Evaluating NBA Perimeter Defense: Establishing and Using a Rubric

Insert player is a plus perimeter defender. So and so really does a lot on defense. His off-ball activity is great. This guy offers some great rim protection. Player X is a traffic cone. He adds value as a team defender.

Chances are, you’ve heard or read one of these comments on your favorite podcast or NBA site. I even find myself falling prey to these simplistic and comforting cliches, but I rarely hear anyone interrogating what any of these mean. We grappled with this very question on a recent episode of Sense and Scalability, and Mark and I discussed this on the Premium Hoops Podcast. However, it’s too large of a topic to fully address in an hour.

My goal today is simple: establish a rubric for judging perimeter defenders, use it to evaluate famously (and notoriously) good/bad defenders, and look for specific differences. I want to be clear that I do NOT intend to definitely solve the quandary of scouting and evaluating perimeter defenders. I intend only to start the conversation to be pushed forward elsewhere.

Perimeter Defense Rubric

The following is my first attempt at establishing a rubric for evaluating NBA perimeter defense.

2OutstandingOh Shit!” plays. Either saves a basket while covering from one’s defensive job or defies human athleticism to make a play. It may also include “Pick-6” steals that lead to 1-0 transition plays
1GreatDefensive plays that clearly adds value to the team. Could be a basic secondary block at the rim, any block, a steal at the nail, an off-ball steal, drawing an offensive foul, or a similarly valuable play.
0.5GoodDefensive plays that clearly helps the overall defensive scheme without really disrupting an offense. This might be a basic deflection, a strong switch, or a strong help contest at the rim.
0League AverageDefensive plays that I would expect any functional defensive player in the league to do. This includes keeping a player in front, contesting shots, and making appropriate rotations.
-1DetrimentalDefensive lapses that lead the other team to gaining an advantage. Perhaps a clear blow-by, a missed rotation, being unable to help at the rim, or a bad gamble.

Using the Rubric to Evaluate Perimeter Defenders

To test out this rubric, I decided to look at opposite ends of the spectrum: three players who are commonly known as excellent perimeter defenders and two players who are known as sieves. I also used Basketball Index’s D-LEBRON metric to ensure that their defensive impact statistically matched what the NBA-Twitter intelligentsia saw with their eyes.

For the dominant perimeter defenders, I settled on DPoY darling Ben Simmons (+1.56 D-LEBRON as of 3/31/21), Matisse Thybulle (+2.02), and Jimmy Butler (+1.63). On the other side of the equation, I evaluated Damian Lillard (-2.65) and Trae Young (-2.18). (Note: I decided to scout Thybulle after writing a first draft, so I’ll discuss him separately later).

Before diving into the results, I want to be perfectly transparent: defensive evaluation is extremely difficult. Not only does the evaluator need to know the defensive system, but they must also account for the player’s role in that system. Because of this, I cannot stress enough that there will always be some level of subjectivity. I’ll show some video evidence of what I’ve found, but even then, that’s just the tip of the iceberg.

As I watched them play, I marked down how many plays they made on the rubric. Then, I looked at the number of minutes they played in that period so that I could determine the number of plays per 36 minutes to better normalize the data. Finally, I combined the data from the players either ends of the spectrum so that I could more quickly gather more data. Ideally, I would just use one player’s data for himself, but for the sake of rough estimation, I combined them.

Evaluating Trae Young’s and Damian Lillard’s Perimeter Defense

I began my evaluation with the two worse defenders because I wanted to set a baseline definition of “bad defense.” What popped off the tape for me was the number of -1 or “Detrimental” plays that both Young and Lillard committed. Specifically, both struggled with blow-byes where the defense hinged on either of them keeping a defender in front. By either closing out sloppily or simply not moving their feet quickly enough, they surrendered an advantage that might have led to opponent points (note that I’m looking at what could have happened here as opposed to a results-based analysis).

Neither defensive lapse led directly to a made field goal, but both produced advantageous situations for the offense that could’ve impacted the outcome of the game on a per-100 basis. Sometimes, these lapses generated easy offense for the offense, but again, these are attributed an equal “-1” because I’m avoiding just looking at results.

