Denver Nuggets 2011-12 PER and usage rates (or, why PER needs context)

This post was inspired by a recent discussion I had on twitter with some of the guys at Roundball Mining Company. Follow the link to their blog, and follow their twitter feeds if you haven’t already.

These days the use of PER is becoming more widespread as a kind of “all-encompassing” metric to evaluate the overall efficiency of players in the NBA. But sometimes it can be misleading.

Short, simplistic version: The fewer minutes and the lower usage rate a player has, the less significant the meaning of his PER becomes. Basketball Reference (the source of these stats) defines usage rate as “an estimate of the percentage of team plays used by a player while he was on the floor”. In other words, the percentage of all possible chances for one player’s team to use the ball that were used by that single player.

So here’s the chart I made:

It should come as no surprise that the trio clustered in the upper right with both high PERs and high usage rates are among those who pass the “smell test” of players who made big impacts.

But more interestingare players like Chris Andersen and Kosta Koufos in relation to Al Harrington, Nene, Danilo Gallinari, Andre Miller and Arron Afflalo. Birdman and KK have higher PERs than the latter five players, but I don’t think many fans would seriously argue they made a bigger impact or helped win more games for the team.

Or in other words, their higher PERs are “devalued” by their lower usage rates, in a manner of speaking.

PER tends to be most useful in comparing star players with big minutes and high usage rates, or at least in comparing players with similar minutes and usage rates. But taken out of context it can be misleading: “Chris Andersen has a much higher PER than Danilo Gallinari.” So when you see PER discussed, always keep the context of minutes and usage in mind. It’s not that the stat itself is “wrong”, it’s just that it can be deceptive when it stands alone.