Thursday, October 20, 2005

Football Wonk - Better than Expected

The Blogpoll Rountable this week had a question about margin of victory and whether it should be accounted for in the BCS computer rankings. A few years ago I got to thinking about margin of victory, and while I actually think it can mean something, I got to thinking that a different margin should mean more. The margin between how a team performed against an opponent and how that opponent normally performs. Example: Team X scores 35 points against Team Y. Team Y has one of the best defenses in the country, averaging giving up 10 points a game. Team X therefore score 25 points better than expected against that team.

So in a way to focus on achievement, I looked at the statistics for the seven undefeated teams as of today (USC, Texas, VT, UGA, Alabama, UCLA and Texas Tech). Then I went game by game and tallied up how much better or worse each team played against its opponent in each game. I looked at scoring offense, scoring defense, total offense and total defense. I'd take the scores and yards from the head to head matchup out of the equation, then compared the scores and yards against the averages for the opponent in all games other than the one at hand. Example: UGA-Boise St. I'd take the offensive scores for UGA (not counting defensive and special teams scores), subtract that from Boise's Defense total scores against, re-average BSU's scoring D, and then compare it to the actual score for the Dawgs. In this case, Georgia scored 22.8 points more than they "should" have had Boise State given up their average. I did this for each game played by the unbeatens (not counting TT's 1-AA games). Then I tallied it all up and averaged it out.

Again the best reason for this system is that playing inferior or superior opponents doesn't really change the equation. If you're playing BFE State, and they normally give up 45 points a game, blowing them out doesn't say anything. But beating a team that's normally very good and blowing them out does, like the Alabama-Florida result.

Here are the results:

Scored 124.9 points more than expected. That's 20.8 points more per game. 3 TDs!
Gave up 84.9 points less than expected. 14.2 points less per game.
Totalled 1415 more yards than expected, 235 more per game.
Conceded 647 fewer yards than expected, 108 fewer per game.
Basically, when teams play USC they give up three touchdowns more than normal, and score two touchdowns less than normal. This is incredibly impressive. Also, their victory over Oregon might be the most impressive win of the year, as compared to the other team's averages (scored 3 TDs more than normal, gave up 4 fewer TDs, and gained 280 more yards than normal while giving up 263 fewer yards than normal) .

Scored 85.2 points more than expected. 14.2 per game more.
Gave up 86.8 points less than expected. 14.5 per game less.
Totalled 886 more yards than expected, 148 more per game.
Conceded 909 fewer yards than expected, 151 fewer per game.
Basically, it's a 4 TD swing when you play the Horns. You're going to give up two more than normal, and score 2 fewer than normal. Also, the Longhorn D's strength really shows here.

Scored 57.2 points more than expected. 9.5 per game more.
Gave up 83.2 points less than expected. 13.9 per game less.
Totalled only 92 more yards than expected. 15 per game more.
Conceded 827 fewer yards than expected. 138 fewer per game.
VT gets a lot of bang for their buck. They don't gain a ton of yards, but they make it count in the scoreboard. 25 point swing is still very good.

Scored 35.3 points more than expected. 5.9 per game more.
Gave up 73 fewer points than expected. 12.2 per game less.
Totalled 512 more yards than expected. 85 more per game.
Conceded 527 fewer yards than expected. 88 fewer yards per game.
The Dawgs haven't been blowing people away. But also, they're the reverse of VT, in that they seem to get less bang for their buck, yardage-wise.

Scored 31.8 points more than expected. 5.3 per game more.
Gave up 100.5 fewer points than expected. 16.8 per game less.
Totalled 327 more yards than expected. 55 more per game.
Conceded 575 fewer yards than expected. 96 fewer yards per game.
Alabama's D really has ben fantastic. Against the average, they've performed the best.

Scored 70.4 points more than expected. 11.7 per game more.
Conceded only 3 fewer points than expected. 0.5 per game less.
Totalled 360 more yards than expected. 60 more per game.
And actually they've given up more yards than their opponents normally gain. 278 total more, for an average of 46 more per game. And this has been the case in 5 of their 6 games.
UCLA are going to be in trouble when they play a team that can stop them. Their defense hasn't stopped anyone. Their numbers are probably the worst of the seven.

(remember, only 4 games, so the numbers might be skewed)
Scored 73.6 more than expected. 18.4 more per game.
Gave up 32.9 fewer points than expected. 8.2 less per game.
Totalled 929 more yards than expected. 246 more per game.
Conceded 59 fewer yards than expected. 15 per game.
As is self-evident, the Red Raider offense is really good, while their D leaves some to be desired. We'll see what happens when they play the Longhorns this weekend.

So there you have it. When you look at the data all together, it seems like USC is clearly the cream of the crop. Texas is a clear second. Alabama, Georgia and Va. Tech are the next level, with Texas Tech not too far behind (and I'd include them too if I had a little more data showing the same results - that FIU game skewed it a little too much). UCLA is the least dominant of this group, and it wouldn't surprise me to see them lose a game they probably shouldn't (though they're probably my second favorite team in this group, the facts is facts).