Kawhi Leonard, NBA

By Mike O’Connor

Individual statistics were created because we, as humans, are unable to capture all the happenings of a game with the naked eye. They exist to quantify, highlight and uncover the many parts of a game our brains fail to store.

In their most rudimentary form, they do so by counting things. This means points, shots, rebounds, etc. This method captures offensive impact imperfectly, but cannot begin to quantify defensive impact.

Good defense is characterized by the absence of things. The best defense is not a missed shot attempt, but rather the forcing of a pass. When Kawhi Leonard swarms or intimidates a player so badly they had to pass, his defensive play goes completely unrecorded. This is the most ideal byproduct of good individual defense, and yet there is no method of counting to capture it.

To make up for this, we’ve created countless advanced stats to measure relative and individual impact. But even those have massive, glaring flaws of varying types and causes. Let’s discuss.

Tier 1: The Absolutely Never, Ever Use:

1. Synergy Play-Type Defensive Stats

Synergy is a magnificent offensive tool. But defensively, synergy stats are the quintessentially flawed defensive stat. They measure only possession-ending events. That means that the above scenario with Kawhi Leonard forcing a pass is simply not counted. Or, in this example, Kemba Walker seeing Joel Embiid and respectfully declining to challenge him does not show up in Joel Embiid’s defensive pick and roll stats.

Even in more extreme cases where Embiid swarms the ball handler but the guard is able to get a pass out, Embiid’s defense simply goes unrecorded. Consider this play, as well:

If Embiid had been defending, perhaps Goran Dragic turns the corner, is dissuaded and swings the ball back out top to reset the possession with 10 seconds left on the shot clock. Instead, Jahlil Okafor’s pick and roll stats get rewarded for arguably the worst defense ever played in professional basketball.

Such is the problem with only measuring possession-ending events. It overlooks the best type of defense entirely, and judges the player based on the fickle result, rather than the process it took to get there.

*Note*: The one area where this is an exception is post-ups. It’s perfectly valid at that point to assume that a shot is coming and judge a player based on how they defend it.

2. Overall Opponent Field Goal Percentage

This stat is terribly misleading for the same reasons listed above, with two additional concerns.

First, every shot assigns an opposing defender to it. If a player shoots a wide open three from the corner, the big man with his back turned may get tagged as the defender and his field goal percentage against will suffer.

Second, it’s also important to consider how replicable all defensive recoveries are. By the time an open player receives the ball to shoot, the damage has been done. Using field goal percentage against to measure defensive players based on how they close out often labels the player scurrying for the recovery as the culprit as opposed to the one who caused the defensive breakdown to begin with.

There is one exception to this stat, which we’ll circle back to.

3. Individual Defensive Rating

Let me make one clarification to begin with. Individual defensive rating is NOT a measure of the team’s defensive rating while the player is on the floor. It is an estimate of the player’s defensive impact, but still almost entirely based on team context. 28 of the top 50 players in defensive rating were Spurs, Warriors, Hawks, Jazz, or Pistons. It is simply a massively incomplete way to compare players to each other league wide.

Tier 2: Be very, very careful

1. Defensive Win Shares 

Like many defensive stats, many people may not know how it is even calculated. Defensive Win Shares is calculated by multiplying per minute efficiency by minutes played, thus removing the per-minute basis from the stat entirely. For example, Giannis Antetokounmpo ranks ninth in the NBA in defensive win shares, but ranks 47th on a per-minute basis courtesy of the conversion from nba.com/stats.

2. Real Plus/Minus and Box Plus/Minus

Real plus/minus is possibly the most empirical advanced defensive stat given it combines the box score data from box plus/minus and combines it with lineup data. Still, it rewards players who record counting stats (steals, blocks, etc.) more so than it does normal, fundamentally sound defense. Timothé Luwawu-Cabarrot, for example, ranked as the 14th best shooting guard in the NBA in defensive real plus-minus. While I love TLC, he has no business being ranked ahead of Klay Thompson, Justin Holiday, Nicolas Batum, and Dwyane Wade.

