Assessing NBA’s tricky DPOY race, Point Zion and the importance of beating other contenders

NEW ORLEANS, LOUISIANA - APRIL 09: Zion Williamson #1 of the New Orleans Pelicans is defended by Ben Simmons #25 of the Philadelphia 76ers during the third quarter of an NBA game at Smoothie King Center on April 09, 2021 in New Orleans, Louisiana. NOTE TO USER: User expressly acknowledges and agrees that, by downloading and or using this photograph, User is consenting to the terms and conditions of the Getty Images License Agreement. (Photo by Sean Gardner/Getty Images)
By Seth Partnow
Apr 13, 2021

Quality competition

About a month ago, I described some of my misgivings about how well some of my favored metrics and modes of analysis for overall team strength might hold up in what has proven to be a … shall we say eventful season. Those worries aside, I do think it’s useful to take a lay of the land in terms of where teams have stacked up against various tiers of opposition. Especially in a season with lineups that vary so widely from night to night, there is no perfect way to quickly categorize a game as having been against a top-, middle- or bottom-tier opponent, so we’ll stick with the rubric I’ve used in the past:

Team Performance By Opponent Tier
Team
  
Top Tier Win%
  
TT Net
  
TT Rank
  
Mid Tier Win%
  
MT Net
  
MT Rank
  
Bottom Tier Win%
  
BT Net
  
BT Rank
  
57.1%
3.7
1
80.0%
10.1
1
85.7%
14.4
1
57.1%
3.2
2
62.5%
4.0
8
82.4%
13.4
2
57.1%
2.8
3
65.0%
4.4
7
68.4%
7.2
11
41.2%
2.3
4
66.7%
2.7
10
57.1%
1.4
22
58.3%
1.9
5
80.0%
9.8
2
70.0%
7.2
12
41.7%
1.3
6
33.3%
0.7
12
44.4%
-0.5
25
72.7%
-0.1
7
77.3%
7.1
3
55.0%
2.7
21
36.4%
-2.9
8
72.7%
5.9
5
80.0%
9.4
7
43.8%
-3.3
9
38.9%
-1.9
21
52.6%
2.9
20
50.0%
-3.4
10
23.8%
-8.7
26
40.0%
-1.0
27
35.3%
-3.5
11
47.4%
-0.3
15
73.3%
8.7
6
37.5%
-3.6
12
44.4%
0.0
14
70.0%
7.8
9
25.0%
-4.1
13
19.0%
-9.0
28
47.1%
3.2
19
35.7%
-4.3
14
45.5%
0.5
13
77.8%
8.3
8
36.4%
-4.7
15
56.5%
6.9
4
84.2%
11.7
3
23.1%
-5.3
16
66.7%
5.8
6
84.2%
9.7
4
23.1%
-6.7
17
44.4%
-1.0
17
69.6%
7.3
10
35.3%
-8.7
18
38.1%
-3.4
23
53.3%
-0.9
26
28.6%
-9.0
19
45.0%
-1.2
19
78.9%
6.1
15
23.1%
-9.3
20
47.4%
-1.1
18
68.4%
4.0
17
28.6%
-9.6
21
55.0%
-0.9
16
66.7%
6.4
13
7.1%
-10.2
22
45.0%
-1.9
20
66.7%
6.1
16
0.0%
-10.9
23
25.0%
-8.9
27
42.1%
0.5
24
21.4%
-10.9
24
64.7%
-2.8
22
77.3%
8.8
5
18.8%
-11.3
25
39.1%
-0.9
29
57.1%
-3.9
30
7.7%
-11.9
26
56.5%
3.8
9
68.8%
3.7
18
23.5%
-11.9
27
33.3%
-9.4
30
53.3%
0.8
23
13.3%
-12.9
28
21.7%
-6.1
25
43.8%
-1.8
28
25.0%
-13.0
29
42.9%
1.1
11
75.0%
0.6
14
20.0%
-16.0
30
23.8%
-4.8
24
52.9%
-3.5
29

Top-tier opponents are those on pace to win 50 or more games in a normal (82-game) schedule. Middle-tier squads are those on pace to win between 35 and 50, with bottom-tier teams on pace to win under 35. This has generally broken the league down into thirds across the years.

As I mentioned the last time I discussed this sort of opponent split, only one team in the modern era has won a title without playing to at least a decent record against top-tier opposition: the 2006 Heat, which went 3-11 in those games. Every other eventual title winner was within two games of .500. The 2010-11 Mavs went 10-12, the 2006-07 Spurs were 9-10, and the 2018-19 Raptors 9-9, with the rest of the champs posting winning records against top opponents.

Advertisement

Under normal circumstances, such history would make me a little worried about the Lakers, who are 3-10 against teams on a 50-plus-win pace. But assuming LeBron James and Anthony Davis are healthy a month from now, these Lakers seem poised to be a team that confounds future playoff predictions based on regular-season performances.

