Analyzing Lebron's defensive activity via DFGAs

[VanWest82]

Edited title for better clarity.

When engaged, Lebron can be one of the more imposing defensive players in the game, even at age 35. But for most of the last 10 years, he hasn't regularly guarded the most imposing threat on the other team. That job was usually reserved for guys like Shane Battier, Iman Shumpert, Tristan Thompson, Jeff and Danny Green, and AD, to name a few.

If he's not guarding top options, how much is Lebron really contributing defensively? Let's examine.

It stands to reason that if Lebron is helping lots and/or guarding the ball lots that he'd wind up with lots of shots defended. Besides being common sense, we know this to be true from DFGAs leaderboards every year which are basically a who's who of the best (and worst) help defenders and shot contesters in the NBA.

But we don't even need to go there, and instead we can look at where Lebron stood vs. his own teammates in shots defended as that should give us a better idea who was actually finishing possessions defensively (independent of outcome).

DFGAs per 36

14 Heat RS: 1. Battier (18.1), Bosh (15.0), Lewis (16.0), Birdman (14.3), Chalmers (10.4), Wade (9.0), 7. Lebron (9.1)
14 Heat PS: 1. Lewis (14.6), Birdman (13.9), Bosh (13.4), Allen (10.6), Chalmers (9.9), Wade (9.8), 7. Lebron (7.2)

15 Cavs RS: 1. Mozgov (19.4), Varejao (17.5), TT (16.0), Love (14.6), Shump (12.2), Kyrie (10.8), JR (9.3), 8. Lebron (7.9)
15 Cavs PS: 1. Mozgov (18.7), TT (15.6), Delly (13.3), Kyrie (12.0), Shump (10.3), JR (9.1), 7. Lebron (8.3)

16 Cavs RS: 1. Mozgov (19.2), Love (15.3), TT (16.2), Delly (12.7), Shump (10.6) Kyrie (9.8), JR (8.8), 8. Lebron (8.5)
16 Cavs PS: 1. TT (16.1), Love (15.7), Frye (14.7), Kyrie (11.1), RJ (12.3), Shump (11.3), 7. Lebron (9.3), JR (9.1)

Spoiler:

17 Cavs RS: 1. Frye (19.0), TT (17.8), Love (15.5), RJ (13.9), Shump (11.7), 6. Lebron (10.9), JR (10.3), Kyrie (10.2)
17 Cavs PS: 1. TT (18.0), Love (15.9), Korver (12.5), Shump (12.0), DWill (11.1), JR (11.0), Kyrie (10.7), 8. Lebron (10.5)

18 Cavs RS: 1. Nance (17.3), Green (14.8), Love (14.2), Crowder (14.0), TT (13.7), Korver (13.0), 7. Lebron (10.8), JR (10.4)
18 Cavs PS: 1. Nance (18.2), Love (15.5), TT (14.3), Green (13.0), Hill (12.9), Korver (12.8), 7. Lebron (9.9), JR (9.6)

19 Lakers RS: 1. McGee (20.0), Hart (15.8), Lonzo (13.8), BI (12.4), Kuz (12.4), Rondo (11.6), KCP (10.9), 8. Lebron (10.3)

20 Lakers RS: 1. McGee (19.1), DH (16.6), Green (15.7), AD (15.3), Caruso (13.7), Kuz (13.1), KCP (10.1), 8. Lebron (8.1)
20 Lakers PS: 1. DH (17.4), Morris (14.6), AD (14.5), Green (14.0), Caruso (12.4), Kuz (10.3), 7. Lebron (10.1), KCP (8.2)

21 Lakers RS: 1. Gasol (20.1), AD (16.1), Harrell (15.6), Matthews (13.2) Kuz (12.7), Schroder (12.6), 7. Lebron (9.6), KCP (7.9)

Basically Lebron defends among the least amount of shots per 36 of rotation guys on all his teams. He either isn't guarding the ball or isn't helping on shot attempts or both. It's also worth noting that although Lebron has some remarkable DFG% numbers in 2016, most of the other years range somewhere between 0% and -3% diff, so he's having consistent, marginally negative impact on shooters but it isn't quite at the level of the best defenders in the game (who are likely also guarding tougher scorers in addition to way more shots as he typically takes the weaker assignment).

