Stat Relevance Watch, Part One

There are, as they say, lies, damn lies, and statistics. The numbers mean something, yet often we know not what. Here SMQ will look at the final regular season statistics in more than a dozen  major categories to suss out who succeeded in what and how that statistical success correlated to overall success in terms of final record. SMQ does not have the luxury of a high-powered supercomputer or degree-type qualification in mathematics or statistics, but his analysis will be driven as deep as his egghead, tinfoil cap curiosity and cell phone calculator will take it. That is to say, quasi-scientific at best.

Part One: Which stat correlates most closely to success?

There are many ways to define "success;" SMQ has done so for simplication purposes by stating simply: level of success=record. This is not strictly true, of course, as an 8-4 team from the Big Ten, because of the higher level of competition it must overcome to achieve that record, has likely had more success than an 8-4 Sun Belt team, if such a thing existed. But such distinctions are too fine here.

The method: From the NCAA's rankings in 13 major statistical categories (rush offense, pass offense, total offense, rush defense, pass efficiency defense, total defense, turnover margin, third down efficiency rate, fourth down efficiency rate, third down efficiency defense, fourth down efficiency defense, penalty yards and time of possession), pull out the top 20 teams and bottom 20 teams in each category. Combine the records of those teams into a winning percentage for the best and worst of each category. Rank categories by number of wins of the top 20 and the bottom 20, as well as the margin between the two; the "most important" category, it would follow, would be the one with the best records at the top, the worst records at the bottom and, therefore, the greatest disparity.

Notice the term "correlate," which of course is not "cause." There are a number of reasons for the records in each category to look the way they do, and numerous possible cause-effect relationships to be drawn (i.e., teams that trail often have to throw to catch up, and throwing therefore is inherently tied to bad teams; vice versa in the case of rushing). So the fact that, for instance, the collective records of the best passing offenses do not match those of the best rushing defenses is not an argument against ever throwing the ball.

The most obvious conclusions to be drawn from this set are a) being good at anything correlates to winning; the top teams in every category are collectively in the black, record-wise, all but the least penalized winning more than 63 percent of the time. Remember as we go on that we're talking on the whole about winners at the top of each category across the board, and dealing mostly with just a few percentage points difference in degree between them as groups. And b) defense comes out looking pretty good. The top three and four of the most five successful categories are defensive. The best total offensive teams in terms of yards can compete with the high rates of success by their defensive counterparts, but by category, the best rushing, third down conversion and especially passing offenses lag a bit behind.

But the relevance of a statistic shouldn't be measured only by the relative success of teams that perform well in a given category, but also by the relative failure of those that don't. On that level:

Here again, the teams at the bottom of each category are, predictably, losers without fail or hope. The order of success, or lack thereof, from the winning categories above in some cases is merely reversed, but with the noticeable trend that teams that were bad on offense, or that turned the ball over, fared worse than teams with forgiving defenses, and generally lost to a greater degree than their successful counterparts in the same offensive categories won. Which could be interpreted as, "Defense wins games, but offense keeps from losing them." Or something to that effect.

So let's move on to what SMQ sees as probably the most relevant measure of this analysis, the relationship between the winning percentages on the high and low ends of each category:

First counterintuitive result: the most penalized teams were slightly better as a whole than the least penalized teams. Penalty yardage, over the course of an entire season, had no discernible effects on winning and losing. You can probably think of a situation that would specifically argue otherwise, cuz penalties are definitely bad, mmmkay?, but they're bad more as situational mistakes than an overall, cumulative drain.

What's more interesting is that offensive categories in general come out looking far more important in the relative measure, which reveals just how truly horrid teams that couldn't move the ball or convert or third downs really were. At the extremes, winners played lights out defense, but big losers offer some evidence that offensive deficiency was slightly more difficult to overcome this season than a defensive one. Again, on the whole.

Now, to the asterisk by fourth down conversion efficiency: seven teams nationally converted 70 percent or better of their fourth down attempts, and those seven teams (Boise State, Auburn, LSU, Michigan, Southern Cal, Louisville, Ohio State) won more than 92 percent of their games. Six of them are in the BCS; the seventh is ten-win Auburn. And nobody on the list went 3-4 or anything - all seven attempted at least ten fourth down tries, and USC converted 19 of 27 fourth downs. After those seven, below the 70 percent threshold, it evens out and drives the top 20's winning percentage down for the entire group. But no other statistical category can claim anything like that sort of success at the very top. More than any other number, very good fourth down percentage equated to a very good team (though not necessarily vice versa).

Extrapolating: want to be good? Play good defense. Want to be very good? Convert fourth downs. Want to not completely suck? Move the ball and convert a few third downs. Penalties? In the end, irrelevant.

But this is just one way to measure statistical relevance, of course. Part Two will look at what the very best teams were very best at, exactly, and further installments planned beyond the second will go into much greater detail.