LeBron has the worst plus/minus of any player on a good team

[Rust_Cohle]

Taj in shambles

[JonFromVA]

No, that's noise. The 56th best player in the league isn't a negative player...

I mean he literally is a negative player ... when you add up all his floor time numbers they come up negative. It's not noise, it's a hard cold fact.

The meat and potatoes are in the lineup data, but per usual people would rather scream about noise than dig in to the numbers.

I mean RAPM and EPM aren't going to sort this **** out, you've got to watch his games (something I don't do anymore) and dig in to those lineups.

And fwiw, LeBron started half assing it during his second stint with the Cavs (slowing down pace, resting on D, launching J's rather than attacking); but he'd just get his act together in time for the playoffs.

RAPM and EPM say he's a plus. Raw plus minus says he'd negative. All else equal, I'll go with the two metrics that are actually predictive.

Funny thing, but when we're talking about what's happened int he past, we aren't interested in "predictions"; but those stats do a lot of things. They may have box score priors - meaning they use old box score data not necessarily even from this season or really who knows what unless we're talking about a public model.

But even assuming 100% good faith in the design of the model, the intention of APM models is to account for the affect teammates have on a player which is important to understanding why LeBron ended up negative, but like I said the number itself doesn't tell you why - it's buried in the lineup data.

We know the actual APM model itself starts to become really good after multiple seasons of data (not that I've ever seen someone quantify what that means regarding an individual player rather than over the entire population of the model on average) but what does 38/39 year old LeBron have to do with 39/40 year old LeBron?

You can usually get a quick idea about +/- numbers by looking at who has a better +/- on the team and who has a worse and look at the distribution of minutes playing with those guys. For instance, James had some pretty atrocious minutes playing with Knecht and Reddish. It suggests he can't just uplift random units like he used to *or* perhaps he'd even stopped trying to.

[dhsilv2]

I mean he literally is a negative player ... when you add up all his floor time numbers they come up negative. It's not noise, it's a hard cold fact.

The meat and potatoes are in the lineup data, but per usual people would rather scream about noise than dig in to the numbers.

I mean RAPM and EPM aren't going to sort this **** out, you've got to watch his games (something I don't do anymore) and dig in to those lineups.

And fwiw, LeBron started half assing it during his second stint with the Cavs (slowing down pace, resting on D, launching J's rather than attacking); but he'd just get his act together in time for the playoffs.

RAPM and EPM say he's a plus. Raw plus minus says he'd negative. All else equal, I'll go with the two metrics that are actually predictive.

Funny thing, but when we're talking about what's happened int he past, we aren't interested in "predictions"; but those stats do a lot of things. They may have box score priors - meaning they use old box score data not necessarily even from this season or really who knows what unless we're talking about a public model.

But even assuming 100% good faith in the design of the model, the intention of APM models is to account for the affect teammates have on a player which is important to understanding why LeBron ended up negative, but like I said the number itself doesn't tell you why - it's buried in the lineup data.

We know the actual APM model itself starts to become really good after multiple seasons of data (not that I've ever seen someone quantify what that means regarding an individual player rather than over the entire population of the model on average) but what does 38/39 year old LeBron have to do with 39/40 year old LeBron?

You can usually get a quick idea about +/- numbers by looking at who has a better +/- on the team and who has a worse and look at the distribution of minutes playing with those guys. For instance, James had some pretty atrocious minutes playing with Knecht and Reddish. It suggests he can't just uplift random units like he used to *or* perhaps he'd even stopped trying to.

All metrics have always been judged on their ability to predict. That's really the only way to test if the number it spits out was accurate in the past. We have to somewhat assume player's don't just fall off a cliff on the whole to see if this works. And thankfully it does.

Yeah, agree if you want to dig into what Lebron is clearly a plus player but has s negative on off...we'd need to dig into lineups. We'd need to look at rotations. And we'd need to look at outliers...games with HUGE swings that might cause more noise than value. And that's what RAPM metrics attempt, and actually do a reasonable job at.