Again, we know Lebron is a good defender based on eye test and the fact that he's generally had negative net Drtg on/offs over the years. But there seems to be a major disconnect between the observable data and common narrative surrounding Lebron's defensive activity, particularly in help defense. I don't doubt that he's doing things to deter good offense but the easiest way to do that is contest a shot and make them miss, and that's not something Lebron has done with any meaningful frequency. Given this, how much defensive credit should he get despite being a relatively low activity player?

[Statlanta]

The only thing this thread needs is quotes from the Cavs organization because it is assumed that LeBron and TT lead the Cavs defense via leadership and production respectively. We know Spo and Vogel would never leave home without a good defense. Who was responsible for the 2015-2018 Cavs defense and how did it get so bad?

[VanWest82]

One analogy for Lebron’s defense is that of a good shooter with low usage and a good rep. We know there’s value (e.g. gravity) beyond just the made FGs, but ultimately you’re still dependent on the other four guys to make up all that extra volume.

[The Master]

Again, we know Lebron is a good defender based on eye test and the fact that he's generally had negative net Drtg on/offs over the years. But there seems to be a major disconnect between the observable data and common narrative surrounding Lebron's defensive impact, particularly in help defense. I don't doubt that he's doing other things to deter good offense but the easiest way to do that is contest a shot and make them miss, and that's not something Lebron has done with any meaningful frequency.

LeBron's strength on defense since 2014 is based on his team defense effort: great rotations and 'switchability' - he is really great in terms of defensive awareness, picking up his spots and rebounding, and while he can't defend guards on regular basis anymore, he's still much above average for 4s in guarding faster players. That's why last season according to eye test he was 'resurgent' defensively, while his advanced stats were pretty similar to his Cavs days (14-17 seasons) - but his team was much better defensively, so his defensive effort was more exploited in well organized defensive team.

The fact that he's not getting defensive coverage as big as other guys is a reason why in most seasons he was only 'good/very good' defensively, and not on All-NBA D level, because to be really elite in modern game you have to be both great on individual and team D level, and physically LeBron is unable to be as impactful as he used to be in the past. I don't think this data tells that much about his 'help defense' though, because what does it mean? If you mean by 'help defense' rim protection, then yeah, LeBron isn't any factor in this department outside of chasedown blocks, but there are other ways to play great team D. Most of time, LeBron in his 30s has been pretty good in conserving energy on defensive end AND playing decent D simultaneously.

[uberhikari]

I've seen a lot of crazy posts on realgm, but this one takes the cake. Using DFGAs per 36 minutes is wild crazy. When Gobert deters someone from even taking a shot attempt are we to assume that he's not really playing defense because he hasn't actually defended a FGA?

When LeBron has eidetic memory, memorizes your team's playbook, and is perfectly positioned so no one can even attempt a shot, we should just assume that LeBron isn't really playing defense. Ok, sport.

[VanWest82]

I've seen a lot of crazy posts on realgm, but this one takes the cake. Using DFGAs per 36 minutes is wild crazy. When Gobert deters someone from even taking a shot attempt are we to assume that he's not really playing defense because he hasn't actually defended a FGA?

Gobert is number one in the NBA in DFGAs. He not only deters shots from happening he causes many others to miss too. Doing it per 36 makes sense to help normalize the numbers and get a better idea of who was actually defending the end of possessions on the team. Why wouldn't we want to look at that?

When LeBron has eidetic memory, memorizes your team's playbook, and is perfectly positioned so no one can even attempt a shot, we should just assume that LeBron isn't really playing defense. Ok, sport.