One critique here might be that Lillard is supposed to be insulated from post-ups, so this is outside of his expectations in the Blazers’ system. On the contrary, the Blazers often switch perimeter defenders, so if Lillard were replaced with a better post defender, their overall defense would improve.

Besides the increase in “Detrimental” plays, Young and Lillard contributed a dearth of positive plays. Between them, I counted very few “Great” plays and zero “Outstanding plays.”

Lillard and Young’s Numerical Valuation

During this part of my analysis, I was still developing the +0.5 or “Good” defensive plays, so I didn’t mark those down. I also didn’t tally the number of 0 or “League Average” plays because I thought my time would be better spent counting other kinds of plays.

Overall, here are the results of using the perimeter defense rubric on Lillard and Young.

Detrimental Plays/36 minutesGreat Plays/36
Lillard/Young4.241.18

Since both players average about 36 minutes per game, this translates to 4.24 Detrimental plays per game and around one “Great” play every game.

Evaluating Ben Simmons’ and Jimmy Butler’s Perimter Defense

Just like most theories, the conclusions are mildly boring. However, the nuance is what I’m interested in. Neither Simmons nor Butler were devoid of defensive errors, but they occurred at a much lower rate. This might be the only true blow-by that I saw, and some may quibble that it’s not a true blow-by. Regardless, Wiggins clearly gains an advantage because of Simmons’ defense.

Likewise, these two defensive mistakes from Butler clearly lead to easy baskets. In the former, Miami’s scheme relies on Butler switching onto Sabonis, but Sabonis takes advantage. In the latter, Butler pauses for a moment giving up an open lob for McDermott.

Again, these three plays constitute a much higher percentage of the total “Detrimental” plays that either player made compared to the Young/Lillard section.

Furthermore, Simmons and Butler generated far more +1 “Great” plays but few +2 “Outstanding” plays. Butler drew a couple of offensive fouls which one could be interpreted as equivalent to a block at the rim, and both stripped opposing ball handlers.

Where this differentiation becomes difficult is in a pick-6 situation when a steal or defensive play produces an efficient transition possibility for the offense. For example, take these two passing lane steals.

In the first play, Butler makes a good off-ball read, but it seems like more of a lazy pass. Furthermore, he scores by using his guile to beat the one-on-one break. In the latter, Simmons clearly creates the steal through his quick lunge which generates an easy 1-on-0 break. To me, I’d be willing to mark Simmons’ play as a +2 “Outstanding” steal, but I’d be hard-pressed to do the same for Butler’s +1 “Great” steal. The line of demarcation between the two is where this exercise becomes messy.

Butler, Simmons, and Thybulle’s Numerical Valuation

Here is the same table as above with Butler, Simmons, and Thybulle added to it.

Detrimental Plays/36 minutesGreat Plays/36Outstanding Plays/36
Lillard/Young4.241.180
Butler/Simmons/ Thybulle1.642.360.36

It bears repeating that the “per 36” in terms of these four players is easily translated to “an average game” for them.

Matisse Thybulle’s Perimeter Defense

You’ve been reading long enough, so I’ll throw you a hot take: Matisse Thybulle is the best perimeter defender in the NBA. To me, it’s only hot because I haven’t watched all other perimeter defenders in the league and not because I think that Simmons is better. I want to be clear: Thybulle is a more impactful perimeter defender than Ben Simmons. Alex Fidanza pushed back on Twitter, so I want to address why this isn’t so hot.

Based on smallish sample of hand-tracking, Thybulle made more “Outstanding” and “Great” defensive plays while essentially making “Detrimental” plays at the same rate. This aligns with a comparison of Basketball Index’s player profiles.

Since my hand-tracking matches the statistical outcome, I’m confident in calling Thybulle better. However, Alex’s major point is that he trusts Simmons more to lock down a specific player. Perhaps that is true and perhaps it’s true across multiple positions, but I’m convinced that the difference between the two in that situation is effectively negligible.

Furthermore, why wouldn’t Thybulle be a better defender? Simmons has to exert significantly more effort on offense, and it’s simply not possible to go equally hard on defense. Maybe Simmons would be better in Thybulle’s role, but that’s what what we have to analyze. Much like Tony Allen a decade ago, a defensive specialist should be just that: signficantly more impactful on defense.