The other item of note is that even though Utah has been consistently great regardless of opponent tier, its +3.7/100 Net Rating against top-tier opposition is the lowest “best” mark since the 2012-13 Knicks led the league that season at +1.8/100, and would be only the third time since 2004-05 that no team played at least +4.0/100 against top-tier opposition. I think this is reflective and indicative of what is shaping up to be one of the most wide-open championship runs in recent history.

DPOY problems

Last week, our own Zach Harper went deep on his Defensive Player of the Year leaderboard. I don’t want to go too far down the rabbit hole of litigating who should win a close contest with almost a third of the season still left. But I do want to address part of why arguments about defensive awards can be so tricky: individual defensive stats are messy and hard to interpret, especially with players for whom versatility is a major defensive selling point.

A few weeks back, Daryl Morey appeared on the Rights to Ricky Sanchez podcast, and in the course of answering a question about the state of analytics with respect to individual defense, had this to say: “I would say public-domain, all-in-one defensive measures are all really bad. I think internally, several teams have some pretty OK ones, not good. Defense is hard…it’s so dependent on what the coach wants them to do.”

That last part is key, because even the best “single number” metrics are not measuring how good a player is. Rather they are measuring how effective they have been within the scheme and role in which they have been deployed. This is always a sticking point in the application of “impact” metrics. For example, the three-season Regularized Adjusted Plus/Minus (“RAPM”) for 2017-18 thru 2019-20 available on NBAshotcharts.com is not claiming and should not be used as evidence for Otto Porter being the 11th-best player in the NBA over that stretch, nor was Steven Adams 16th or Robert Covington the 19th-best player. It does mean that they were among the most impactful role players, but that should not be taken as a direct comparison between Porter and James Harden (tied at 11th with an estimated +4.58 points/100 over that span). Further, even across three seasons, a measure like RAPM should be taken much more as a directional indicator than as a ranking. Harden’s .01/100 edge on Kevin Durant means that according to this methodology, Harden is slightly more likely to have had a stronger impact than Durant than for the reverse to be true.

Advertisement

Somewhat fortunately, the difference in player offensive roles is readily understandable enough that it is largely understood that Patty Mills (ninth in 3-year Offensive RAPM) and Devin Booker (10th) aren’t really being graded on the same curve. There are enough examples of solid offensive role players’ efficiency and productivity falling off a cliff when asked to do too much in a larger role that most people avoid direct comparisons across roles. Unfortunately, the same doesn’t really hold for defense. There is no readily understood defensive analog to “Usage Rate” which would allow for a quick and dirty categorization of a player’s defensive role.

There have been a number of noble attempts at creating this kind of rubric based on available defensive data, but I’m not really sold on any of them, especially using public data. The inputs are simply not good enough to allow for the kinds of discernment needed, which is a little of what Morey is getting at in the above quote. Even the data we do have in terms of Synergy defensive play types defended or NBA.com matchup data don’t tell us about defensive schemes in nearly the requisite level of detail.

Getting back to Ben Simmons and Rudy Gobert, this matters when evaluating their defensive impact metrics. Part of the reason bigs are easier to evaluate on defense is first that rim protection is one of the only areas where we do have solid individual numbers. Without belaboring the point (again!) on how little control individual defenders have on the jump-shooting accuracy of opponents, the same cannot be said of shots around the rim, where players like Gobert (as well as Joel Embiid, Brook Lopez, Myles Turner and so on) consistently hold opponents to lower-than-expected rates of rim finishing when they are around the basket. This consistency is driven at least in part by the fact that it isn’t hard to figure out where a dominant defensive big makes their greatest contribution.

The Jazz employ Gobert in a very specific fashion. They know it works, so why mess with it? Nobody told Mariano Rivera to stop throwing cut fastballs, and at least in the regular season, there isn’t any more reason to have Gobert do anything differently. For most other defenders, especially those with the range of defensive skills possessed by Simmons, assigning them to the role where they have the highest individual impact while also allowing as many other players on the squad to do the same is a much bigger challenge. Almost inevitably, a versatile player like Simmons will be better at some things than others, but the difficulty in both defining those roles and measuring the outputs can make narrowing in on the single best spot difficult. Further, a defensive Swiss Army Knife is likely to get used to filling in gaps across a defense, whether or not each of those roles best showcases their skills.

To run back a simplified and completely hypothetical example I used over the weekend, let’s compare Simmons to Covington. At this point, I think it is pretty well understood that Covington is an elite help defender, but mediocre at best as an on-ball stopper. In contrast, Simmons is a similarly elite help defender, but also better than decent on the ball. In part because it’s hard to figure out in which role he is most impactful, but equally because opposing teams present different offensive challenges night-to-night, Simmons gets shuffled between responsibilities a fair amount both from game to game and even within a single contest.

Advertisement

Let’s suppose we knew Simmons was a +4/100 defender in a help role, while only +1/100 in a stopper role. Meanwhile, suppose Covington is +3/100 in a free safety role, but a very mediocre, borderline-replacement -2/100 as a “shutdown corner.” Because Simmons is good at both skills, he ends up playing two-thirds of the time as a stopper, while Covington is used exclusively guarding non-primary scorers. On average, Simmons would only appear to be a +2/100 defender, while Covington would show up as +3/100 despite being inferior to Simmons in either possible role.

