Put Me in Coach: Playing Time and Player Quality

Mandatory Credit: Derick E. Hingle-USA TODAY Sports

One of the staples of analytics is to look at per minute or per possession statistics when evaluating player performance, in order to control for opportunities the player receives. Still, it is common not only outside of analytic circles, but within, to attempt to make inferences about a player’s abilities outside of pure efficiency measures based on the amount of playing time.

The logic is pretty simple, it goes something like:

1. None of the statistics we have capture all of what a player is doing on the court.
2. The coaching staff has both an incentive to play the best players and access to information outside of the box score or SportVU data, including game film, practice reps and their basketball experience and knowledge.
3. Therefore. there should be some signal of a player’s abilities beyond the stats from how much time the player is allowed on the court.

Indeed, multiple statistical studies have shown that there is evidence of independent informational value based on minutes played. Daniel Myers Box Plus Minus (BPM) and Neil Paine’s Statistical Plus Minus (SPM) both show a positive relationship for minutes played in terms of long term Plus/Minus contributions beyond box score data.

A couple of my own studies have given similar indications. Simply plotting the results of a linear efficiency measure that I (and I alone) tend to favor shows that measures of efficiency tend to converge to a more accurate reading with more minutes, as one would expect. But also that the pattern look like a log linear increase in value, even as measured by box score efficiency.

The indication I took from this was to infer that a player was ‘replacement level’ rather than average without seeing adequate on court production, or some other reliable indicator.

In another study of rookies projected forward to the end of their rookie contracts, I found that rookie production, age, pre-NBA production and minutes played were the best predictors of future performance in that order. Draft spot was also correlated with minutes played for it to show independent value and minutes was more reliable in cross validation testing. But, again, minutes played was the weakest and least important predictor in the model.

All of these cases are indications that on average the amount of playing time a NBA player gets is an additional, but not overwhelming, signal of their abilities beyond their recorded production.

But applying that inference to a particular player is even more complicated than that, since each player faces different competition at their position (which may be above or below average) and is coached by a particular coach (who may be above or below average in personnel decisions). And, face it, some play on teams with different levels of motivation to actually win.

For example, the cases of Andrew Wiggins and Zach LaVine have drawn considerable discussion, due in part to the wide gap between their time on the court and their playing efficiency (as well as this article at FiveThirtyEight, by Neil Paine). The sheer amount of playing time for NBA players of their age is nearly unprecedented, with only 12 players 19 or younger seeing that much court time.

So what to take from that?

• Some of that playing time is just due to circumstances on the Timberwolves this year. Without injuries to Kevin Martin and Ricky Rubio, it is unlikely that either rookie sees anywhere near the playing time they have received.
• The organization clearly does believe in them as players, however, given that there are still other options and the price paid to acquire them in the first place.
• Some of the weight one gives those minutes played depends on one’s degree of belief in Flip Saunders as a coach and talent evaluator. So, there is some internal inconsistency, I think, if one claims both rookies deserve credit for getting the starting nod and then goes on to claim that the same coach has no idea how to use them properly.
• For both Wiggins and LaVine their extreme youth is a much stronger claim to a bright NBA future than counting minutes on the court. There is a long and established record of players tending to improve as they age.

Ultimately playing many many minutes poorly is a double-edged sword. Yes it shows organizational confidence, but it also gives us a much larger and, therefore, more reliable sample of poor play. The best indicator of future production is a record of efficient production, something Wiggins has started to put together in a string of very efficient games recently. If Wiggins does that for a whole bunch of minutes on the court at his age we can make much more optimistic inferences about his future[1.For LaVine it is somewhat encouraging that his biggest weakness appears to be turnovers, which is an likely to improve with experience].