The data analytics revolution in sports has taken on a life of its own. In basketball, the creation of PER (Player Efficiency Rating) brought the blessed relief of one statistic which is a synthesis of all the others. Since its introduction, PER has become probably the single most used “argument ender” statistic. While I like PER’s efficacy for individual players, I wondered how meaningful it is when applied to teams.
To get some sort of answer, I decided to apply PER to the Toronto Raptors, and see where it leads. My thesis is simple: a winning team should have a higher PER than a losing one, otherwise PER is of dubious value. I added up all the PERs of the ’13-’14 Raps’ starters, then divided the result by five. What’s our average starter’s PER? 16.46, which is barely above 15, the number which the stat’s creator tells us is what the most average player in the NBA attains. The closest last season was Boston’s Brandon Bass, at 15.01. The closest mark to our average starter was the 93th highest-ranking player, the Knicks’ Tyson Chandler, at 16.48.
A New York Knick, huh? Let’s use them as a comparison. I couldn’t do an exact one, as the Knicks were roiled by injuries last season, so I did what I thought was the next best. I selected their five by minutes played, which not surprisingly corresponded closely with the “games started” stat. That gave us Carmelo Anthony, J.R. Smith, Raymond Felton, Iman Shumpert and Tim Hardaway to look up and divide. Well! The Knicks’ “starters” (Hardaway was the outlier here, playing the fifth-most minutes while entering 81 games, yet starting only one) had a PER of 14.77.
I confess I don’t know what to make of this result. A few tentative conclusions come to mind. (Let’s remember the Raptors won the Atlantic Division with a 48-34 record, while the Knicks needed to end the season on a 7-3 spurt just to finish at 37-45.) If our virtual Raps starter was just a smidgen better than the league’s average player, our bench must have been terrific. PER doesn’t work for teams. The Knicks’ coaching wasn’t sound.
We’re not through with this topic. In the meantime, I’d enjoy hearing from you, Rapture Nation, about this admittedly unscientific comparison. I’m particularly interested in the thoughts of PER fans, and those of you who think my methodology was poor (don’t just say that – tell me how).