I would submit this has happened to a degree:

DRAPM Rank Comparisons
SeasonGobert DRAPM RankSimmons DRAPM RankCovington DRAPM Rank
2013-14
58
-
201
2014-15
27
-
91
2015-16
17
-
69
2016-17
1
-
2
2017-18
5
47
1
2018-19
3
252
12
2019-20
24
108
101
2020-21
1
51
232

We only have somewhat reliable matchup data going back the last three seasons, but Covington has to guard primary offensive options much more frequently in Portland than he did in past years. Over 30 percent of his time on defense has seen him matched up against players with usage rates of 25 or higher, as compared to the 21-23 percent range he’d seen in the previous two seasons. While this is an exceedingly blunt manner of assessing defensive role, and the data isn’t nearly sufficient to prove causality, it is at least suggestive. This has been, by far, Covington’s least effective defensive season according to DRAPM since he became a regular rotation piece in his second year — at the same time he has been asked to do things outside his best role more frequently than in recent years.

Meanwhile, Gobert keeps chugging along with DPOY-level impact season after season because stationing a tall, agile and angrily aggressive dude near the basket is as effective today as it was in the days of Russell and Chamberlain. Does this make him a “better” defensive player than Simmons? On some level, that’s an epistemological question, but I will say that having perfect clarity on how to most effectively deploy a player has to be a positive in terms of ensuring they turn their talents into impact.

Closing with Zion

While the New Orleans Pelicans are battling for their playoff lives, sitting fifth among the six teams battling for the Western Conference play-in spots, in one very important long-term sense, they can already consider this season a success. We’ve written in the past about the absurdity of Zion Williamson’s ability to get to the basket at a rate nearly 50 percent higher than anyone in recent league history, but for a graphical refresher:

To put the size of the gap between Williamson (20.1 FGA/100 from 5 feet and in) and Montrezl Harrell (second at 14.1/100) in perspective, among the 382 players with at least 1,000 such attempts since 2004-05, the difference between first and second is as large as the space between Harrell in second and Andrew Wiggins in 142nd!

But a more important development for the Pelicans and coach Stan Van Gundy this season has been the realization that not only can Williamson be a battering ram of a scorer the likes of which the league has never seen, but also his ability to handle the ball and make plays for others means that he’s the sort of ball-dominant closer championship-level teams are always searching for. New Orleans’ rough start to the year was characterized by a number of close losses with a sputtering fourth-quarter offense.

However, month by month, the end-frame offense has shifted away from Brandon Ingram and towards Williamson. According to tracking data provided by the NBA Advanced Stats group, Williamson has become the main ballhandler more and more often in fourth quarters for New Orleans:

Pelicans' 4th Q Time of Possession %'s
Player
  
Dec
  
Jan
  
Feb
  
March
  
April
  
19.1%
11.9%
12.5%
11.1%
15.0%
15.5%
15.1%
7.0%
7.9%
10.9%
5.7%
3.7%
4.9%
5.5%
7.0%
17.7%
16.3%
14.7%
12.7%
13.1%
16.6%
16.1%
14.8%
15.6%
20.0%
11.1%
13.0%
11.6%
11.2%
14.0%
6.2%
6.2%
9.7%
16.1%
18.9%
Clutch ORTG
61.5
105.8
110.7
123.5
118.2

While Ingram’s time on the ball in the fourth has bumped back up recently, that likely has as much to do with the rash of backcourt injuries New Orleans has dealt with in the last week or so than a concerted effort to give him the ball more at Williamson’s expense. In the relatively limited sample of “clutch” situations in these fourth quarters, the Pels offense has performed better with Williamson at the helm. Meanwhile, Williamson’s own fourth-quarter numbers have jumped substantially:

Zion in the 4th Quarter by Month
MONTHGPMPGPPGAST%UsageeFG%TS%
December
5
7.4
3.8
5.3%
24.2
40.0%
50.1%
January
12
8.4
5.7
18.5%
25.7
55.6%
61.2%
February
15
8.5
5.9
28.4%
24.2
65.2%
71.0%
March
11
8.9
7.5
31.0%
29.7
62.8%
68.0%
April
3
8.9
10.0
22.2%
35.4
68.4%
72.3%

I feel safe in saying New Orleans won’t win the title this year, but the steps taken by the Pelicans’ young star should make them feel like they’ve gotten closer to that long-term goal.

(Photo: Sean Gardner/Getty Images)

Get all-access to exclusive stories.

Subscribe to The Athletic for in-depth coverage of your favorite players, teams, leagues and clubs. Try a week on us.

Seth Partnow

Seth Partnow provides NBA and basketball analytics for The Athletic. He resides in Milwaukee and was formerly the Director of Basketball Research for the Milwaukee Bucks. Follow Seth on Twitter @sethpartnow