[JonFromVA]

Fact is most players anywhere near LeBron's age are either out of the league or being used very selectively. He should be given lots of rest, he should be used in lineups that protect him, he should be used in matchups that don't exploit him; but apparently the Lakers and James are happy to try to squeeze whatever is left out of him and it's not surprising at his age that even more cracks are starting to show.

Were people really surprised to hear Windhorst say the Lakers are planning their future around Doncic and not James?

[eminence]

He's clearly well into his decline, but this sample likely overstates it.

[Ainosterhaspie]

He started slow, then was playing very well, then got injured and was back to where he started. His bad +/- numbers aren't that surprising if you zoom out like that.

Slow starts, injuries, and long recovery from injuries are part of the package at this age. When everything is right, and the stakes are high, you're still getting a highly impactful player, but those moments are increasinlgy rare.

He really needs to be the team's second or third best offensive player and same on defense.

[JonFromVA]

RAPM and EPM say he's a plus. Raw plus minus says he'd negative. All else equal, I'll go with the two metrics that are actually predictive.

Funny thing, but when we're talking about what's happened int he past, we aren't interested in "predictions"; but those stats do a lot of things. They may have box score priors - meaning they use old box score data not necessarily even from this season or really who knows what unless we're talking about a public model.

But even assuming 100% good faith in the design of the model, the intention of APM models is to account for the affect teammates have on a player which is important to understanding why LeBron ended up negative, but like I said the number itself doesn't tell you why - it's buried in the lineup data.

We know the actual APM model itself starts to become really good after multiple seasons of data (not that I've ever seen someone quantify what that means regarding an individual player rather than over the entire population of the model on average) but what does 38/39 year old LeBron have to do with 39/40 year old LeBron?

You can usually get a quick idea about +/- numbers by looking at who has a better +/- on the team and who has a worse and look at the distribution of minutes playing with those guys. For instance, James had some pretty atrocious minutes playing with Knecht and Reddish. It suggests he can't just uplift random units like he used to *or* perhaps he'd even stopped trying to.

All metrics have always been judged on their ability to predict. That's really the only way to test if the number it spits out was accurate in the past. We have to somewhat assume player's don't just fall off a cliff on the whole to see if this works. And thankfully it does.

Yeah, agree if you want to dig into what Lebron is clearly a plus player but has s negative on off...we'd need to dig into lineups. We'd need to look at rotations. And we'd need to look at outliers...games with HUGE swings that might cause more noise than value. And that's what RAPM metrics attempt, and actually do a reasonable job at.

Statistical predictions are wonderful for betters who are willing to make a large number of bets and just need the slightest edge.

But hey if you can dig up an error bracket with 99.73% surety (3 sigma) for LeBron's EPM or RPM, I'd be glad to give the numbers some more credence.

[Doctor MJ]

Only including players on teams with winning records who play at least 20mpg. He literally has the worst plus minus among all of them.This is a damning statistic if there ever was one and should be considered a black mark on his career. The only thing that matters is winning and he makes his team significantly worse when he’s playing because he cares more about stat padding. If i were a lakers fan I’d be praying he retires after this season

Okay, so as someone who regularly uses raw +/- in my analysis, and has notice what the OP says about LeBron this year, here are my thoughts:

1. Whenever I use raw +/-, I don't do it as a final step. I think it's a great early step, but obviously it doesn't factor in correlations between players and other context.

2. While I'm generally skeptical about stats that purport to be luck-adjusted, luck is certainly a thing, and there are indicators this year that LeBron's been specifically unlucky.

3. While LeBron has been THE single most impressive +/- player of the databall era (going back to '96-97). he has had blips like this before which means I don't think this issue is just about noise.

4. However, all those blips are regular season stuff, and I think it's clear right now that LeBron is one of the most capable defenders on the Lakers. When it's critical, LeBron can do it, and given that he's like 60 years old, I'm not that judgy when he coasts a bit in the RS.

