Monday, August 17, 2009

The Superlative Teams of the Decade

Part 3 – Margin of Victory (and Football "IQ")

In our first two parts of this piece we looked at the top Scoring Offenses and Scoring Defenses as defined statistically using standard deviation for the decade (thus far).

We did the Top Ten MOV teams of the decade in an earlier piece, but as you will see this will be a little different.

The MOV average is 0.88 for the decade with a standard deviation of 11.3 points per game. No teams fell more than 3 standard deviations higher than the mean which would not be surprising under a “normal distribution” as we would only expect this to occur 0.1% of the time, or perhaps once in our sample size of 1054 teams. The closest was the 2001 Miami team at 33.78, or about 1 point short of 3 standard deviations.

On the other hand we would expect 22 teams to fall 2 standard deviations above the mean in a normal distribution (2.1% of the time) and we got almost exactly that with 24 teams.

The Superlative teams of the decade by MOV are -

1. 2001 Miami (Fla.) 33.78
2. 2005 Texas 33.75
3. 2002 Kansas St. 32.97
4. 2000 Florida St. 32.12
5. 2008 Florida 30.71
6. 2004 Louisville 30.05
7. 2001 Florida 29.72
8. 2008 Southern California 28.54
9. 2000 TCU 27.67
10. 2003 Oklahoma 27.63
11. 2000 Miami (Fla.) 27.14
12. 2002 Boise St. 27.12
13. 2008 Oklahoma 26.64
14. 2007 Kansas 26.37
15. 2005 Southern California 26.28
16. 2003 Boise St. 25.9
17. 2004 Utah 25.83
18. 2001 Texas 25.47
19. 2004 Southern California 25.15
20. 2008 Boise St. 25
21. 2000 Toledo 24.96
22. 2008 Penn St. 24.54
23. 2008 Texas 23.61
24. 2003 Miami (Ohio) 23.6


BCS title winners in this group include –

1. 2001 Miami
2. 2005 Texas
5. 2008 Florida
19. 2004 Southern California

The other BCS title winners fell accordingly –

27. 2000 Oklahoma
28. 2003 LSU
58. 2007 LSU
92. 2006 Florida
93. 2002 Ohio State

Even at 93rd the 2002 OSU team had a MOV of 16.19 or about midpoint through 1 standard deviation above the mean.

For comparative purposes let’s use an example most people are familiar with – the Intelligence Quotient, or IQ. IQ is believed to fall along a standard Gaussian Bell Curve, or normal distribution, with a mean of 100 and a standard deviation of 15. Using this as a guide we can equivocate an IQ score for the BCS Championship teams for the decade (as compared to their MOV).

The “Football IQ’s” of the BCS Champions (rounded) –

2000 Oklahoma – 131
2001 Miami – 145
2002 Ohio State – 121
2003 LSU – 130
2004 USC – 133
2005 Texas – 145
2006 Florida – 122
2007 LSU – 125
2008 Florida – 141

American Mensa accepts a 132 score on the Stanford Binet IQ test, so by that measure Miami ’01, USC ’04, Texas ’05 and Florida ’08 are the football “geniuses” of the pack.

The Truly Miserable

A single team fell 3 standard deviations below the mean in our sample, or had a Margin of Victory worse than -33.05 points per game. That team was the winless 2005 Temple Owls with a MOV of -35.57. The football IQ equivalent of the ’05 Owls is … 53. My candidate for worst team of the decade.

What Now?

As I mentioned before Dr. Saturday made a list of his top 10 teams of the decade so far. Since we have all this data at our disposal I plan on doing the same from a statistical viewpoint. I also agree that winning the BCS in any given year doesn’t necessarily mean you are the “best” (just a champion). In making my top 10 list MOV will also be important, but I plan on adding one more statistical input – strength of schedule. The 2002 Kansas State team, for example, appears to be one of the top teams of the decade by MOV (as well as Offensive and Defensive Scoring). Should they be considered a top team?


Next – My Top Teams of the Decade.

2 comments:

Jams said...

would love a graph to show how close to a bell curve MOV is. By the results, it seems like it's pretty normally distributed, which makes sense, since MOV is not bounded above or below.

anyway, sweet analysis.

Mike said...

Why do you guys feel so compelled to over analyze college football? It all seems to be an exercise in futility.

The sport is as imperfect a human endeavor as ever invented. In any one game multiple situations - first downs; touchdowns; and even yards gained are determined by the capricious spots of the football made by the zebras. The coaches, through analysis, may use scripted play calling but most freely admit that when it comes down to the critical decisions they use their gut to make the call. Let's not forget the calibre of competition and the personnel matchups that give statistically inferior teams a better chance to win then their records/achievements reveal.

The detailed results of your analyses make this football fanatic's eyes roll back in his head. If you are targeting the engineer crowd you have probably succeeded. Congratulations.

If you feel like flexing your standard deviation muscles again, let's go for broke - let's get each players vital statistics (weight lifting, jumping and speed/distances measurements), determine the protein/caloric intake since their last digestive cycle, make notes of each play's atmospheric conditions (wind direction, field temperature, etc), postion of each player on the tilt of the peaked playing field and any other variables you decide to include. Then match each player against his opposite number (keeping in mind zone or one-on-one coverage of the receivers and stunting schemes by the O and D lines) during each play.

That kind of per game analysis, season summaries, and team rankings by decade and all-time best would be worth hearing about.