Tuesday, November 17, 2009

Power Ranking

Here are our week 11 Power Rankings, with the methodology explained here.

TCU moves up to 2nd overall based not only on their outstanding offensive and defensive rankings, but also their SOS which places them 22nd in the country, or five spots ahead of Florida.

And yes, that is 4 loss Oklahoma in 5th place, a result I’m not entirely comfortable with. But much of the reason I created this Power Ranking system was to remove bias from the procedure – notably my bias – and base the results on statistical fact, not impression. Oklahoma has lost 4 games, but by a total of 12 points (14-13, 21-20, 16-13 and 10-7). They are playing outstanding defense with the nation’s 5th best by Scoring Defense, and have the nation’s 16th best offense at 33.3 points per game. Finally, schedule-wise they are 8th in the nation.

Looking at it another way, the only team with a both higher Scoring Offense and Scoring Defense than Oklahoma is Florida. And if you consider all three metrics (Offense, Defense and Schedule Strength) no team has higher values in all three areas than Oklahoma.

In other words, Oklahoma is one pretty good 4 loss team.

My top 25 in Power Rankings with their scores –

1 Florida 2.2107
2 TCU 2.1878
3 Texas 2.1307
4 Alabama 2.1017
5 Oklahoma 1.9064
6 Ohio St. 1.8605
7 Virginia Tech 1.8582
8 Cincinnati 1.7677
9 Penn St. 1.7491
10 Oregon 1.7382
11 Clemson 1.6924
12 Oklahoma St. 1.6683
13 Iowa 1.6451
14 Boise St. 1.6334
15 Arkansas 1.6273
16 Wisconsin 1.5985
17 Rutgers 1.5860
18 North Carolina 1.5419
19 Oregon St. 1.5413
20 Miami (FL) 1.5374
21 Georgia Tech 1.5374
22 Pittsburgh 1.5356
23 LSU 1.5257
24 Nebraska 1.5126
25 Arizona 1.4836


Once again I use these rankings for game score predictions, and have done so with some reasonable success. How this works is thus –

Take the rankings of each team and multiply by 27 (average points scored in a NCAA game).

Subtract the higher from the lower.

Adjust by 3 points (or 2.5, depending on your preference) for home field.

You have your spread. For example in the Colorado at Oklahoma State this Thursday -

Oklahoma State = 1.6683
Colorado = 0.8135

1.6683 * 27 = 45.04

0.8135 * 27 = 21.96

Difference is 23.08. Add 3 for Okie State at home, and they are 26 points better than Colorado. (The spread is about 14, so take the Cowboys – for fun only, of course).