5. And yeah, I do think LeBron remains someone who can be a star-caliber player on a championship team. He ain't what he used to be, but his collection of gifts on both sides of the ball is still enough that I don't think you need to trade him away to be a contender necessarily.

6. But, he's also not an ideal fit with Luka which was why I said at the moment of the trade I don't expect the team to be able to actually contend for a title until they fill the hole left by AD.

7. This means that over this critical off-season what I'd really like to do is keep Luka & LeBron (as well as Reaves, DFS, etc), and add a nice big. To me if you can do that, that's your best bet for a Luka-LeBron chip.

8. Of course, in the long term LeBron won't be around. While it seems unlikely that the Lakers would trade, or be able to trade, LeBron for that missing piece, the reality is that if the Lakers don't win a chip next year, they shouldn't be counting on LeBron as a critical player after that. This is Luka's team now after all.

[runtmc]

I'm trying to look at this from a statistical point of view. Its obvious single game +/- isnt meaningful. But for the entire season, and considering almost everyone around him seems to have a positive value, and his entire team has a hugely positive outcome, i try to find an explanation to prove the hypothesis wrong that he is in fact a negative on an otherwise positive team. Can you explain?

I'm confused why you're asking me this question when I was one of the few who defended the significance of the OP.

Since you insist on making me play devil's advocate, here's an example to help you:

Team X is a juggernaut with 5 all stars in its starting lineup. 4 of the all stars get hurt, so the other is left playing with vet mins and G Leaguers for much of the season. They perform as expected, terribly, until the other 4 guys come back from injury at the same time. All of a sudden they're a juggernaut again. 4 stars put up majorly positive +/- and on/off while the other guy's numbers look terrible by comparison. Not because he's a negative impact player but purely as a result of circumstances out of his control.

It's an outlier example but demonstrates pretty clearly how a non negative player can appear negative. Hope that helps. Outliers happen and context is a complex beast with countless factors to consider.

Thank you that would be one hypothetical case where it could happen. However in teh OP we see the entire Lakers roster including AD himself, so everyone who played this season is on there (in your example, the imaginary 4 vet min guys woould obviously have the worst +/- by far). Of all players who played for the lakers this season, LeBron is amost dead last. I still look for a good explanation for this concrete season and this data, explaining how this doesnt clearly shows his negative impact.

What i understand is that no matter who he played with, those players got worse by being on the floor with him,, and the only reason the Lakers had a positive record overall is him sitting out. And it matches the eye test - he was often criticsed for his visible lack of defensive effort.

A lot of posters have ridiculed the OP for quoting raw +/- , maybe one of those experts can give a interpretation that explains the numbers. The only way i could twist this in favor of leBron is that for an entire season, while he played good, he had the incredible misfortune to catch every team mate in their worst minutes.

Raw +/- *is* incredibly noisy. Incredibly. James for example (for a game) ranges from a +29 to a -30 this season. That's an enormous spread (aka very noisy).

To give a toy example, imagine flipping a coin. If you look at say a group of 50 10-flip samples, some of those 10-flip samples are going to be nearly all heads, some nearly all tails, just based on chance. But if you look at 50 samples of 100-flips, 1000-flips, 10,000-flips, what youll find is the larger the sample, the less of a spread -- 10-flip samples might have a spread of say .1 to .9 (meaning the proportion of heads), 100-flip samples might range from .35 to .65, 1000-flip samples from .45 to .55, etc. (Im making the numbers up to illustrate the point -- Im too lazy to calculate what the actual ranges for these would look like) I assume all of this seems obvious.

Raw +/- is like looking at 10-flip samples in the example above. Its why people look at multiple years of +/-, to try and reduce the variance/noise, which is like looking at 100 or 1,000-flip samples in the example.

