Top 25 peaks of the 2001-25: #15-#16 Spots

[f4p]

You don't think the box score likes nash enough, but like I said, am I supposed to believe harden, who is one of i think 10 players ever with a box score triple crown in 2018, was less impactful than Nash but just magically created a team that practically never lost when healthy? He just got lucky that eric Gordon and PJ tucker, who had each played on exactly one +0.500 team before they joined harden, were actually the key to it all? Or that capela, who hasn't done jack since leaving harden, was the secret? The team ran basically their entire offense through a guy who wasn't impactful, and that guy wasn't helping on defense, but 65 wins just fell out of the sky. The 2019 rockets had cp3 looking like trash, had a replacement level bench, and they won 53, crushed a +5 team in the 1st round, and gave a team with 2 (maybe 3) peaks above harden, all they wanted but weirdly it wasnt the guy putting up 35/7 driving it. If Nash looks weak by the box score, I argue regular season RAPM underrates harden even more, especially since the playoff stuff tends to really like him.

As to Harden, this is the second thread that Harden has been on my ballot, so obviously I think he was a great player. But I do think some of what you’re saying is a bit contrary to what actually happened. You say Harden “magically created a team that practically never lost when healthy” and say somewhat sarcastically that “The team ran basically their entire offense through a guy who wasn’t impactful, and that guy wasn’t helping on defense, but 65 wins just fell out of the sky.” But it’s important to recognize that the 2018 Rockets had a +4.92 net rating when Harden was off the court. So yeah, while Harden was certainly impactful, I’d say that team functioned very well even without Harden, regardless of what you happen to think about the other players on the team.

And if we talk about those games with Chris Paul, where the Rockets went an insane 44-5, the Rockets were +12.14 in Harden’s minutes. That’s really good, but it’s not an on-court number that would usually result in a 74-win pace. A huge thing that drove that 44-5 record was that the Rockets had a +4.94 net rating even without Harden on the floor. This is incredible for a team to do without their best player. For reference, I posted about how great teams generally do without their best player here: https://forums.realgm.com/boards/viewtopic.php?p=119652544#p119652544. The upshot is that great players on historically great teams usually have teams that have about a -2 net rating with them off the court.

Didn't the thunder just play +6 without Shai and the 2024 Celtics play +11 without Tatum?

That post of mine noted that Duncan was really lucky to have a team that had a positive net rating with him off the court, but even the Spurs only produced a +2.70 net rating overall in the data we have for them without Duncan. So yeah, that +4.94 net rating without Harden on the floor was abnormally good. It was probably partly just luck that they did that well, but luck or not it certainly still plays into how well the team did and isn’t something we can credit to Harden.

And, by the way, even in the 23 games Harden played without Chris Paul in that season, the Rockets still had a +3.04 net rating without Harden on the floor (compared to +6.34 with Harden on the floor). Even that is better than what we generally see from historically great teams without their best player.

Seems like James harden was just incredibly lucky to play with so many guys who were so great right when they played with him. I mean even without cp3 and harden they were playing +3 basketball? That's a 49 win team. I feel bad for all those casinos who set the rockets over under at 56. Guess they either thought harden and Paul were only worth 7 wins or they didn't realize what a juggernaut cast we had on our hands. Eric Gordon winning like 30% of his career games and then jumping to a +9.9 on/off waa great. And PJ going from loser teams and +0 on/offs right up to +6 for a season. Good for james.

I think you posted the EPM numbers that had eric Gordon at Draymond's level (3.0 for Gordon vs 2.8) and capela at KDs level ( 3.9 for capela vs 4.2). Very fortunate for James that 2 of his role player teammates were actually superstars. I guess that would help create an all time team.

Although having said how great all of hardens teammates actually were, 2017 to 2021 hardens on offs do look quite suspicious. Like 2017 to 2019 is almost the worst 3 year stretch of hardens prime. Right when Mike Dantoni, an offensive genius, said I need to build an entire system around harden and make him one of the highest usage guys ever (to the point that he was basically the genesis of the heliocentric term), and the team suddenly spiked in wins and playoff success with that system, it apparently actually all happened as hardens impact was significantly decreasing. Amazing for so many other guys to simultaneously get good and pick up the slack for harden. Of course it was quite unfortunate for hardens championship chances that that +5 2018 supporting cast decided to play -13 ball in the playoffs. Eric Gordon and his career -8 playoff on/off apparently turned back into a pumpkin (-8.7).

