Stat Relevance Watch Addendum: Now Tempo-Free?

At the risk of alienating the fine people coming to SMQ through an old hotlinked pic that's found its way into high-ranking rotation on Google Image Search results for "Pamela Anderson," SMQ would like to engage the substantially smaller portion of his audience with a penchant for amateur stat wonkery.

Last basketball season, hoops blog Big Ten Wonk began using "tempo free stats" as a way to get a clearer picture of offensive and defensive prowess than afforded by outright point per game totals, which are skewed heavily by pace - a team that runs up and down the court like SMQ's undefeated Seattle SuperSonics on NBA LIVE 99 may score more than a more methodical offense, but because the number of possessions for each team is going to be virtually the same regardless the pace, a better measure is the efficiency of "points scored per possession" and "points allowed per possession." This has caught on pretty well in basketball circles.

Maybe because there's so much less data (about ten times fewer total possessions in a season than in basketball), the same measure has not, to SMQ's knowledge, been applied to football. Brian Cook mentioned Big Ten Wonk and the idea of "points per possession" in his harangue against the archaic emphasis on time of possession in football last October, but there's been no corresponding effort to actually produce those numbers.

That may be because it doesn't tell us that much. Here is last year's per possession data for two very different offenses, Texas Tech on top and Michigan on bottom, excluding obvious kneeldown possessions of only two or three plays or a couple yards at the end of a half. What's in question here is the relevance of these numbers as a comparative tool:

High flyin' Tech went further per possession than Michigan, but the teams were almost identical in terms of scoring efficiency. As much as the Raiders are considered a "high octane" offense, and Michigan is considered stodgy and boring because it doesn't put up 650 yards and 50 points on the Baylors of its schedule, the average offensive possession of each team yielded about the same result. Take out Southeast Louisiana in Texas Tech's schedule and the scoring averages become basically identical.

This isn't so much to compare the relative success of two very different styles of offense, though - especially when it doesn't include any nod to field position, presumptively a big key in Michigan having to travel shorter distances for its points in the far right column, where it is more efficient and far, far, far more consistent than Tech - as it is to test the viability of this measure. On one hand, Michigan comes out here looking every bit as good offensively as Texas Tech, which cuts against the conventional wisdom. On the other, there's not much variability in the number of possessions for each team; Michigan has a few less, but it's close enough that standard per game averages should tell more than they do in basketball, where there is much greater variability in the number of possessions depending on style of play. In this case, Texas Tech was sixth in total offense and 13th in scoring; Michigan ranked 38th in total and 26th in scoring, so it is somewhat enlightening that Michigan was about as dangerous on a per possession basis. But maybe only somewhat, and maybe ony due to factors not shown here.

SMQ's initial thought is that the "yards per point" category - or possibly "points per yard," not calculated above but done so easily - might be the best measure of efficiency. But it doesn't take into account field position, which is a big deal. What do you think - can "tempo free stats" lead to a better understanding of the game? Is it worth the tedium to expand on these numbers? Is there any reason defensive numbers might be more or less relevant than offensive? Digame.