As mentioned by someone else in this thread, adjusted plus-minus models (as opposed to raw plus-minus) run a regression to adjust the value based on other factors, namely who they're playing with. Some of the fancier models start taking into account all sorts of other things that they then try to correct for, like who they're playing against, etc. I cant remember which model, but one for example even corrects for technical free throws.

As for how you can get a situation where Lebron would be dead last using GeorgeMarcus's example above, you mentioned that the G-leaguers should all be worse than LeBron in his example. That's true on a per-minute basis, but not true on an aggregate basis, because you're assuming there were only 4 G-Leaguers and they were playing the same number of minutes as Lebron. If there were say 10 G-Leaguers + Lebron holding down the fort while the other 4 All-Stars were out, and Lebron were playing significantly more minutes than any of them, in the aggregate, he could have a larger negative +/- than any of them individually. To put some numbers to it to make the example more clear, say Lebron's true per-minute +/- is +0.1 (about +5 over a full game), and the G-Leaguers are each -0.1 per minute. The lineup would be -0.4 on the whole. So say Lebron plays 40 minutes, and each of the G-Leaguers plays 20 minutes, he'd have a -16, and they'd each have a -8. Once the juggernaut lineup came back, if they had missed a lot of games, they might not be a large enough aggregate positive to get Lebron back above the -8 of the other G-Leaguers, making the other all-stars all very positive +/-, the G-Leaguers negatives but not as negative as Lebron, and Lebron the worst +/- on the team, despite having a "true" positive per-minute +/-.

To give a more real-world hypothetical, say you had a bench unit that was lacking playmaking. You might start Lebron, play him for a few minutes with the starters, then sub him out say around the 4m mark. Let the starters and a 6th man play together for another say 4m, then bring in the subs at the 8m mark, along with Lebron (to make sure the bench unit has a playmaker/scorer), and that lineup finishes out the quarter, maybe making some subs along the way for different bench players. It still looks in the boxscore like Lebron and the other starters all played the same number of minutes, 8m, but while the other starters all got to play half their minutes with all starters, and half with 4 starters and 1 bench player, Lebron was playing half his minutes with all starters and half his minutes with 4 bench players, instead of just 1. Then, if you mix around who the bench player is that subs in as the 6th man among a bunch of different players, you can find yourself in a scenario where a lot of the bench players end up with better +/- scores than Lebron because they're playing a big chunk of their minutes with the starters as well (eg they play 4m as the 6th man, and only 2m with Lebron + 4 bench players). Then you can throw in other factors, like maybe the Lebron + 4 bench players lineup starts the 2nd q regularly, where they play against starters, and then the 4 starters + 1 bench player lineup subs in when the other team goes to their bench, because the coach prefers those matchups. This makes the Lebron lineup look even worse, and the starter lineup without Lebron look even better. And on and on. There are tons of factors that influence raw +/- data. This is why people look at/use adjusted models, because they adjust for these sorts of things.

To that point, the adjusted models say Lebron is still a positive player, just not as much as he used to be. His LEBRON for example is about +2.3 this season, compared to ~+9 at his peak. As recently as 2019, he was still nearly a +6, then +4, then a few seasons of +3, and now +2. He's a slight negative -0.3 on defense, and about +2.6 on offense. Someone earlier mentioned his defensive impact has really fallen off -- according to the LEBRON model at least, he's been a pretty average defender for awhile. In 11 of the past 12 seasons he's been between -0.3/+0.6, and its not since 2012 that he was consistently a +1 or better defender every year (he was a +0.9 in 2015, his only other good year in the past 12 years). Its mostly been his offense that has fallen off -- he used to consistently be in the +4-6 range, and peaked as high as +7, but now he's down at +2.6.