Yeah that helps, and I did call him a great defender. But from the standpoint of defensive activity, DFGA says Lebron is defending the end of possessions less than his teammates. Some of that is he's being attacked less due to his rep or eidetic memory, but some of it is he's not always inviting the challenge defensively either. Lebron is great at defending his own area of the court.

If this were any other stat about any other player, we'd want to talk about it and discuss the implications.

[bondom34]

FWIW, Seth Partnow's been doing a series on shot quality on The Athletic recently. I'd asked him about using DFGA and percentages, and statisitcally at least they're highly variant. It's not something I'd use to measure defensive impact unless it's for shots at the rim by a rim protector (where the effect on a shot being contested is just way higher).

I'd asked:

For non-paint shots (or perimeter jumpers) is there any defensive correlation for this as well? For example, Jrue Holiday contests a certain number of shots greater than 15 feet from the basket but players are shooting above average percentages this season, is this mostly a variance thing? I'd always thought the perimeter numbers were more variance based and at the rim could be effected more by bigs.

And seems the response, from an analytics side:

There is very little evidence that individual players have much control of opponent jumpshooting percentages over and above other factors. Opponents shooting a lower percentage "against" a guy is largely a function of differences in the mix of shot qualty and shooter quality faced by differen defenders with a heavy dollop of variance thrown on top. I'll definitely get into this more in a later installment perhaps as early as next week

[VanWest82]

FWIW, Seth Partnow's been doing a series on shot quality on The Athletic recently. I'd asked him about using DFGA and percentages, and statisitcally at least they're highly variant. It's not something I'd use to measure defensive impact unless it's for shots at the rim by a rim protector (where the effect on a shot being contested is just way higher).

I'd asked:

Spoiler:

For non-paint shots (or perimeter jumpers) is there any defensive correlation for this as well? For example, Jrue Holiday contests a certain number of shots greater than 15 feet from the basket but players are shooting above average percentages this season, is this mostly a variance thing? I'd always thought the perimeter numbers were more variance based and at the rim could be effected more by bigs.

And seems the response, from an analytics side:

There is very little evidence that individual players have much control of opponent jumpshooting percentages over and above other factors. Opponents shooting a lower percentage "against" a guy is largely a function of differences in the mix of shot qualty and shooter quality faced by differen defenders with a heavy dollop of variance thrown on top. I'll definitely get into this more in a later installment perhaps as early as next week

You're talking about the DFG% diff. I'm just talking about DFGA as a very simple measure of defensive activity, like rebounds or steals or deflections (high level stuff, I know). What level of defensive impact are you really having if you're MIA on a bunch of the defensive box stats? I assume we acknowledge that counting stats are still important? Like if you're Rudy Gobert defending everything or Dame Lillard scoring everything you're probably more valuable than say Kyle Korver who doesn't shoot as much because everyone is so afraid to leave him open.

[bondom34]

FWIW, Seth Partnow's been doing a series on shot quality on The Athletic recently. I'd asked him about using DFGA and percentages, and statisitcally at least they're highly variant. It's not something I'd use to measure defensive impact unless it's for shots at the rim by a rim protector (where the effect on a shot being contested is just way higher).

I'd asked:

Spoiler:

For non-paint shots (or perimeter jumpers) is there any defensive correlation for this as well? For example, Jrue Holiday contests a certain number of shots greater than 15 feet from the basket but players are shooting above average percentages this season, is this mostly a variance thing? I'd always thought the perimeter numbers were more variance based and at the rim could be effected more by bigs.

And seems the response, from an analytics side:

There is very little evidence that individual players have much control of opponent jumpshooting percentages over and above other factors. Opponents shooting a lower percentage "against" a guy is largely a function of differences in the mix of shot qualty and shooter quality faced by differen defenders with a heavy dollop of variance thrown on top. I'll definitely get into this more in a later installment perhaps as early as next week

You're talking about the DFG% diff. I'm just talking about DFGA as a very simple measure of defensive activity, like rebounds or steals or deflections (high level stuff, I know). What level of defensive impact are you really having if you're MIA on a bunch of the defensive box stats? I assume we acknowledge that counting stats are still important? Like if you're Rudy Gobert defending everything or Dame Lillard scoring everything you're probably more valuable than say peak Kyle Korver who doesn't shoot as much because everyone is so afraid to leave him open.

