Beating Shitty Defenses: the 2008 Texas Longhorns

I don’t think so. Look at the defenses OU’s “Greatest Offense of All-Time” faced last year. From the top of my head I can only think of three good defenses they faced last year TCU, Texas, and Florida and only one was truly elite.

The fact is most big time programs face shitty defenses because the teams that play great defenses are the ones you face maybe three times a season.

You can also say the opposite for Florida’s great defense last year. Apart from OU, name a great offense they faced? Georgia? Alabama? Florida State? I’m not disputing that Florida’s D was great. It was. But the fact is elite teams rarely face teams that are elite in defense, because they are the ones that field the elite defense and you only get those matchups about 3 times a season.

In order to do this properly, what you have to do is recalculate the defenses’ rankings by eliminating the team in question’s performance against them. For example, to calculate the ranking of the defenses of Texas’ opponents, you have to re-average everything without their game against Texas.

The reason you do this is, if Texas shows up to your school and hangs 100 on ‘em, it’s going to affect the average for that school.

Actually what you’re best doing is calculating the factor of what percentage of the team’s defensive average points allowed that Texas scored on them. That gives you a far better idea of how well Texas’ offense did against any given defense. So if a team gives up 23 ppg on average (except for us) and we scored 45, our offense was pretty fookin’ effective.

I actually have a script that does this for NFL teams each week. It will actually do this 4 times for any given pair of teams, multiply those factors by points allowed per game and scored per game, and produce a prediction of the final outcome of the game. Parity in the NFL makes the standard deviations enormous and the predictions (from a pure raw score standpoint) only slightly better than average (and dubious quality vs. the spread), but the factors are very valuable in determining which teams have good offenses and which have good defenses by taking into account the quality of the defenses and offenses (respectively) they’ve faced.

One of these days I hope to get it set up for college. The advantage of college is you can usually use last year’s output to feed into this year’s to some extent…

—Rimbo

Rimbo, my adjusted stats are the same concept taken to the extreme. Mathematically it’s the same thing as comparing a game performance to the opponent’s average then adjusting the team in question’s average. Then iterating as many times as necessary until the ratings stabilize.

PrimeTime -

In my opinion, the comparison I presented demonstrates that not all good teams face bad defenses. In 2005, the national championship team faced an average opposing defense ranked #51.5. In 2004, a very good Longhorns team faced an average opposing defense ranked #54.5. Like I said, between 1999-2005, Texas had only one season in which opposing defenses averaged below the national median. Oklahoma may have faced poor defenses this past year, but I don’t think that every top team will play as lousy a set of defensive units as the Longhorns saw last season.

Rimbo -

I’ve actually run the “percentage above opponents’ YPGA” calculation. That was the number I started this whole process to obtain, since I’ve always argued that Texas outperforms bad defenses by a wide margin, but hovers around the average against quality defenses. The recent run of weak opposing defenses between 2006-2008 has largely prevented me from determining whether that theory has any validity post-Vince.

I was hoping not to have to dig to a high level of detail for this particular post, since I believe the enormous discrepancy between the non-adjusted average ranking in 2008 and pretty much every other year is evidence enough. Adjusting the rankings won’t make that much difference. If anything, it will only emphasize the drop-off in quality compared to our national championship season. In 2005, Texas faced defenses with an average rank of #51.5 and outperformed them on average by 47.4%. In 2008, Texas faced defenses with an average rank of #83.9, but only outperformed their YPGA average by 23.9%. So, an adjustment to 2005’s rankings is likely to have a more significant effect in terms of increasing the opponents’ average rankings than would a 2008 adjustment. In other words, 2008’s slate of opposing defenses is even worse in comparison to 2005’s than my simplified analysis indicates.

In fact, here are the adjusted rankings for 2005 and 2008:

2005: #42.8; -16.2
2008: #82.3; +22.3

Unadjusted, the swing from 2005 to 2008 was 31.36 compared to the median. Adjusted, the same measure is 38.5. As I had suspected, the difference between these two seasons is more significant after adjusting to remove Texas’ numbers against its opponents.

However, adjusted rankings will likely compress the difference between 2008 and, say, 2006 and 2007, but probably not by much.

Huck -

I was hoping you’d chime in. Do you have the averages based on a more sophisticated ranking?

You mean greg’s high powered offenses with solid running schemes would suck ass in the SEC? no way!? I still hate the fat man and can see through the haze of having superman in 05’ and a kid completing 80% of his passes in 08’; that would make anyone look good.

Playing the Spread game after game will do that to defensive rankings. In other words, the higher a conference’s offensive rankings go, the lower the defensive rankings will tend.

Huck –

I was hoping you’d chime in. Do you have the averages based on a more sophisticated ranking?

What do you mean? What averages? Season total yards per game allowed?

I don’t have any data from years before 2008, so nothing I have will assist with prior years. I could run some quick and dirty analysis at some point, though. The problem is that I would have to use published NCAA totals but only D-1A games in the schedule for previous seasons. Some error is introduced there.

Stats are for losers… Will Muschamp

Huck —

There’s also the Football Outsiders, who do this sort of thing on a per-player, per-play basis.

But really, as Muschamp says, stats are for losers.

“Stats are for losers.”

I agree. Just check Huck’s posts that demonstrate the strong correlation between bad statistical performance and losing.

[PhD Nutrition] Pharma Whey: I agree completely with Mike. Strawberry tastes odd, it has a very bitter edge when used with water, for no apparent reason. Maybe some of their extra chemicals are giving it a harsh after-taste. Tastes better with milk, but not fantastic. The strawberry y does froth up loads after 5 seconds of shaking. You have to wait a while for it to settle before you can drink it like a normal shake. Try Boditronics Express Whey in chocolate. It s not amazing, but it s very easy to drink and mixes well.