Lastly, Ill point out some lineup data I looked up. Lebron apparently has had by far the worst +/- when playing as a PF. According to 82games, Lebron is +30 when playing as a SF, +0 when playing as a C, and -86 as a PF, which I thought was interesting. Also, his worst 2-man lineup combos are with Knecht, Russell, Vincent and AD. AD and Knecht make some sense in that it means Lebron is playing PF, and both AD/Knecht were net negative +/- on the season (AD had a positive on-court, but net negative because the team was even more positive with him off the court). Russell and Vincent being two of his worst 2-man combos is interesting in that both Russell and Vincent both were net positives and on-court positives. The fact that both of them are PGs though, might make me think that LeBron does better when the ball is in his hands. Then again, last season, Russell was one of the better 2-man combos with Lebron, and these are all relatively small samples, so coincidence/luck is obviously a huge part (raw +/- is hugely noisy), so take all of this with a grain of salt.

[Djoker]

^Great post about by runtmc. The only thing I would add is that one whole regular season isn't a tiny sample. It's not a big sample but it's not tiny. Also worth noting that the typical issue with raw ON-OFF for star players is the OFF sample because that's the much smaller sample than ON. Stars tend to play a lot more than they sit so the OFF sample is typically only 600-900 minutes.

However, in Lebron's case, because he only played 34.9 mpg and missed 12 games on top of that, the OFF sample is actually quite substantial with 1502 minutes with Lebron OFF. That's still not a big sample but it's equivalent to ~31 full games. All in all I would still take the good OFF numbers without Lebron on the court with a grain of salt especially after Luka arriving because the two of them staggered a lot. But I wouldn't completely discount it.

And the Lakers being -0.5 with Lebron ON is genuinely a bad look for him. His last three seasons being -2.1, +5.1, and +4.1 ON also don't paint the picture of superstar impact. Neither do adjusted metrics like LEBRON, MAMBA etc. of course. Lebron looks like a top 15 ish player possibly fringe top 10 looking at it through the most positive lens IMO.

[Doctor MJ]

^Great post about by runtmc. The only thing I would add is that one whole regular season isn't a tiny sample. It's not a big sample but it's not tiny. Also worth noting that the typical issue with raw ON-OFF for star players is the OFF sample because that's the much smaller sample than ON. Stars tend to play a lot more than they sit so the OFF sample is typically only 600-900 minutes.

However, in Lebron's case, because he only played 34.9 mpg and missed 12 games on top of that, the OFF sample is actually quite substantial with 1502 minutes with Lebron OFF. That's still not a big sample but it's equivalent to ~31 full games. All in all I would still take the good OFF numbers without Lebron on the court with a grain of salt especially after Luka arriving because the two of them staggered a lot. But I wouldn't completely discount it.

And the Lakers being -0.5 with Lebron ON is genuinely a bad look for him. His last three seasons being -2.1, +5.1, and +4.1 ON also don't paint the picture of superstar impact. Neither do adjusted metrics like LEBRON, MAMBA etc. of course. Lebron looks like a top 15 ish player possibly fringe top 10 looking at it through the most positive lens IMO.

Beat me to it, thank Djoker.

The solution to noise is more sample.

Now, I don't just use raw +/- because there's a limit to how much sample we can get in NBA basketball (among other reasons), but generally I do see a season-long sample as something other than just noise.

[tsherkin]

^Great post about by runtmc. The only thing I would add is that one whole regular season isn't a tiny sample. It's not a big sample but it's not tiny. Also worth noting that the typical issue with raw ON-OFF for star players is the OFF sample because that's the much smaller sample than ON. Stars tend to play a lot more than they sit so the OFF sample is typically only 600-900 minutes.

However, in Lebron's case, because he only played 34.9 mpg and missed 12 games on top of that, the OFF sample is actually quite substantial with 1502 minutes with Lebron OFF. That's still not a big sample but it's equivalent to ~31 full games. All in all I would still take the good OFF numbers without Lebron on the court with a grain of salt especially after Luka arriving because the two of them staggered a lot. But I wouldn't completely discount it.

And the Lakers being -0.5 with Lebron ON is genuinely a bad look for him. His last three seasons being -2.1, +5.1, and +4.1 ON also don't paint the picture of superstar impact. Neither do adjusted metrics like LEBRON, MAMBA etc. of course. Lebron looks like a top 15 ish player possibly fringe top 10 looking at it through the most positive lens IMO.

