Friday, July 15, 2011

Efficiency Experts

If you’re anything more than a casual fan, you are likely aware of the ongoing debate between proponents and detractors of the use of so-called advanced metrics in NBA analysis. Art vs. Science. Hell, sometimes it feels like Science v. Religion.

In one corner we have the willfully ignorant old guard, entrenched in the position that sports have long had all the statistics necessary to facilitate effective analysis. This camp will argue that the proliferation of advanced statistics “takes the fun out of the game” and that no amount of numerical data can take the place of actually watching the action. Keep all open flames away from those straw men, fellas.

Meanwhile, there are the rest of us. Fans, bloggers, most team executives, etc, that have welcomed advanced metrics into our basketball lives. In my case, I’ve always enjoyed looking for meaning in large pools of seemingly disparate data. Thanks to a number of outstanding sites (’s Stats Cube, HoopData, Basketball Reference’s Play Index, and BasketballValue, just to name five), such immense pools of NBA data are now readily accessible and waiting to be filtered, spliced, compared and analyzed. This, combined with the aforementioned affinity for numbers and a possibly unhealthy fetish for NBA history, has provided me with the perfect outlet.

I’ve mentioned this before, but it bears repeating. This is born neither of a hatred for the viewing of basketball (as my wife will happily, or not, attest) nor- to the best of my knowledge- a subconscious fantasy of turning the NBA experience into some soulless computer simulation. The objective of advanced metrics is to shed light on the granular components of statistical achievement. We are now able to use numbers to paint a more descriptive picture of a player’s career, and the individual seasons of which it’s comprised. Applied properly, these statistics allow us to look past the what, 
and examine the how. 

For instance, it should surprise no one that of the 25 highest single-season Usage Rates (percentage of team’s possessions ending in a shot, turnover or free throw attempt by a given player; only tracked since 1977-78), Michael Jordan (6), Allen Iverson (6), Kobe Bryant (3) and Dwyane Wade (2) are responsible for 17. Shoot-first alpha dogs all. The remaining seven? LeBron James, George Gervin, Dominique Wilkins and Jerry Stackhouse, along with two apiece for Tracy McGrady and Bernard King. Even at this basic a level, these stats certainly reveal a “type.”


In today’s NBA, simply accumulating numbers by any means necessary is no longer acceptable. Raw statistics are taking a backseat to (or at least having to share the front seat with) those that tell the story of how exactly the numbers are tallied. It’s unlikely that there will ever be a “silver bullet formula” that perfectly accounts for all aspects of the game and spits out a single figure by which every player’s game can be judged on a level playing field. To that end however, we do have Player Efficiency Rating, or PER, developed by ESPN’s John Hollinger, which aims to “sum up a player's positive accomplishments, subtract negative accomplishments, and return a per-minute rating of a player's performance."

As advanced metrics have ascended to a more prominent place in the basketball world, they have elevated the status of certain concepts and terms with them. More than any other, the idea of “efficiency” has gained prominence in the area of player assessment. That a guy scores 25 points per game is all well and good, provided he doesn’t waste too many possessions doing it. 12 rebounds per game? That’s great. Well, maybe. How many rebounds were “available” to him?

While it is in fact a very good gauge of on-court productivity, PER ironically falls short in measuring, wait for it, actual efficiency. This is not an indictment of the stat itself, but rather an argument advocating its use in concert with other metrics in order to paint a more complete picture. Since the 1977-78 season, 50 players have played 400+ games and posted a cumulative PER of 20 or higher. Topped by Michael Jordan (27.9), followed by LeBron James (26.9), Shaq (26.4), David Robinson (26.2) and Dwyane Wade (25.7), the list is not overflowing with surprises.

However, when we look deeper into the legendary role call, we can identify the NBA’s “Efficiency Experts” of the past 35 years. Which brings us me to what I am finding to be an excellent measure of true efficiency- the ratio of PER to Usage Rate. In each of the aforementioned 50 instances, the player in question produced at an elite (or near-elite) level. This is fact. What we can deduce from PER/USG, however, is which stars burned the fewest (figurative) calories en route to their numbers. Here, for your consideration, is that list, beginning with the stars that got the most bang for their possession buck:

And for those of you that like your numbers represented by geometric shapes of varying color:

A few takeaways:

The numerical particulars. The average PER for this sample is 22.3, with MJ topping the list, and John Drew (of late-70s/early-80s Hawks and Jazz fame) coming in at an even 20. As for Usage Rate, MJ leads the way again at 33.3, with John Stockton 50th with a rate of 18.9. The average career PER/USG for these 50 players is .8540, with a standard deviation of .1083. This suggests that the half dozen guys with a PER/USG greater than 0.9623 were/are the most efficient of the lot, while the eight below .7456 represent the least efficient of the bunch.