Peak Kyle Korver, as an example, defended more shots/game (and I think per 36 by rough math in my head) than Thabo Sefolosha or Kent Bazemore on the 16 Hawks. Enes Kanter and Serge Ibaka are defending near identical numbers of shots.

On the perimeter, Kyrie is defending more shots than Lowry at the moment this season. Same for Harden. Just looking at shots defended, given their effect on both opponent shooting and that it can be effected by so many other things (players being targeted on defense, etc) isn't really informative.

It's not really measuring defensive activity either, it might just be telling us "the offense is trying to go at this guy" for all we know. The tracking stats aren't really measures of any sort of impact just telling things that happened, and even then there are some deeper questions as to how they're tracked if I recall.

Edit: And this may not be totally fair to Korver, who I don't think was as bad a defender as his reputation. But without looking into it more I'd say Thabo and Bazemore were likely more important defenders.

[VanWest82]

Peak Kyle Korver, as an example, defended more shots/game (and I think per 36 by rough math in my head) than Thabo Sefolosha or Kent Bazemore on the 16 Hawks.

Not the point I was making, but Korver was an underrated defender.

Enes Kanter and Serge Ibaka are defending near identical numbers of shots.

and have increasingly similar defensive impact these days.

On the perimeter, Kyrie is defending more shots than Lowry at the moment this season. Same for Harden. Just looking at shots defended, given their effect on both opponent shooting and that it can be effected by so many other things (players being targeted on defense, etc) isn't really informative.

Lowry has taken a step back defensively this year and isn't doing as much. It's an unspoken truth over at the Raptors board. He's still a brilliant player though.

It's not really measuring defensive activity either, it might just be telling us "the offense is trying to go at this guy" for all we know. The tracking stats aren't really measures of any sort of impact just telling things that happened, and even then there are some deeper questions as to how they're tracked if I recall.

There's definitely some of that going on, and surely some guys are also trying to avoid going at Lebron. But when you grade out the same way for 7+ years straight I think the numbers are harder to argue. The reason why Shump and RJ and Crowder and AD and Danny Green, etc., etc, guarded more shots than Lebron is because they drew the tougher assignment most nights.

I'm not trying to say DFGA is anything more than what it measures. Dismissing it out of hand because some guys show up a little higher than you might think because they're being attacked more doesn't make sense when we have good evidence and reasoning to explain those cases.

What we might say is that because Lebron has low DFGA, blocks, steals, and deflections for someone with his size, ability, and reputation, these measures together would suggest that his defensive activity level is low outside of rebounding.

[bondom34]

Not the point I was making, but Korver was an underrated defender.

Underrated? Sure. One of the Hawks best perimeter defenders? I'm struggling there.

Enes Kanter and Serge Ibaka are defending near identical numbers of shots.

This is a pretty hard one for me to really buy into.

Lowry has taken a step back defensively this year and isn't doing as much. It's an unspoken truth over at the Raptors board.

Taken a step back isn't saying he's a poor defender, and I don't think the same could be said of Irving.

There's definitely some of that going on, and surely some guys are also trying to avoid going at Lebron. But when you grade out the same way for 7+ years straight I think the numbers are harder to argue. The reason why Shump and RJ and Crowder and AD and Danny Green, etc., etc, guarded more shots than Lebron is because they drew the tougher assignment most nights.

I'm not trying to say DFGA is anything more than what it measures. Dismissing it out of hand because some guys show up a little higher than you might think because they're being attacked more doesn't make sense when we have good evidence and reasoning to explain those cases.

What we might say is that because Lebron has low DFGA, blocks, steals, and deflections for someone with his size, ability, and reputation, these measures together would suggest that his defensive activity level is low outside of rebounding.