Beat me to it, thank Djoker.

The solution to noise is more sample.

Now, I don't just use raw +/- because there's a limit to how much sample we can get in NBA basketball (among other reasons), but generally I do see a season-long sample as something other than just noise.

I think it's pretty reasonable to look at the past 252 RS games, which have him at +1.7 ON and +3.7 ON/OFF and say he's not where he used to be when he was a full superstar, for sure. It's also worth remembering that LA was experimenting with Lebron as a 5 a bunch in 2022 when he had that first negative ON rating. And he only played 56 games.

This year, I mean, he's 40. I don't think it's a grand surprise that he's tailing off some. He was coasting on D for a chunk of time to start the season, and then as you noted Doc, he's been staggered a bunch with Luka. So while I think his negative ON is probably a little overstating things, it isn't wholly inappropriate to look at his RS sample and say "yeah, he wasn't a stunner over the balance of the season."

His EPM numbers for the RS look better than his raw ON and ON/OFF, of course, at +1.8 O-EPM and -0.0 D-EPM, which left him 65th in the league in EPM.

This series, it's been pretty bad. He's at -9.7 ON and -11.5 ON/OFF, but there is almost no OFF sample from which to draw any conclusions, so that's a lot of noise, as has also been discussed.

[Doctor MJ]

^Great post about by runtmc. The only thing I would add is that one whole regular season isn't a tiny sample. It's not a big sample but it's not tiny. Also worth noting that the typical issue with raw ON-OFF for star players is the OFF sample because that's the much smaller sample than ON. Stars tend to play a lot more than they sit so the OFF sample is typically only 600-900 minutes.

However, in Lebron's case, because he only played 34.9 mpg and missed 12 games on top of that, the OFF sample is actually quite substantial with 1502 minutes with Lebron OFF. That's still not a big sample but it's equivalent to ~31 full games. All in all I would still take the good OFF numbers without Lebron on the court with a grain of salt especially after Luka arriving because the two of them staggered a lot. But I wouldn't completely discount it.

And the Lakers being -0.5 with Lebron ON is genuinely a bad look for him. His last three seasons being -2.1, +5.1, and +4.1 ON also don't paint the picture of superstar impact. Neither do adjusted metrics like LEBRON, MAMBA etc. of course. Lebron looks like a top 15 ish player possibly fringe top 10 looking at it through the most positive lens IMO.

Beat me to it, thank Djoker.

The solution to noise is more sample.

Now, I don't just use raw +/- because there's a limit to how much sample we can get in NBA basketball (among other reasons), but generally I do see a season-long sample as something other than just noise.

I think it's pretty reasonable to look at the past 252 RS games, which have him at +1.7 ON and +3.7 ON/OFF and say he's not where he used to be when he was a full superstar, for sure. It's also worth remembering that LA was experimenting with Lebron as a 5 a bunch in 2022 when he had that first negative ON rating. And he only played 56 games.

This year, I mean, he's 40. I don't think it's a grand surprise that he's tailing off some. He was coasting on D for a chunk of time to start the season, and then as you noted Doc, he's been staggered a bunch with Luka. So while I think his negative ON is probably a little overstating things, it isn't wholly inappropriate to look at his RS sample and say "yeah, he wasn't a stunner over the balance of the season."

His EPM numbers for the RS look better than his raw ON and ON/OFF, of course, at +1.8 O-EPM and -0.0 D-EPM, which left him 65th in the league in EPM.

This series, it's been pretty bad. He's at -9.7 ON and -11.5 ON/OFF, but there is almost no OFF sample from which to draw any conclusions, so that's a lot of noise, as has also been discussed.