Still no silver bullet. This calculation is hardly perfect. It skews heavily in favor of power forwards and centers, who combine to account for 20 of the top 30 spots, but just six of the last 20. This makes sense. Bigs typically do not handle the ball a great deal and rely heavily on offensive rebounds (a PER plus) and the whims of their perimeter teammates to receive it- in scoring position more often than not. Given this, it’s not terribly surprising that shooting guards and small forwards, of which there are 19 on the list, but only five in the top 30, are punished by the metric. These players tend to handle the ball a great deal, are not as active on the offensive glass and attempt a greater number of lower percentage shots than big men. It thus makes sense to isolate the wings players and compare their exploits within that sample.

Larry Bird: king of wings. Sort of. In terms of pure basketball aptitude, Kobe Bryant and Michael Jordan are pretty firmly cemented in the top two spots in NBA history. However, in terms of efficiency- the kind we’re talking about here- The Legend sits at the head of the SF/SG class. His formula was simple. A virtually automatic outside shot- the result of thousands of hours of work. He also produced some fantastic rebounding numbers, averaging at least 8.5 per game in every season in which he played at least seven games, with a 6+ Offensive Rebound Rate in each of his first seven seasons. Larry Bird’s spot atop the list of wing players considered here is more confirmation than revelation. Not confirmation that Bird was a all-time great basketball player, but that our numbers aren’t batshit loco. Everyone that watched him play in his prime can agree on that.

A couple of other observations on the collection of wing players below. First, I found it surprising that Manu Ginobili came in second, but in thinking about it I realized I was hard-pressed to think of another guy that consistently has a bigger impact in limited burn. Manu’s played just 28 minutes per game for his career, during which he’s posted a True Shooting Percentage of 58.1%, made his free throws (83.4%; 86%+ each of the past five years) and consistently ranked in the top 10 in Steal Rate.

Turning our attention to the bottom of the list, just ahead of John Drew we have Carmelo Anthony, Allen Iverson, George Gervin, Dominique Wilkins and Vince Carter. All spectacular talents, and anyone that knows me can attest to my lve of A.I. and ‘Nique, but… yeah. You’re not surprised that these guys are where they are, are you?

Lead guards: quality over quantity. At the other end of the positional spectrum we have point guards. Point guards, who tend to have lower Usage Rates than players at other positions despite handling the ball a ton (assists do not constitute “usage”), occupy an especially interesting place in this group. Despite representing just 10% of the overall list, all rank in the top 13, with the three highest PER/USG rates since 1977-78 belonging to lead guards.

John Stockton epitomized efficiency. Given his demeanor and style of play, it comes as no surprise that advanced metrics cast a favorable light on his body of work. He averaged just over nine field goal attempts per game (making 51.5%, including 38.4% from 3), hit his free throws, 16 times ranked in the top 10 in Steal Rate (sixth all-time) and turned the ball over less than three times per game, despite having it glued to his hands for 1,504 games over nearly two decades. Still, I must admit to a little bit of surprise, as heading up a list like this is kind of a big deal. I guess that John Stockton though- stealthily efficient.

However, more (pleasantly) surprising than seeing Stockton at #1, was the appearance of Kevin Johnson at #13. I am no Suns fan- and KJ was central figure in breaking my heart in the springs of 1990 and 1993- but I have long been of the belief that Kevin Johnson is a top-12 (maybe top-10) all-time point guard. In 11 NBA seasons, he averaged at least 19.7 points and 9.3 assists six times, 20- 10 three times and averaged 10+ assists four times. However, despite his role in the Suns’ run in 1990s, a multi-pronged job title which included “prolific scorer,” he averaged more than 15 shots per game just once in his career, never had a Usage Rate above 25, was always near 50% from the field (49.3% career; 58.5% TS%), was a regular at the free throw line (it helps when you can slice into the paint at will), where he made 84.1% of his attempts.

At his best, Kevin Johnson was to opposing fans what Chris Paul is today- devastating and breathtaking. The kind of player you hated even more because he made you love him while he was kicking your ass. I am a big believer in leaving preconceptions at the door and letting the numbers lead the way, but I’ll admit that I’ve been hoping for an advanced metric that would give KJ his due. I’m glad we did this.

These are just a few of my thoughts. Looking at so many players, each with a unique skill set and resume, across so many eras, including the appropriate level of analysis for each would be prohibitive in terms of both word count and time (both mine and yours). The topics covered herein will be revisited, as will others inspired by the work done here. In the meantime, however, I welcome, and look forward to your observations, feedback, effusive praise and book proposals.

1 comment:

jollyrogerwilco said...


This bit of roundball manna from heaven demands multiple readings.

Well done, good sir.