Some Quick Thoughts on Turnover Margin

I have much to say on this. Suffice to say, of all the “stats that mean something” on defense, turnover margin, and ESPECIALLY fumbles recovered, are the most flukish. In fact, there is really no statistical correlation year to year in fumbles recovered at all. It is pretty much all a function of luck. Turnover margin has more of a correlation because bad offenses turn the ball over. Turnovers forced by defenses have very little correlation.

Who led the nation in fumble recoveries last year…..Buffalo, and by a lot.

Where’s the love?

Gene, that’s kind of my point. OU had to be several standard deviations to the good on both the fumbles recovered and fumbles lost bell curve. Chances are they’ll get worse in both cases, and by a significant margin.

With interceptions being more predictive, I’d say Texas gets better in this department. I love it when stats agree with my homerism and lack of objectivity. Hence my post.

Keep in mind I’ve had a busy summer consisting of drinking Mai Tai’s while eluding angry black men.

You go, girl!

I hate you with the heat of a thousand fiery suns.

Considering OU was +23, UF was +22, what the hell would this team have done if they were at that level instead of +2?

And how did they only get 6 picks in the Big XII? I’d almost bet my first born that it at least doubles to 12 this year.

Chris Applewhite’s “regression to the mean” article is all fucked up. Ah, and he talks like a fag, and his shit’s all retarded.

“Considering OU was +23, UF was +22, what the hell would this team have done if they were at that level instead of +2?”

I can tell you what we would have done if we were +3.

Our defense averaged one forced turnover less per game than any season during the Mack Brown era

Wow.

“Several Standard Deviations to the good” would put them inside the top 3ish% of the population (D1 CFB teams in this instance), if I am thinking about things clearly. I am fast and loose with data, as opposed to concepts, so I might not have my percentage right. Some engineering imbecile could probably break it down a little better, but they’d probably also kill the thread.

Whatever the case, I think the SD for this population is about 3.6. The mean was 10.5 fumbles gained last year. 15 is outside of 14.1, so they were more than an SD away, making a regression more plausible, but not nearly 2 SDs away. Looking at their historicals in this regard under Stoops would give you a better indicator of whether or not they are commonly that far away from the center of the normal distribution. Again, I am probably bastardizing it, but my point is that their point on the curve for the fumbles gained stat is far from being an anomaly, although it is relevant.

They’re fumbles lost number is a legitimate outlier. LSU fits that as well. They are more than 2 standard deviations away from the mean for that stat. I do agree that they will come back to the pack here, specifically because this particular category isn’t really a controllable item for a team, although I am fairly certain Sooner fans will attempt to tell us that it is and Stoops and Wilson have arbitraged the sport in developing the handling of this item.

Not trying to nitpick, just trying to put the two things in the right lens.

that’s enough for me, get rid of coach boom! he’s mean to me.

Thanks for the clarification. I was fast and loose with the std’s on the recovered fumbles side, but I never said it was an anomoly. I actually cited two BCS schools that had better fumbles recovered numbers.

I think turnovers made, to a certain extent, can be controlled by coaches. Mack benches fumblers, and the conservative pass routes (I’m convinced) are designed to avoid pick-risky passes over the middle. If Mack (or almost any other coach) had to choose between a punt returner that would break 5 in a season for TDs, but also drop 4 punts, and a returner who broke no punts but caught every one, he would choose the latter.

Phil Steele tracks turnover margin. For the last seven years, here are the TO margins for selected B12 schools, what they had last year, and last year’s difference from the average (note- positive TOM means more TOs gained than lost):

Team Ave TOM 2008 TOM Delta TOM
UT +6 +2 -4
OU +69 (!) +23 +13
OSU 0 +5 +5
TT -1.5 +8 +9.5
BU -6 (!) +16 (!) +22 (!)
TAMU +2 -10 -12
KU 0 +4 +4
Mizzou +5.5 -4 -9.5

BTW, notable extreme years for teams are 2008 OU (23), 2008 BU (16), and 2007 KU (+21) to the good and 2007 BU (-18), 2002 TT (-10), and 2008 TAMU (-12) to the bad. It’s hard to get a feel on whether the 2008 BU and TAMU TOMs are outliers because they are new staffs. Johnson threw a lot of picks, and Griffin had very few- do we expect that to flip?

In short, expect mild TOM improvement for Texas, and mild worsening for OU, hopefully all on the same day in October.

Fast and loose with std’s. Bravo!

Well, he is from El Paso.

I thought you’d like that.

Outside of 2 SDs is, by definition, an anomaly. You didn’t need to use the term as the context implied it.

Ok, rpongetti.

Just don’t have Sailor Ripley ban me and turn my posts into Latin, man.

Gene Claude -

You will be quite interested in my next stats post. The only thing I’m lacking for publishing it is data entry and I figure I’m about 5 hours of pure keymashing away from being ready.

How much of our lack of interceptions can be attributed to our DB’s, especially the CB’s, not turning around and playing the ball?

I know last year was training wheels year for the DB’s, but we’ve seen this for a few years now and if it doesn’t change I would not expect the INT’s to go up by any meaningful amount.

Gene Claude -

Maybe that is why Auburn didn’t hire Turner Gill. They where regressing to the Gene???

 http://upload.wikimedia.org/wikipedia/commons/4/4c/GeneChizikAU.jpg