Tier 3: Use with caution

1. On/off splits 

I find these stats generally do a pretty accurate job of isolating how a team performs with or without one or two players. But there’s one major caveat.

Using the on/off splits to compare players on the same team should be used with extreme caution, particularly when these two players play the same position. More often than not, it means you’re looking at how an entirely different lineup performs compared to another. For example, are the Spurs really only 2.8 points worse defensively per 100 possessions with David Lee compared to Dwayne Dedmon, or are we just observing that Lee played 250 more minutes in lineups with Kawhi Leonard and Patty Mills? The latter, most likely.

We also must never forget to ask, “why?” The investigation into Kawhi Leonard’s mysterious defensive rating sent an appropriate wave of shock through the NBA twittersphere. The Spurs’ defense was 8.2 points better with Kawhi on the bench this season. Huh?

This conundrum went largely unsolved until Bo Schwartz of Nylon Calculus discovered a tremendously strange pattern of three-point shots with Kawhi on and off the court. Opponents shot a league-best percentage from three with Kawhi on the floor, and a league worst with him on the bench.

This maddeningly complex dynamic should call us to question not only this stat, but also the aforementioned field goal percentage against. Again, defense is very arbitrary!

2. Lineup data

These stats can help to uncover defensive solutions for a team if they are having success with one particular lineup. For example, holy crap the Sixers are a lot better in lineups with T.J. McConnell than Sergio Rodriguez! Still, the issue here is sample size. Even lineups with 100 or more minutes together could simply be benefitting from an easy stretch in the schedule in which the coach began playing them together. For that reason, these stats don’t tell much until the later portion of the season.

Tier 4: Go Ahead!

1. Field Goal Percentage against at the rim 

Remember that exception I said we’d come back to? This is it.

Field goal percentage against as measured from six feet and in is a very reliable indicator of the degree of rim protection a player provides. It is not plagued by the same issues of the overall field goal percentage against because:

  • The defender is likely always close and not arbitrarily chosen
  • At this point in the court, a shot attempt is imminent. We are not concerned about whether a good defender would force a pass.
  • No such area on the court is as easy to measure a player’s ability to disrupt the shot. Players drill jump shots in defenders’ faces all the time regardless of who is defending. Finishing over Rudy Gobert is not nearly as fickle.

Go ahead and let this one fly.

2. Pace-Adjusted Counting Stats

While I am the first to point out that counting stats do not begin to quantify defensive impact, using pace-adjusted counting stats can be very useful for nuanced purposes. It is important to consider who are the best shot blockers or pocket-pickers in the league. This is not to say that they encompass a player’s full value, but rather just that there are no flaws in it.

The Verdict

Notice the two least flawed stats are two of the most nuanced. We are still left with an outright failure to use defensive stats to quantify wholesome impact. For this reason, we should avoid using one or two advanced defensive stats for narrative-forming purposes.

It’s tempting and understandable to fall in love with stats of all kinds. They show a player is this good, or has that type of impact on his team. But they simply cannot capture these impacts, especially on defense.

That’s not to say all defensive stats are completely worthless or inherently false. It simply means we must take each stat with a massive grain of salt and not forget to ask, “why?” to each one. Because our current level of discussion has failed to do that, leading to a series of lazy narratives.

It has created a conundrum where most can see Hassan Whiteside, Andre Drummond, Russell Westbrook, John Wall and many more are spaced out off-ball defenders who get burnt on a regular basis. But there is no way to prove it. Many people will cite stats, or even tens of plays where the opposite is true about each of those players. But it is maddeningly difficult to quantify or even begin to discuss our opinions on these issues because the inability to rely on stats create a wide scope of subjectivity.

As with anything, the correct answer is always somewhere in the middle, and the right answers are achieved through a mixed and thoroughly considered blend of stats and eye test. But going forward, we must proceed with far more caution.

[newsbox style=”nb1″ display=”tag” tag=”oconnor” title=”More from Mike O’Connor” number_of_posts=”2″ show_more=”no” nb_excerpt=”0″]

Mike O'Connor

View all posts

Subscribe on YouTube

The Podcast

Subscribe on YouTube