Reasonable thoughts. I'm not going assert defiantly that LeBron had a negative average overall impact this year, to me the key things are a) I think LeBron can be a major 2-way positive when fully engaged, but b) old guys due tend to conserve their energy and so it's entirely possible that LeBron's play over the course of the year tipped negative even if that doesn't reflect what he can do when he's putting 110% into a critical game.

I thought the Olympics were so telling. LeBron & Steph are nowhere near as good 82 game players as Tatum right now, but if rested and at a critical moment, both may well still be better than Tatum.

[tsherkin]

Reasonable thoughts. I'm not going assert defiantly that LeBron had a negative average overall impact this year, to me the key things are a) I think LeBron can be a major 2-way positive when fully engaged, but b) old guys due tend to conserve their energy and so it's entirely possible that LeBron's play over the course of the year tipped negative even if that doesn't reflect what he can do when he's putting 110% into a critical game.

I absolutely believe he was metering his defensive effort. He basically provided none at the start of the year and dialed it up later on. He's older, he's geriatric by NBA standards, so that's quite well expected, I agree.

I thought the Olympics were so telling. LeBron & Steph are nowhere near as good 82 game players as Tatum right now, but if rested and at a critical moment, both may well still be better than Tatum.

Yep. Honestly, Tatum is a contemporary player. If his 3 isn't falling, he struggles a lot. He is actually a pretty good isolation scorer who draws fouls well, but his entire game is so patterned on top of 3pt volume and he isn't an elite ATB 3pt shooter, so it makes me sad. He could be a lot better. Lebron, of course, is still a humongous 6'9 man who sees the floor very well and has a good series of scoring tools, but as we've agreed, he needs to have his gas tank metered out pretty carefully at this age.

Unfortunately, LA has no depth worth mentioning, so against stiff competition, they're in deep crap and have to over-rely on the old man. Especially when Luka and Reaves come up so dry late in the game.

[MrPainfulTruth]

^Great post about by runtmc. The only thing I would add is that one whole regular season isn't a tiny sample. It's not a big sample but it's not tiny. Also worth noting that the typical issue with raw ON-OFF for star players is the OFF sample because that's the much smaller sample than ON. Stars tend to play a lot more than they sit so the OFF sample is typically only 600-900 minutes.

However, in Lebron's case, because he only played 34.9 mpg and missed 12 games on top of that, the OFF sample is actually quite substantial with 1502 minutes with Lebron OFF. That's still not a big sample but it's equivalent to ~31 full games. All in all I would still take the good OFF numbers without Lebron on the court with a grain of salt especially after Luka arriving because the two of them staggered a lot. But I wouldn't completely discount it.

And the Lakers being -0.5 with Lebron ON is genuinely a bad look for him. His last three seasons being -2.1, +5.1, and +4.1 ON also don't paint the picture of superstar impact. Neither do adjusted metrics like LEBRON, MAMBA etc. of course. Lebron looks like a top 15 ish player possibly fringe top 10 looking at it through the most positive lens IMO.

I dont think an entire season is a small sample, especially not when you get the data of the entire roster - its the complete picture. I will have to re-read the very detailed post by runtmc, i also greatly appreciate the work he put into it. I dont think more seasons will help, rather the opposite - its panel data and the hypothesis is that LeBron's contribution has regressed into the negative, so how would adding more seasons (neccessarily older ones) help prove that right or wrong? It is evident that he regressed, we dont really need to examine that. The point in question is wether he is objectively detrimental to his team (see OP).

[parapooper]

Tatum had a negative on/off over the last 2 season's combined, and gets 2 all-nba 1st team selections
LeBron had a positive on/off over the last 2 season's combined, but only gets a 3rd + 2nd all-NBA team - what gives?

According to this thread Tatum only cares about stat padding and plays no defense. The Celtics should trade him. Paying a player who makes his team worse $60M/year is insanity. Boston's best player is clearly Luke Kornet.

Ant also has a negative on/off this season, the wolves should get rid of him and build the team around positive on/off stud Nickeil Alexander-Walker.

