Commentary: PER Underrates Kobe Bryant
John Hollinger, the stats guru and creator of the Player Efficiency Rating, has a new article on ESPN where he claims that Manu Ginobli's actual production is similar to that of Kobe Bryant's. You can see Hollinger's thought process—he's always thought that Ginobli is a superstar and what better way to prove it than compare him favorably with Bryant?
I don't dispute that Manu is a superstar. With apologies to Brandon Roy, it was clear that Ginobli should've been an All Star this year. Duncan is still the Spurs' best player, but Manu is their heart and soul.
The problem I have with Hollinger's analysis is that it purposely obscures the facts to make Ginobili seem more productive than he is vis-a-vis Bryant. Which, in turn, makes Bryant seem less productive than he is compared to other superstar players.
I am particularly concerned because so many basketball fans rely on the Player Efficiency Rating and John Hollinger's analysis. It is concerning because Hollinger makes no attempt to mention the limitation of his methodology or formula in his articles. PER, as with every stat, is useless without context.
So here I attempt to explain why PER underrates Kobe Bryant.
Player Efficiency Rating (PER) is a stat designed to measure a player's impact per minute. It uses a variety of criteria like rebounding rate and true shooting percentage to determine just how effective a player is when he is on the floor. One of the most important variables is pace. Hollinger adjusts for pace so that players playing on a fast pace team aren't given a greater advantage over players that play on slower paced teams.
Hollinger uses pace as his primary argument for why Ginobili produces at a similar rate as Bryant. "Wait, it gets better. Kobe plays on one of the league's faster-paced teams; Manu is on the second-slowest. Plug Bryant's numbers into the Spurs' sluggish pace and you get 26.9 points, 5.9 boards and 5.2 assists, making the comparison a dead heat," Hollinger writes.
Normally, the stat works as it should. After all, most teams play at roughly the same pace throughout the entire game and run the same offense. Some teams, however, run a completely different offense and play at a different pace during different points of the game.
Take the Los Angeles Lakers for example. The Lakers starting unit at the beginning of the season was Derek Fisher, Kobe Bryant, Luke Walton, Lamar Odom and Andrew Bynum. This unit ran the triangle offense, which is predicated on ball movement and a series of player movements based on the position of the ball. Indeed, the triangle is a half court offense and running it effectively takes time.
Contrast that to what has happened for most of the season beginning at the end of the 1st quarter. Coach Phil Jackson benches all of his starters except for Lamar Odom. Jordan Farmar, Sasha Vujacic, Ronny Turiaf and Trevor Ariza come into the game and Kobe sits for about six minutes. As Phil Jackson recently revealed, with the second unit on the floor the Lakers run a completely different offense similar to the Phoenix Suns'. It is run and gun designed to push the ball in transition and take shots early in the shot clock.
Until recently, Kobe Bryant had not been playing many minutes with the second unit. Prior to Ariza falling victim to injury, Lakers' Coach Phil Jackson would play the bench as a unit with Bryant sitting out.
The problem with Hollinger's argument is that PER is based on the pace of the team and not based on the pace when the player is actually in the game. To explain the concept another way, while running the triangle, the Lakers might be on pace for hypothetically 88 possessions per game. The second unit comes in, pushes the pace and now the Lakers play at a pace of 96 possessions per game. Obviously, if Kobe's playing on a unit that is playing at a different pace than the second unit, PER distorts his numbers to make him look worse than he actually is.
Another way to think about this is that 28/6/5 in 38 minutes with his true shooting percentage looks much better on a team that plays at a pace of 88 possessions versus a team that plays at a pace with 96 possessions.
Think I'm making this up? When Ariza went down, Bryant was 10th in PER. Now that he's playing more minutes with the second unit, his PER has shot to 7th in just 4 weeks. Indeed, pace heavily influences PER and one has to wonder what Kobe's actual PER is without this flaw.
Hollinger does nothing to account for this obvious discrepancy. Ginobili on the other hand, tends to speed up the game for the Spurs when he is on the floor but gets the benefit of playing on a slower paced team. Ginobili's PER overall is reflective of his per-minute contributions since he, alone, doesn't increase the pace of the Spurs' game significantly. Still, PER severely underrates Bryant and any analysis that tries to compare the two players based on PER and pace is bound to be inaccurate.
Obviously, the pace problem has larger implications because when you compare Bryant's PER to other elite players, it won't look like he is producing as effectively while on the court when in actuality, he is. It is a shame that Hollinger promotes his player efficiency rating without a discussion of the limitations of his formula.
While I think PER is a incredible statistical tool and useful in evaluating players, it is only useful when examined in the proper context.
Nate Smith is a regular contributor to Bleacher Report. He does a mailbag every Monday. Please send your basketball related questions or comments to NateSmithBR@gmail.com.
What is the duplicate article?
Why is this article offensive?
Where is this article plagiarized from?
Why is this article poorly edited?