Junky's Top 30

If Jeff Sagarin can create a computer poll that is used by the BCS to rank teams to determine who plays for the national championship in college football, what stops anyone else from doing the same thing? Well, it probably helps that Sagarin was a mathematics graduate of MIT and has an MBA from Indiana University. But does plugging numbers into a computer and having it spit out some score really help us understand how good a football team is in comparison with the its competition?

 

Well, I've often wondered that myself. So, while waiting for college football season to start this last summer, I attempted to create my very own computer rankings using mathematical values I felt were proper and fair in evaluating the performance of all Division 1-A college football teams. The ranking system took several months to built and rebuild until I felt comfortable with the finished product. Alas, we have BroncoJunky's College Football Rankings (or just simply Junky's College Football Rankings).

Now, you might ask, what makes this computer ranking any different from others out there? That's a good question, because I really have little to no idea what criteria other computer polls use for generating their team rankings. But, I can tell you what I was attempting to do with my own rankings. It has always bothered me that there are over 100 teams in Division 1-A but only 11-12 games a year for each team to play. How can one evaluate how good a team is when they play only about 10% of all the teams in their same division? When creating this system, I wanted to show how a team played against its own opponents in comparison with how everyone else played against those same opponents. To do this, I created what I call the "Game Value Score." Simply put, this score (for example) would show how Hawaii played versus Alabama in comparison with how every other team played Alabama during the season. Also, I wanted to reward teams who played, and won, on the road and who didn't lose at home. In addition, rather than just counting wins and losses, I wanted to reward teams for winning against teams with good records and for not losing to teams with poor records. Taking these into considerations, I came up with what I call the "Power Score," "Average Win Value," and "Average Loss Value" scores. Taking these together with a teams strength of schedule and margin of victory, you have the basis behind Junky's rankings.

I certainly don't have time to go in depth about the values used to calculate a team's computer ranking, but if you are curious, there is a link at the end of this article to Junky's College Football Rankings website. There you can find more information on how these rankings are calculated. Plus, there are links to Junky's final rankings for 2003 for all 117 Division 1-A teams as well as current rankings for all those teams. Below is Junky's Top 30. Please note that it is very early in the season, and therefore, difficult to really evaluate teams based on these rankings. Some teams have still only played one game, which makes it nearly impossible to compare their performances with other teams. Also, there are no preseason values assigned, so everything is based purely on each team's performance for this season. Give it a couple more weeks, and these rankings should start to make much more sense.

The website and these rankings will be updated each week (generally each Monday), so you can follow along for the remainder of the season.

 

Junky's Top 30

(as of Sept. 13, 2004)

 

Rank

Team

GVS

PVS

SOS

PWR

SUBSC

WINVAL

LSVAL

SCORE

W

L

1

Southern California

10.551

0.424

6.000

4.500

21.476

33.072

0.000

54.549

2

0

2

Auburn

4.467

0.420

9.500

5.000

19.387

33.376

0.000

52.764

2

0

3

Stanford

2.703

0.430

9.500

4.000

16.633

35.790

0.000

52.423

2

0

4

Utah

4.625

0.352

9.500

5.000

19.477

32.790

0.000

52.267

2

0

5

Purdue

9.978

0.470

5.375

4.000

19.823

32.366

0.000

52.190

2

0

6

Texas

4.644

0.407

8.875

5.000

18.926

32.334

0.000

51.260

2

0

7

Oklahoma St.

2.633

0.314

6.000

5.000

13.946

34.011

0.000

47.957

2

0

8

Troy St.

3.863

0.293

5.375

6.000

15.530

30.648

0.000

46.179

2

0

9

Ohio St.

3.830

0.327

5.375

4.000

13.532

30.584

0.000

44.117

2

0

10

Boston College

1.017

0.345

5.375

5.000

11.737

32.173

0.000

43.910

2

0

11

Florida

7.391

0.415

7.000

7.000

21.806

21.132

0.000

42.939

1

0

12

Colorado

-2.626

0.283

6.000

5.000

8.657

33.574

0.000

42.231

2

0

13

Alabama

4.731

0.380

5.375

4.000

14.486

27.033

0.000

41.519

2

0

14

Boise St.

6.058

0.371

1.875

1.000

9.304

30.174

0.000

39.479

2

0

15

Arizona St.

0.681

0.351

2.500

2.000

5.533

33.274

0.000

38.808

2

0

16

Oklahoma

4.071

0.368

2.500

1.000

7.939

30.100

0.000

38.039

2

0

17

Louisville

0.000

0.396

2.500

2.000

4.896

31.714

0.000

36.611

2

0

18

Fresno St.

0.000

0.342

2.500

3.000

5.842

30.659

0.000

36.502

2

0

19

Virginia

0.706

0.362

2.500

2.000

5.569

30.597

0.000

36.166

2

0

20

Tennessee

1.774

0.356

2.500

1.000

5.630

30.527

0.000

36.157

1

0

21

Kansas

0.894

0.416

2.500

1.000

4.810

31.121

0.000

35.931

2

0

22

Iowa

0.000

0.384

2.500

1.000

3.884

31.312

0.000

35.196

2

0

23

Louisiana Tech

0.000

0.301

2.500

1.000

3.801

31.264

0.000

35.065

2

0

24

Maryland

-0.493

0.309

2.500

1.000

3.316

31.510

0.000

34.827

2

0

25

Wisconsin

-0.360

0.426

2.500

1.000

3.566

30.911

0.000

34.478

2

0

26

Texas Christian

2.276

0.336

1.875

1.000

5.487

28.680

0.000

34.167

2

0

27

Indiana

1.251

0.338

1.875

2.000

5.464

28.523

0.000

33.988

2

0

28

Louisiana St.

0.251

0.379

1.875

1.000

3.505

29.600

0.000

33.105

2

0

29

West Virginia

1.522

0.351

1.875

2.000

5.748

26.981

0.000

32.730

2

0

30

Georgia Tech

1.124

0.322

9.500

5.000

15.946

16.749

0.000

32.695

2

0

Visit Junky's College Football Rankings website:

Junky's College Football Rankings


Bronco Country Top Stories