:format(webp)/cdn.vox-cdn.com/uploads/chorus_image/image/18564173/20130606_kkt_st3_441.0.jpg)
Earlier this summer, after J.R. Wilco posted this optimistic piece about Kawhi Leonard, AaronStampler wrote a sobering response about why Leonard was not a star just yet, and he used one of the most widespread composite stats in basketball, John Hollinger's PER, to make a point:
Most disconcerting of all, Leonard's PER actually went down his second season, and it's not like the 16.6 he posted as a rookie was world-beating.
Compare his numbers to the first years of the Big Three and it's just not a pretty picture. Tim Duncan's PER improved from 22.6 to 23.2 in his second season, and he's perhaps the worst example of the three because he nearly came to the pros as something of a finished product after four years at Wake Forest. Ginobili had a far more dramatic improvement, from a 14.7 PER as a role-playing rookie to 18.5 his second season to a full-fledged star by 2004-2005, putting up a 22.3 mark. In terms of PER anyway, Ginobili's prime extended far longer than the average fan would guess, for a full eight seasons up to his age 34 year, as he put up something between 21.7 and 24.3 each year. Tony Parker's probably the best guy to compare Leonard to because they both came into the league so young, but Parker improved his PER from 11.7 to 16.5 his second season.
You should go read it in its entirety, if you haven't already. I found that part especially interesting so I decided to take a look at Kawhi to see if his low PER was a cause for concern. When I showed him my findings, J. Gomez decided to offer his take as well.
Disclaimer: PER's flaws have already been explored by people smarter than us, but we wanted to figure out why, specifically, Kawhi's was low and if it meant anything.
How is PER calculated?
You can find the actual equation used to calculate PER here, but a simpler, if a bit different, explanation is this from Bleacher Report.
Now let's use the formula to show how Kawhi stacks up against their own teammates:
PSherman42: In almost all the categories where a player beats Kawhi, it's by a fairly large margin -- especially field goals made, where Tony has almost twice as many field goals as Kawhi. Leonard actually beats Manu in nine of the 12 categories.
J. Gomez: It makes sense for Kawhi to have low numbers in some categories as he wasn't relied upon to create for his teammates.
PSherman42: We will get back to that, but that's not its biggest flaw, this is:
PER doesn't measure defense
PSherman42: The only two defensive stats that are taken into account are steals and blocks, something which even John Hollinger, the creator of PER, has acknowledged as a flaw:
Bear in mind that this rating is not the final, once-and-for-all answer for a player's accomplishments during the season. This is especially true for players such as Bruce Bowen and Trenton Hassell who are defensive specialists but don't get many blocks or steals.
There are players that get big steal and block numbers (like Monta Ellis and Deandre Jordan) despite being subpar overall defenders. Think about that for a second: PER actually rewards a guy like Ellis on the defensive end more than Kawhi Leonard.
J. Gomez: After reading that, I went and looked at Bruce Bowen's career PER. It was 8.2. What a scrub. So the contributions of low-usage, defensive-minded players, which is what Kawhi was last season, will never be appropriately gauged by PER.
PSherman42: I'm glad you mentioned usage because that is the second biggest reason why Leonard's PER is underwhelming.
PER doesn't take usage rate into account
PSherman42: Usage rate is an important statistic for determining how many possessions are run through the player. The significance of usage rate with respect to PER is that if a player spends a lot of minutes on the court mostly because of his defensive work, he will be a low usage player and therefore won't have a lot of field goals made, three-pointers made, assists, or free throws made, which are the primary contributors to PER.
Kawhi's usage rate of 16.4% makes him the 245th most used player offensively in the NBA last season, and eight out of the nine Spurs that played more than 1,000 minutes last season ranked higher. This essentially means that Kawhi seldom had the ball in his hands on the offensive side to significantly increase his PER
J. Gomez: It really does seem like a high usage rate is required to have a high PER. I checked for players having less than 20% usage rate that had a PER above 18, which is considered to be indicative of a good second option, and only 12 players popped out: eleven big men (including our own Tiago Splitter) and Jose Calderon. So it seems that in order to have a high PER as a role player you either need to rebound like crazy or have ridiculous assist to turnover ratio.
PSherman42: It is important to note that a player's PER can also be negatively impacted by an increase in usage rate. This is because an increased usage rate should lead to more turnovers, missed field goals, missed free throws and fouls -- as well as all of the good things. In theory, one would expect a player's PER to trend upwards as his usage rate increases, but Kawhi showed otherwise in his second season as a Spur. While his usage percentage saw a relatively significant, almost two percent, increase, his PER got two decimal points worse.
J. Gomez: But since the decrease wasn't substantial, you can make the case that Kawhi actually did get better, as his role increased, but his overall efficiency didn't.
PSherman42: Exactly. Another, perhaps better, example is Paul George. After his 2012-13 regular season and playoff performances, George is now seen by most as a bona fide star. Much like Kawhi's stats, however, George's numbers say otherwise. George's 16.8 PER isn't indicative of a star, and he got it with a much higher usage rate (23.5%) than Kawhi.
J. Gomez: But because he got that PER while having to carry the offensive load for a bad offensive team, we should take his PER with a grain of salt. The Pacers lost Danny Granger and they needed someone to become their first offensive option, so George's usage percentage (and minutes played) took a huge leap last season, which caused his shooting numbers to dip while his turnovers spiked. So the fact that George's PER saw a small spike last season, even under those circumstances, suggests he is a very good player.
PSherman42: Yeah, the number isn't everything. Let's use Andray Blatche as an example. Blatche has a usage rate of 26.5% and a ridiculous PER (21.9). Under Hollinger's reference guide, this would actually mean that Blatche is in between "Borderline All Star" and "Bona Fide All Star." Is there a more damming argument against PER than that?
J. Gomez: Well, that might be the opposite case from George. Blatche went from being the Wizards' offensive centerpiece, to logging back-up minutes for the Nets and benefiting from the attention his teammates got. Not to mention playing against reserves instead of starters. That, combined with PER's lack of accounting for his still subpar defense, is very likely what allowed Blatche to achieve such a stellar PER. But any rational observer would take George's 16.8 PER while playing great defense for 37 minutes per game over Blatche's 21.9 PER while playing 19 minutes of mediocre defense off the bench on a team that was eighth in the league in offensive rating.
So, should we be worried about Kawhi's PER?
PSherman42: No. PER occasionally has its uses, but comparing PERs is often nothing more than comparing points per game; it's irrelevant most of the time because different players play different roles on different teams. Last season, the PER statistic did not benefit Kawhi because of the way the stat is calculated and multiple injuries that made him miss 24 games. However, as the seasons progress and the Big Three take smaller roles, we should expect to see Kawhi's PER skyrocket to a much higher level than his mere 16.4 as more possessions are run through him.
J. Gomez: I don't know about his PER skyrocketing with more touches; after all, it didn't happen for Paul George. But at least we will have a better sense of how good Kawhi can be as a first or second option when he is actually asked to fill that role. For the time being, I'll find solace in the George vs. Blatche examples you presented. It seems clear that you can be the best player on a good team and still not see your value accurately reflected by PER.
In the final analysis, judging Leonard's numbers against the PER of the Big Three during their second seasons -- when they were a big part of the team's offensive attack -- doesn't seem to be a relevant measure when determining Kahwi's capabilities.