Those Troublesome Turnovers, the Brat Stat

Guest Post by Abacus Reveals

I got hooked on crossword puzzles as a kid, and Sudoku number mazes joined the addiction about ten years ago. I’m particularly fond of the ones that provide a good challenge and require some time and thought – maybe even a little trial-and-error along the way. The “Aha!” one feels when the solution finally reveals itself (no peekin’ allowed) is priceless.

Of course, the path to the “Aha!” can be fraught with frustration.

I’ve been Stuck on Frustrated for what seems like the longest time now with an enigmatic basketball statistic. I’ve stared at the raw data, tested out the numbers in various arrangements … but still no “Aha!”

The confounding stat is Team Turnover Percentage, specifically the values provided at basketball-reference (b-r).

A few years back while messin’ around with the number puzzle that is an NBA box score, I conjured up a number combination that I’ve come to call the “SPOR-t” score. The acronym stands for: “Scoring Plus Offensive Rebounds minus Turnovers.” The calculation is simply: FG% + OR% - TO%. For a TO%, I’d simply calculate the ratio of TO’s to total possessions. That seemed logical.

(I use the SPOR-t score as a component in my periodic Power Ratings.)

But my TO% never seemed to match what b-r offered, not even when I came across an alternate method for determining possession and gave it a try, not even with b-r’s own possessions formula.

The authorized stat was always a bit lower/smaller than my personal number-crunching produced. Number sense requires that, in order to obtain a lower TO% from the same number of TO’s, the number of possessions must increase. The required boost to match b-r’s posted data for TO% came to about eight or ten possessions a game.

Clearly, things weren’t adding up – this puzzle was not getting solved.

A visit to the b-r Glossary provided the “Aha!” to the arithmetic inconsistency, but only further befuddlement in the basketball arena.

The problem is one of words, the defining of terms. You see, b-r defines TO% as the ratio of TO’s (not to possessions, but) to something called “plays.” Alas, this mini-dictionary provided no definition for a “play.” But it would seem that when an offensive rebound extends a “possession,” a new “play” has begun – or so I deduce from the formula for TO%. (This might also account for those eight or ten “phantom possessions” the data was hinting at.)

I can’t help but think/wonder/hope that there might be an amusing historical anecdote that accounts for this record-keeping peculiarity.  (Are “plays” rather than “possessions” used in any other statistical measures?)

Maybe the basketball gods just don’t want everything to “add up” neatly.

For example, here’s another “Hoops Aha!” for which I continue to seek:

Is it not both ironic and rather disingenuous that the NBA disciplines its players for flopping, but not its franchises?