I'd tend to side with "the box score doesn't measure defense well". He's been on one of the best/the best defense in the NBA the last few years, and the defensive on/off splits are strong enough even without Davis to indicate he's a smart and strong offball defender which wouldn't be measured by steals/blocks/deflections. I'd look to a guy like Lu Dort on the opposite end. He's been called out for being an amazing defender despite all evidence pointing to him being relatively poor offball, because he puts up good on ball defensive measures.

Ultimately I wouldn't touch any of steals/blocks/deflections/DFGA as a real measure of defensive impact (and I'm not saying this to stick up for Lebron, just as a general stance). There's just a lot more going on than on ball defense, and a box score is a pretty poor measure of it. Of note, even if one was to look into a box score measure, Lebron has been top 20 in the NBA in DBPM this season, and going further was 51st in 3 year LA-DRAPM from 2017-20 (which puts him right next to Lowry oddly enough). His non-AD lineups this year are in the 92nd percentile defensively.

There's just a lot more to defense than on ball, and as a general rule I don't trust the tracking data or raw blocks/steals numbers a ton for partially that reason. Some of this could be someone's off ball defense, some could be a defense just trying to avoid a player, etc. Not even making a case specific here (though I think his defense is being understated a bit if he's being compared to Jefferson or Shump or the like) but just a general thing.

[Baski]

I've seen a lot of crazy posts on realgm, but this one takes the cake. Using DFGAs per 36 minutes is wild crazy. When Gobert deters someone from even taking a shot attempt are we to assume that he's not really playing defense because he hasn't actually defended a FGA?

Gobert is number one in the NBA in DFGAs. He not only deters shots from happening he causes many others to miss too. Doing it per 36 makes sense to help normalize the numbers and get a better idea of who was actually defending the end of possessions on the team. Why wouldn't we want to look at that?

Per 36 would do away with LeBron's advantage over his teammates in sheer volume right? He played monster minutes in all those years and being as active defensively as his teammates per minute would be virtually impossible given his other responsibilities for the team. Per 36 has always been a suspicious measure for me because it leads you to assume that there is zero/minimal added value in being able to stay on the court longer and pace yourself. Boban's per 36 numbers have always looked nice but there's a reason he doesn't play 30 mpg.
It stood out to me that the stat as you've laid out has Birdman and Varejao, two low minute high energy guys whose job is to come in in garbage time and blow their whole wad of energy in 10 minutes and then get out of the game, as seemingly better/more active overall than LeBron. That alone should raise eyebrows, but then including guys who barely played, and certainly not in important possessions like Delly, RJ, Frye kinda undermines the whole thing.

I think to add some measure of transparency/objectivity to your analysis you should add per game DFGA stats, mpg, hell even DFGA totals would help paint a better picture, and then let the discussion include how much we should weigh the sheer volume vs the normalized volume you favour.

[VanWest82]

I'd tend to side with "the box score doesn't measure defense well". He's been on one of the best/the best defense in the NBA the last few years, and the defensive on/off splits are strong enough even without Davis to indicate he's a smart and strong offball defender which wouldn't be measured by steals/blocks/deflections. I'd look to a guy like Lu Dort on the opposite end. He's been called out for being an amazing defender despite all evidence pointing to him being relatively poor offball, because he puts up good on ball defensive measures.

So rather than trust the observable data you prefer to trust that his teammates don't have any impact on their own shots defended. Given we agree Lebron is mostly off ball and not contesting, aren't any changes in line ups around him and resulting data just a reflection of the other players and their performance? After all, they're the ones doing the actual defending/helping. Lebron is just deterring. He's the constant. Feels like we have the variables backwards.

Perhaps the Lebron/AD defensive on/offs have more to do with things like Lebron playing more mins with Caruso and AD more with Schroder, as an example.