And, in case you are curious here are teams 26 to 120 –

26 South Fla. 1.4736
27 Houston 1.4511
28 Mississippi 1.4442
29 Connecticut 1.4345
30 Texas Tech 1.4211
31 Fresno St. 1.3977
32 Boston College 1.3969
33 Navy 1.3837
34 Mississippi St. 1.3786
35 Air Force 1.3704
36 Missouri 1.3653
37 Florida St. 1.3593
38 Southern California 1.3519
39 Tennessee 1.3386
40 Georgia 1.3318
41 California 1.3043
42 Notre Dame 1.2941
43 Michigan St. 1.2933
44 West Virginia 1.2733
45 Auburn 1.2726
46 Stanford 1.2362
47 Kentucky 1.1829
48 Minnesota 1.1821
49 South Carolina 1.1803
50 Central Mich. 1.1780
51 Utah 1.1643
52 BYU 1.1392
53 Troy 1.1340
54 UCF 1.1056
55 Arizona St. 1.0985
56 Iowa St. 1.0768
57 UCLA 1.0759
58 Nevada 1.0727
59 Kansas 1.0670
60 SMU 1.0662
61 Baylor 1.0579
62 Purdue 1.0527
63 Washington 1.0491
64 Louisiana Tech 1.0354
65 Southern Miss. 1.0351
66 Kansas St. 1.0252
67 East Carolina 1.0210
68 Bowling Green 1.0124
69 Wake Forest 1.0013
70 Louisville 0.9976
71 Michigan 0.9917
72 North Carolina St. 0.9881
73 Idaho 0.9750
74 Virginia 0.9682
75 Duke 0.9665
76 Marshall 0.9624
77 UNLV 0.9371
78 Texas A&M 0.9346
79 Syracuse 0.9313
80 Colorado St. 0.9282
81 Buffalo 0.9261
82 Temple 0.9198
83 Indiana 0.9174
84 Northern Ill. 0.9023
85 La.-Monroe 0.8981
86 Wyoming 0.8928
87 Tulsa 0.8841
88 Northwestern 0.8740
89 San Diego St. 0.8729
90 UAB 0.8472
91 Illinois 0.8327
92 Arkansas St. 0.8292
93 Toledo 0.8269
94 Middle Tenn. 0.8226
95 Colorado 0.8135
96 Fla. Atlantic 0.7938
97 Utah St. 0.7814
98 FIU 0.7681
99 Hawaii 0.7561
100 Vanderbilt 0.7538
101 La.-Lafayette 0.7378
102 UTEP 0.7217
103 North Texas 0.7048
104 Akron 0.7040
105 Maryland 0.6932
106 Kent St. 0.6868
107 Ohio 0.6808
108 Western Mich. 0.6778
109 Memphis 0.6708
110 Miami (OH) 0.6436
111 Tulane 0.6384
112 San Jose St. 0.6188
113 Army 0.6086
114 Ball St. 0.5426
115 Washington St. 0.5161
116 Western Ky. 0.4568
117 Rice 0.4540
118 New Mexico St. 0.4413
119 New Mexico 0.4034
120 Eastern Mich. 0.4011

Eastern Michigan – the nation’s worst team.

9 comments:

Floridan said...

So I guess if you want to root for the home team, don't live in New Mexico.

Mergz said...

Ugh. I hadn't seen those 2 in spots 118 and 119.

Unknown said...

FIU better than Vandy? We need to bring our A game.

Robert said...

Having trouble wrapping my head around the concept of a "pretty good" 6-4 team. Yeah, the Sooners don't give up many points, but they also lose to everyone with a pulse that they play. They've racked up those gaudy numbers against the little sisters of the poor and been stiffled by the good teams on their schedule.

Rob said...

"Power" rankings often have things like a 4 loss team highly ranked, becuase they dont care about things like winning and losing.

Iowa was highly ranked in score differential power rankings last year, despite 4 losses.

I wonder if this method better predicts future results than a point differential power ranking of the Sagarin predictor? If it doesnt, I dont see the point (okay, since we dont know the exact formula of Sagarin, I can see why this would be useful, but how does it compare to a pure power ranking?)

Trader Rick said...

So we beat the panthers by 42 points...

jj gator said...

Illinoise absolutely sucks, yet their nincompoop AD wants to keep Zook around. Birds of a feather flock together, and those two moRons deserve each other.

Floridan said...

Although the Power Ranking does not correlate with wins & losses, it does provide a useful perspective.

If, for instance, a team like Oklahoma has a PR muchhigher than their record would indicate, it should lead one to look deeper into the causes of that disparity.

It could mean untimely injuries, outlier upsets, poor coaching decisions or other factors that undermine the natural strength of a team.

Pton98 said...

Looking at your power ranking formula, I had a question: do a few lopsided wins throw off the scoring offense and scoring defense numbers in a way not accounted for by the strength of schedule? Extreme wins and losses can affect avg without pushing the strength of schedule numbers a corresponding amount the opposite way.

For example, Oklahoma's numbers factor in 76-0 and 45-0 against Idaho State and Tulsa. Throw out those games and I bet that their numbers are less impressive.

Home blowouts against bad teams who are paid to show up in September and lose horribly don't really tell you much about how a team will play against more even competition.