[MrPainfulTruth]

Beat me to it, thank Djoker.

The solution to noise is more sample.

Now, I don't just use raw +/- because there's a limit to how much sample we can get in NBA basketball (among other reasons), but generally I do see a season-long sample as something other than just noise.

I think it's pretty reasonable to look at the past 252 RS games, which have him at +1.7 ON and +3.7 ON/OFF and say he's not where he used to be when he was a full superstar, for sure. It's also worth remembering that LA was experimenting with Lebron as a 5 a bunch in 2022 when he had that first negative ON rating. And he only played 56 games.

This year, I mean, he's 40. I don't think it's a grand surprise that he's tailing off some. He was coasting on D for a chunk of time to start the season, and then as you noted Doc, he's been staggered a bunch with Luka. So while I think his negative ON is probably a little overstating things, it isn't wholly inappropriate to look at his RS sample and say "yeah, he wasn't a stunner over the balance of the season."

His EPM numbers for the RS look better than his raw ON and ON/OFF, of course, at +1.8 O-EPM and -0.0 D-EPM, which left him 65th in the league in EPM.

This series, it's been pretty bad. He's at -9.7 ON and -11.5 ON/OFF, but there is almost no OFF sample from which to draw any conclusions, so that's a lot of noise, as has also been discussed.

Reasonable thoughts. I'm not going assert defiantly that LeBron had a negative average overall impact this year, to me the key things are a) I think LeBron can be a major 2-way positive when fully engaged, but b) old guys due tend to conserve their energy and so it's entirely possible that LeBron's play over the course of the year tipped negative even if that doesn't reflect what he can do when he's putting 110% into a critical game.

I thought the Olympics were so telling. LeBron & Steph are nowhere near as good 82 game players as Tatum right now, but if rested and at a critical moment, both may well still be better than Tatum.

That would be the way of nature and not worth discussing. The interesting point imho is that he is very selective with his energy management for a couple of years. Its not like he played on 60% all the time. There is a very evident difference in energy he "invests" bezween offense and defense. I believe he can instantly turn into a net positive when he decides to turn it on on D, certainly sacrificing some offensive output. For this roster, his "utility function" as combination of offensive impact * offensive energy + defensive impact * defensive energy , optimized for probability of team success would certainly be leaning more towards defense than he is choosing. I think that is the major debate - he produces "impressive" offensive numbers for his age while the team could be performing a lot better would he play more defense instead. The eye test confirms the thesis and we all know that he is trying to set records for the ages.

[NyKnicks1714]

On/off and +/- CAN be valuable over a large sample size, which we do have here, but it can also be faulty. If it goes against every other advanced (and traditional) metric, which is the case here, then yeah it's likely a product of lineups.

Examples: Jalen Williams had an On/off of -8.0 this season. Jalen Brunson had an on/off of -4.9. Obviously as with LeBron, both teams are much better off with those players on the floor than off the floor, despite what the numbers say.

This is why we don't assess players based on single statistics...and it's crazy that this even needs to be said.

[zimpy27]

Because his role was to make the Lakers terrible bench depth sustainable for stretches all year.


Even with this terrible plus-minis he was top 30 in the impact stats that use this plus-minis and adjust for teammates.

[therealbig3]

I feel so bad for this vast cabal of posters who only exist to hate Lebron James. The best basketball player of all-time who has been the face of this league we love for over 20 years with nary a scandal and elite play from damn near stem to stern.

I can't imagine how much one's own life has to suck to spend this much constant time and energy attacking a guy like LeBron. But I sure hope all of you meet someone, or re-connect with family, or find a cause, or just a good dog. You need some positivity in life. This ain't healthy.

You mean the best basketball player “you” have ever seen? You speak for everyone?

Lmao pretty sure he’s seen Jordan.

Btw, most people that believe LeBron is the GOAT have also seen Jordan. Shocking I know. And I know it just shatters the #1 argument people have against LeBron: “well you just haven’t seen Jordan”.