Ultimately I wouldn't touch any of steals/blocks/deflections/DFGA as a real measure of defensive impact (and I'm not saying this to stick up for Lebron, just as a general stance). There's just a lot more going on than on ball defense, and a box score is a pretty poor measure of it. Of note, even if one was to look into a box score measure, Lebron has been top 20 in the NBA in DBPM this season, and going further was 51st in 3 year LA-DRAPM from 2017-20 (which puts him right next to Lowry oddly enough). His non-AD lineups this year are in the 92nd percentile defensively.

Lowry hasn't taken the tougher defensive assignments either for years. 2017 was the last year and it didn't go as well. He's definitely more active than Lebron as a help defender though.

I don't like all-in-one stats as there's way too much noise even with all the adjustments. The combo ones like RAPTOR make way more sense to me. You don't get to have all the credit for doing the least, but if the team generally performs better with you on the court you should get some credit for that.

There's just a lot more to defense than on ball, and as a general rule I don't trust the tracking data or raw blocks/steals numbers a ton for partially that reason. Some of this could be someone's off ball defense, some could be a defense just trying to avoid a player, etc. Not even making a case specific here (though I think his defense is being understated a bit if he's being compared to Jefferson or Shump or the like) but just a general thing.

I'm just saying those guys guarded the better offensive players, and defended more shot attempts because of it. It's not a comparison. It's a statement of fact.

I see no reason not to trust the tracking data, and have yet to see anything suggesting the cameras are faulty or don't work. It doesn't tell you why or how but it's not trying to. It's just observable measurement, which when you add all of it together gives you a lot of information about a lot of different possessions that you didn't have. The idea that getting a deflection or help defending a shot attempt is useless information doesn't make sense to me.

[Blackmill]

Basically Lebron defends among the least amount of shots per 36 of rotation guys on all his teams. He either isn't guarding the ball or isn't helping on shot attempts or both. It's also worth noting that although Lebron has some remarkable DFG% numbers in 2016, most of the other years range somewhere between 0% and -3% diff, so he's having consistent, marginally negative impact on shooters but it isn't quite at the level of the best defenders in the game (who are likely also guarding tougher scorers in addition to way more shots as he typically takes the weaker assignment).

Again, we know Lebron is a good defender based on eye test and the fact that he's generally had negative net Drtg on/offs over the years. But there seems to be a major disconnect between the observable data and common narrative surrounding Lebron's defensive impact, particularly in help defense. I don't doubt that he's doing things to deter good offense but the easiest way to do that is contest a shot and make them miss, and that's not something Lebron has done with any meaningful frequency.

One thing you've missed in your analysis (see the underlined portion) is that the best way to stop the offense isn't to contest the shot. It's to prevent the shot from even happening. For instance, imagine LeBron closes out to a three point shooter, and because LeBron is fast and large the shooter doesn't attempt the three. Maybe the play is now dead and the offense resets. Or maybe the shooter attacks the close out and gets his hips past LeBron, but even that's acceptable, because AD or Bosh is waiting in the paint. So the guy passes the ball anyways and it doesn't appear as a dFGA for LeBron. Or he takes a tough shot over an athletic center. But the goal (read: to prevent the open three) was accomplished. There's a few other ways that this type of statistical argument can fail to describe reality (i.e. the importance of communication, the unreliability of dFGA stats, etc.) so I don't put much weight on it. Anyways, sometimes it's better to trust the film than the stats.

[VanWest82]

One thing you've missed in your analysis (see the underlined portion) is that the best way to stop the offense isn't to contest the shot. It's to prevent the shot from even happening. For instance, imagine LeBron closes out to a three point shooter, and because LeBron is fast and large the shooter doesn't attempt the three. Maybe the play is now dead and the offense resets. Or maybe shooter attacks the close out and gets his hips past LeBron, but even that's acceptable, because AD or Bosh is waiting in the paint. So the guy passes the ball anyways and it doesn't appear as a dFGA for LeBron. Or he takes a tough shot over an athletic center. But the goal (read: to prevent the open three) was accomplished. There's a few other ways that this type of statistical argument can fail to describe reality (i.e. the importance of communication, the unreliability of dFGA stats, etc.) so I don't put much weight on it. Anyways, sometimes it's better to trust the film than the stats.

