Thursday, August 16, 2007

Saurian Sagacity’s National Power Ranking

After chewing up Stewart Mandel’s rankings of college football “national powers”, I couldn’t help but wonder if there was some sort of unbiased ranking system I might devise using the data available.

Well, here it is.

Before explaining my final methodology, let me at first describe the process in deciding what to use. I wanted to use rankings data that was as unbiased as possible, or at least not influenced by my own opinion. Data I was looking for would include ideas such as winning records, school revenue from football, and recruiting. I was purposefully not going to include in my system any data that related to “national titles”, as the process is highly flawed and the very essence of bias (to say the least). I was also not going to include conference championships, because of the vast difference between conference talent and schedules (not to mention title games).

In the end, I settled on the following 3 data types –

1. 10 Year Winning Percentage – I choose 10 years as being currently relevant, but also a period long enough to reflect some real data points. I think a period of 10 years reflects pretty accurately success on a “national” basis, plus it gives reduced impact to teams no longer currently competitive (remember, Army was once a dominant team).

2. Average Home Game Attendance – I was looking here for a statistic that would reflect revenues generated by individual college programs through all means (donations, tickets and merchandise). However, no private colleges release this information, nor do several public schools (such as Penn State). For those that do release the information, I realized a high correlation between overall football revenues and average game attendance. For instance, the top 10 schools by average attendance, and their place in the REPORTED football revenue rankings, are –

1. Michigan (4th in revenue)
2. Penn St (does not report)
3. Tennessee (12th in revenue)
4. Ohio State (2nd in revenue)
5. Georgia (3rd in revenue)
6. LSU (8th in revenue)
7. Alabama (6th in revenue)
8. USC (does not report)
9. Florida (5th in revenue)
10. Texas (1st in revenue)

Obviously, the comparison is imperfect, but the only complete data set available is the average attendance.

3. Average recruiting class points for the last 4 years – Obviously, the ability to recruit well is as strong an indication as a “national power” as anything. Also obviously, it is a subjective exercise. However the bias isn’t mine. Plus, even though we use the rankings from Scout.com in this system, they are pretty close to the other ranking services.

The Equation

To come up with an overall equation for our ranking system, we kept it pretty simple. First I took the average of each data points for the 115 teams we had data on. The averages were –

1. Winning Percentage - .5135%
2. Avg. Attendance – 44,200
3. Recruiting Class Avg. – 929.96

Next, we assigned a value for each team to the percentage they were above or below the given average. Using Penn State as an example –

1. PSU winning percentage .6116% equals 1.191 (19% better than average)
2. PSU attendance of 107,567 equals 2.433 (143% better than average)
3. PSU recruiting of 1909.75 equals 2.054 (105% better than average)

Lastly, to get an overall score, we add them up – 1.191+2.433+2.054 = 5.678

Presenting Saurian Sagacity’s National Power Rankings (10 Years, with score)

1 Florida 6.796
2 Southern Cal 6.639
3 Michigan 6.602
4 Tennessee 6.548
5 Texas 6.307
6 Georgia 6.197
7 Louisiana State 6.148
8 Ohio State 6.132
9 Penn State 5.678
10 Oklahoma 5.588
11 Florida State 5.542
12 Auburn 5.462
13 Alabama 5.088
14 Notre Dame 5.087
15 Nebraska 5.083
16 Miami-Florida 4.819
17 Texas A&M 4.711
18 Clemson 4.674
19 Virginia Tech 4.631
20 South Carolina 4.488
21 California 4.395
22 Wisconsin 4.362
23 Oregon 4.347
24 UCLA 4.258
25 Arkansas 4.202
26 Iowa 4.087
27 Virginia 3.907
28 Pittsburgh 3.825
29 Texas Tech 3.824
30 Mississippi 3.811
31 Michigan State 3.801
32 West Virginia 3.773
33 Georgia Tech 3.768
34 North Carolina 3.759
35 Maryland 3.633
36 Arizona State 3.621
37 Washington 3.619
38 North Carolina State 3.527
39 Arizona 3.495
40 Purdue 3.447
41 Oklahoma State 3.446
42 Brigham Young 3.372
43 Louisville 3.347
44 Missouri 3.328
45 Kansas State 3.267
46 Oregon State 3.112
47 Illinois 3.068
48 Colorado 3.059
49 Boston College 3.016
50 Mississippi State 2.966
51 Washington State 2.908
52 Kentucky 2.869
53 Minnesota 2.862
54 Stanford 2.816
55 Utah 2.811
56 South Florida 2.803
57 Boise State 2.747
58 Syracuse 2.740
59 Rutgers 2.625
60 Iowa State 2.546
61 Texas Christian 2.537
62 Kansas 2.425
63 Fresno State 2.416
64 Marshall 2.399
65 Southern Miss 2.399
66 Connecticut 2.384
67 Air Force 2.264
68 Wake Forest 2.243
69 East Carolina 2.226
70 Hawaii 2.207
71 Colorado State 2.198
72 Toledo 2.145
73 Central Florida 2.057
74 Northwestern 2.045
75 Memphis 1.999
76 Miami-Ohio 1.964
77 Baylor 1.961
78 Troy 1.952
79 San Diego State 1.942
80 New Mexico 1.909
81 Indiana 1.869
82 Cincinnati 1.867
83 Vanderbilt 1.814
84 Navy 1.814
85 Houston 1.772
86 Bowling Green 1.758
87 UAB 1.751
88 Middle Tennessee State 1.729
89 Tulane 1.721
90 Western Michigan 1.715
91 Northern Illinois 1.714
92 Akron 1.630
93 Duke 1.614
94 Tulsa 1.575
95 Louisiana Tech 1.565
96 Wyoming 1.545
97 Nevada-Reno 1.522
98 Central Michigan 1.479
99 Ohio University 1.477
100 Idaho 1.445
101 Rice 1.434
102 San Jose State 1.430
103 North Texas 1.405
104 Arkansas State 1.370
105 Southern Methodist 1.315
106 Ball State 1.295
107 Louisiana-Monroe 1.253
108 Army 1.249
109 New Mexico State 1.245
110 Kent 1.194
111 Utah State 1.173
112 Temple 1.132
113 Eastern Michigan 1.131
114 Louisiana-Lafayette 1.105
115 Buffalo 0.834

Naturally, the average score comes out to “3.00” using the methodology employed. The standard deviation is 1.517, which means the real statistical outliers are those with scores of 6.03 or more (more than 2 standard deviations). Thus, your “first tier” national powers are –

Florida
Southern Cal
Michigan
Tennessee
Texas
Georgia
Louisiana State
Ohio State

2nd Tier is anyone from Penn State to Virginia Tech. (more than 1 standard deviation).

“Average” teams are listed from South Carolina (20th) to Nevada-Reno (97th).

And the truly pitiful –anything below 97th.

I had hoped Florida would not come out on top (this was not an exercise in data fitting), but they did. And, I guess with the best average recruiting class in the past 4 years, the 9th best average attendance, and 9th best winning record, it is not much of a surprise (The difference between USC and Florida is USC’s 13th overall winning percentage).

Of course if you don’t like it, you can create your own.

But this exercise wasn’t about Florida. It was an attempt to show that, when you slice the data, certain teams belong considered among the elite.

Regardless of elitist opinion.

1 comment:

Henry Gomez said...

I cringed when I saw number 1 because I know the reactions it's going to generate. But this wasn't an exercise, I don't think, about who's number 1. It's about who the top tier is. And honestly if you had done this before the start of last season with the same formula Florida wouldn't have been number 1. You have to attribute that to the season and recruiting class Florida put together in the last 12 months.