Making Sense of All of This

The simplest conclusion here is that great perimeter defenders make significantly fewer “Detrimental” plays and more “Great” and “Outstanding” plays than poor perimeter defenders. While that’s true, I think the more interesting conclusion lies in the magnitude.

Detrimental Plays/36 minutesGreat Plays/36Outstanding/36
Lillard/Young4.241.180
Butler/Simmons/ Thybulle1.642.360.36
Butler/Simmons/ Thybulle Compared to Lillard/Young1/3ish Fewer MistakesTwice as many Great PlaysINFINITY!

Simply put, good perimeter defenders make fewer mistakes and more impactful plays. However, the amount of “Great” and “Outstanding” plays is probably smaller than you’d expect by the way that defense is talked about. The key may lie within the “Detrimental” plays or even “Good” plays which I’ll discuss in a moment.

Circling back to D-LEBRON, this tracks with the per 100 valuations that the metric provides. According to the statistic, Simmons/Butler provide about 1.5 points/100 possessions with their defense while Young/Lillard detract about 2.5/100. Since I want to bring this number as close to a per 36 as possible, I’ll translate those two numbers to +1.125/75 possessions and -1.875/75. That’s about 3-points per 75 possessions (or per 36 minutes or per-ish game) that a great perimeter defender might contribute over a bad defender according to D-LEBRON.

I don’t want to speculate on the actual numerical values here, but that seems like a reasonable assertion. I doubt that the difference would be that much greater. If nothing else, I could see great perimeter defenders providing around 2 points/75 possessions and poor ones (especially point guards) topping out around -1.5.

Estimating Great/Outstanding Perimeter Defense Plays

Based on my criteria, a simple way to estimate the number of “Great” and “Outstanding” plays that a player makes is by adding his steals, blocks, and charges drawn per 36 minutes. Overall, this pretty closely mirrors what I hand-tracked except for in Thybulle’s case who is painted even better through estimation.

Nevertheless, there are a couple of issues with using this estimation. First, it does nothing to differentiate between “Great” and “Outstanding” plays. One would need to watch a players’ steals and blocks to make that determination. Second, some steals especially may be more of a product of luck instead of actual defensive skill. For instance, bad passes might be a clearer error from the offense. Again, one would need to watch and make this determination.

Finally, and most importantly, one cannot make an estimation about a player’s “Detrimental” plays. Since neither the box score nor any metric of which I’m aware contains information about blow-byes or missed rotations, it’s still necessary to hand-track those. Considering that the biggest difference I found between good and bad defenders is the number of “Detrimental” plays, this is a huge flaw with relying solely on estimation.

Limitations and Next Steps

  1. I spent this article only discussing the best and worst perimeter defenders. Next, we need a deeper analysis of the middle or so-called “League Average” defenders. Deciding on who fits that criteria will be difficult enough, but that just calls for a more robust database using my rubric.
  2. I also only discussed “Detrimental,” “Great,” and “Outstanding” plays while ignoring the “Good” section on my rubric. For some reason, I couldn’t conceive of it, so there’s a chance that the differences between the best, worst, and average defenders lies in the amount of “Good” plays they make.
  3. Possibly creating a new category for negative defensive plays. It feels fishy to lump all “Detrimental” plays together. Undoubtedly, there are levels between them.
  4. Better demarcating the sections of the rubric with clearer definitions that more people can collectively use. This rubric is useless if I keep the definition to myself without allowing others to contribute and use it.
  5. Accounting for matchup difficulty and defensive flexibility in these analyses. Basketball Index provides statistics for both which is necessary because a player defending James Harden should be expected to to make more mistakes than one guarding P.J. Tucker.
  6. Making blow-byes a staple box score stat.

If nothing else, I hope to begin pushing defensive conversations towards a more specific and shared vocabulary.

Cody Houdek is a writer and podcaster for Premium Hoops where he co-hosts Sense and Scalability. He also assists with videos for the Thinking Basketball YouTube channel. You can find all of his work (articlesvideos, and podcastshere.

Author

Leave a Reply