Now imagine Lebron doesn't close out to the three point shooter - per tracking, he covers the least amount of defensive ground per min on Lakers, and does so at the slowest pace (even slower than Marc Gasol) - and instead the shooter gets off an open look or someone else has to make up the ground for a late contest. No DFGA for Lebron. You know what doesn't match the eye test? The notion that Lebron is out there deterring all the good shooters and forcing them into bad looks elsewhere. Rather, he spends a lot of time on the weak side of the defense or away from the ball, and switching so he's not having to expend energy fighting through screens.

[RCM88x]

Now tell me how often the other teams score when LeBron is on the floor vs when he is off. Aka the only thing that matters. Why? Because the objective is to outscore your opponent.

[VanWest82]

Now tell me how often the other teams score when LeBron is on the floor vs when he is off. Aka the only thing that matters. Why? Because the objective is to outscore your opponent.

Both this year and last year the Lakers have had a better Drtg with AD on the bench. I guess they should bench him even though he's clearly their best and most impactful defender.

[SweetTouch]

Agree with OP LBJS defense is second to none

Thanks for taking your time for the King

[bondom34]

I'd tend to side with "the box score doesn't measure defense well". He's been on one of the best/the best defense in the NBA the last few years, and the defensive on/off splits are strong enough even without Davis to indicate he's a smart and strong offball defender which wouldn't be measured by steals/blocks/deflections. I'd look to a guy like Lu Dort on the opposite end. He's been called out for being an amazing defender despite all evidence pointing to him being relatively poor offball, because he puts up good on ball defensive measures.

So rather than trust the observable data you prefer to trust that his teammates don't have any impact on their own shots defended. Given we agree Lebron is mostly off ball and not contesting, aren't any changes in line ups around him and resulting data just a reflection of the other players and their performance? After all, they're the ones doing the actual defending/helping. Lebron is just deterring. He's the constant. Feels like we have the variables backwards.

Perhaps the Lebron/AD defensive on/offs have more to do with things like Lebron playing more mins with Caruso and AD more with Schroder, as an example.

Ultimately I wouldn't touch any of steals/blocks/deflections/DFGA as a real measure of defensive impact (and I'm not saying this to stick up for Lebron, just as a general stance). There's just a lot more going on than on ball defense, and a box score is a pretty poor measure of it. Of note, even if one was to look into a box score measure, Lebron has been top 20 in the NBA in DBPM this season, and going further was 51st in 3 year LA-DRAPM from 2017-20 (which puts him right next to Lowry oddly enough). His non-AD lineups this year are in the 92nd percentile defensively.

Lowry hasn't taken the tougher defensive assignments either for years. 2017 was the last year and it didn't go as well. He's definitely more active than Lebron as a help defender though.

I don't like all-in-one stats as there's way too much noise even with all the adjustments. The combo ones like RAPTOR make way more sense to me. You don't get to have all the credit for doing the least, but if the team generally performs better with you on the court you should get some credit for that.

There's just a lot more to defense than on ball, and as a general rule I don't trust the tracking data or raw blocks/steals numbers a ton for partially that reason. Some of this could be someone's off ball defense, some could be a defense just trying to avoid a player, etc. Not even making a case specific here (though I think his defense is being understated a bit if he's being compared to Jefferson or Shump or the like) but just a general thing.

I'm just saying those guys guarded the better offensive players, and defended more shot attempts because of it. It's not a comparison. It's a statement of fact.

I see no reason not to trust the tracking data, and have yet to see anything suggesting the cameras are faulty or don't work. It doesn't tell you why or how but it's not trying to. It's just observable measurement, which when you add all of it together gives you a lot of information about a lot of different possessions that you didn't have. The idea that getting a deflection or help defending a shot attempt is useless information doesn't make sense to me.

