU = 84.34 21 8 75.93( 82) 1- 2 | 2- 4 | 84.89 28 | 85.33 25 | 82.57 45
33 Purdue = 84.30 19 10 78.04( 48) 2- 4 | 7- 6 | 86.25 22 | 84.07 33 | 82.76 44
34 SMU = 84.30 22 6 74.43( 101) 0- 2 | 0 -3 | 83.18 42 | 84.32 29 | 85.01 24

Look at these Sagarin rankings. How can a team (SMU) with no top 50 wins and a SOS of 101 have essentially the same rank as us. Or BYU be ahead of us with all factors again weaker plus a loss to PU.

None of this never makes sense to me or probably ever will.

 

[BoilerinLouisville]

Yeah it makes you want to play a team like BYU to show them who would win head to head, oh wait....
Posted from Rivals Mobile

 

[SYoumans]

I used to create power ratings for college football and they were . . .

published weekly in "The Times" of Northwest Indiana (called the Football Forecast) and can tell you that the #1 complaint from readers was "you have X rated above Y even though Y beat X head-to-head, you moron." Well, power ratings do not and can not rank teams based on head-to-head results.

Using your example of BYU vs. Purdue: so you put Purdue ahead of BYU because we won head-to-head, so then you have to put North Florida over Purdue because they beat us head-to-head, and then you have to put every team that beat North Florida over them, but some of those teams that beat North Florida might have been beaten by BYU, and that breaks the hierarchy and throws the system into a mathematical blackhole. You basically just look at scoring margins and what those scoring margins SHOULD BE compared to schedule strength and calculate a power rating by applying that +/- number to a baseline number like 70. So, say Kentucky's average score margin is +30 and their average team played against is -3 vs. an average team (so their strength of schedule is slightly below average). Their power rating would be 70 + 30 - 3 = 97. Of course, how you calculate a team's average opponent's strength is complicated, but that's the basics.

I don't know much about Sagarin's methodology but the schedule rating difference between #82 and #48 is actually pretty small -- just a hair over 2 points (as in, the average team Purdue plays is 2 points better than the average team BYU has played).