I've posted reasons to not buy the tracking data. It doesn't tell you if the player contesting is doing it because their man shot the ball or if they were forced into a bad switch or if they're being actively attacked by an offense. Maybe offenses are avoiding Lebron in this instance. The inference in the OP from the shot contests is "Lebron isn't giving effort." Maybe it's really "defenses don't challenge Lebron and avoid him". Using a guy like RJ or Shump spending time on certain players also seems to dismiss Lebron's defense a good bit.

Ultimately using shot contests to try to extract defensive effort or ability seems pretty futile. There are about a million things that go into it that aren't in the output. Something like an APM at the least gives an idea of effect on team defense. Circling back to the OP:

Again, we know Lebron is a good defender based on eye test and the fact that he's generally had negative net Drtg on/offs over the years. But there seems to be a major disconnect between the observable data and common narrative surrounding Lebron's defensive impact, particularly in help defense. I don't doubt that he's doing things to deter good offense but the easiest way to do that is contest a shot and make them miss, and that's not something Lebron has done with any meaningful frequency.

As noted by Blackmill above, an easier way to do this is to not allow a shot, or to deter it, or again, to have a defense avoid you. Focus seems to be in the wrong spot.

[Blackmill]

Spoiler:

One thing you've missed in your analysis (see the underlined portion) is that the best way to stop the offense isn't to contest the shot. It's to prevent the shot from even happening. For instance, imagine LeBron closes out to a three point shooter, and because LeBron is fast and large the shooter doesn't attempt the three. Maybe the play is now dead and the offense resets. Or maybe shooter attacks the close out and gets his hips past LeBron, but even that's acceptable, because AD or Bosh is waiting in the paint. So the guy passes the ball anyways and it doesn't appear as a dFGA for LeBron. Or he takes a tough shot over an athletic center. But the goal (read: to prevent the open three) was accomplished. There's a few other ways that this type of statistical argument can fail to describe reality (i.e. the importance of communication, the unreliability of dFGA stats, etc.) so I don't put much weight on it. Anyways, sometimes it's better to trust the film than the stats.

Now imagine Lebron doesn't close out to the three point shooter - per tracking, he covers the least amount of defensive ground per min on Lakers, and does so at a slower pace (even slower than Marc Gasol) - and instead the shooter gets off an open look or someone else has to make up the ground for a late contest. No DFGA for Lebron.

Obviously, if a player doesn't try to contest the shot attempt, they don't register a defensive field goal attempt. And that may come with other consequences. That's not a counter point to what I said which is that defensive field goal attempts is an inadequate stat for assessing defense. Say, because the best defensive plays don't result in a field goal even being attempted. Or as in your own example, because defensive field goal stats don't capture all bad plays. You're making my point here.

You know what doesn't match the eye test? The notion that Lebron is out there deterring all the good shooters and forcing them into bad looks elsewhere. Rather, he spends a lot of time on the weak side of the defense or away from the ball, and switching so he's not having to expend energy fighting through screens.

I disagree. LeBron absolutely deters shooters. Especially in meaningful games. Is he still often playing on the weak side? Yes, and I would say he has more value from that position, at least while the rest of the team matches up well enough. We've seen several times how in close games those role dynamics can change. Prior to last years shut-down, the Lakers won games against the Bucks and Clipper with LeBron spending possessions on Giannis and Kawhi down the stretch. We saw that again against the Nuggets in game four. And in the last game of the 2020 finals, where Butler only took ten shots in part because was LeBron guarding him while the game was close (among other reason, Butler looked tired, and LA as a whole played extremely well, but LeBron did his part).

Jordan at his physical peak had as good a motor as any one. And yet, in the 1991 finals, Jordan needed Scottie to spot him minutes guarding Magic so he could stay out of foul trouble and have something left at the end of games. In the early 2000's when the Lakers often faced the Spurs, Shaq tended not to guard Duncan, and on many possessions Duncan wasn't guarding Shaq. It's very common (and for good reason) that a team's offensive engine doesn't guard the opponent's offensive engine if another teammate can do a good enough job. At least not all the time. And this is especially true when the teammate in this example is a worse help defender.