I mean even the 2021 nets, who went 29-7 with harden and 19-17 without, for a +22 WOWY for harden, somehow resulted in a +0.2 on/off. How many primary guys with a +22 on a game level somehow ended up with 0 on/off on a lineup level? Incredible that the team became a juggernaut when harden played (80% of the time with only 1 or 0 of KD/Kyrie) but it really wasnt harden again. The team was somehow either 0.500 while playing +5 basketball in the games harden didn't play or apparently dominating non-harden minutes in the other games so hard that they were winning at a 66 win pace.

Kind of feels like harden just gets used in lineup situations that aren't necessarily great for his RAPM (at least in the regular season) but lead to his teams winning far more games than basically anyone expected in 2017, 2018, 2019, and 2021. While other players get used in ways that result in crazy on/offs that should seemingly have their team winning 75 games if the supporting cast had literally any NBA level players (like say jokic sporting +20 on/offs for several years, not that I'm putting harden up there with jokic, which means he could be playing with the -10 Process 2015 76ers and still be creating +10 2015 warriors teams). Harden significantly over over performed the over unders in 2017 and 2018 and 2021 (in games harden played) and would have in 2019 if anyone knew the rockets bench would be gerald green and James Ennis and guys like Austin rivers and Kenneth Faried would both be starters in more than 1 game. But on/off says he was just kind of "some guy" helping out where he could.

[LA Bird]

Pop quiz: match the following 3 year RAPM peak to LeBron's teammates (Wade/Bosh, Kyrie/Love, Davis)

3rd
5th
11th
25th
26th

Spoiler:

Answer from best to worst: 10 Wade, 24 Irving, 10 Bosh, 18 Love, 20 Davis

[lessthanjake]

well i don't think anyone said he wasn't great. since we're bringing up box score stuff, all of the pace and league environment stuff is already baked into the big 3 stats from BBRef. Nash looks good but still not spectacular.

23.4 PER
0.164 WS48
4.7 BPM

WS48 is somewhat telling in that WS loves efficiency and nash was incredibly efficient and yet ended up with the same WS48 as 2018 harden, who had one of his lowest efficiency playoffs (54.8 TS%) playing the #1 regular season defense (Utah) for 5 games and the #1 playoff defense (Golden State) for 7 games. and yet, presumably because defense and other factors, harden still basically tied Nash in WS48 (0.163). and crushed him in BPM (8.1).

in terms of overall career, Nash's numbers would rank as follow in harden's career (only 9+ game playoffs counted for harden):

PER: 6th
WS48: 7th (9th with shorter runs included)
BPM: 8th (11th with shorter runs included)


Although I feel like there's an upper limit to how good we can consider Dallas since they're also the team that lost Nash so both they and Nash can't be simultaneously amazing, even if we bumped up his numbers a little for the 2 series you mentioned to 24.9 PER, 0.193 WS48, 7.0 BPM (purposely goal-seeked number, see below), it would be:

PER: 3rd (2018 harden's number behind 2019 and 2020)
WS48: 5th (2011 harden's number behind 2015, 2019, 2020, and 2021)
BPM: 6th (2017 harden's number behind 2015, 2016, 2018, 2019, 2020, and 2021)

So this is part of my point. The “Nash isn’t good by box score” stuff isn’t really borne out of looking at Nash’s raw box score numbers. It’s borne out of the various box-score metrics taking numbers that look pretty great and spitting out a number that looks surprisingly mediocre. Like, even leaving aside that Nash’s 2005 playoff numbers are even better in the two series’s against great opponents, I don’t think we’d look at 23.9 PPG, 11.3 APG, and 4.8 RPG on +9.0 rTS% in the relatively-low-scoring mid-2000s and expect just 4.7 BPM, 0.164 WS/48, and 23.4 PER. Just as an example, those numbers look a whole lot like Chris Paul’s 2008 playoffs, in which he had 24.1 PPG, 11.3 APG, and 4.9 RPG on +4.5 rTS%. Indeed, Nash’s numbers actually look slightly better! Yet Chris Paul had a 11.3 BPM, 0.289 WS/48, and 30.7 PER in those playoffs. That is a massive difference! And the 2005 playoffs are just one example. We could do a similar exercise with various other RS or playoff data for Nash and see similar things. The box-score metrics not only don’t match the RAPM data, but they don’t even really match what we see in the slashlines. So there’s clearly something going on with the box-score metrics regarding Nash that pulls his numbers down a lot. I made a long post earlier positing some explanations as to what I think are the reasons for that, and why I think those reasons are dinging Nash too much. See here: https://forums.realgm.com/boards/viewtopic.php?p=119775666#p119775666. That same post also addressed a huge mechanism through which I think all box numbers (whether box metrics or just slashlines) undersell Nash’s impact in a way that I don’t think any box metrics can really fix without throwing the metrics out of whack for a ton of other players.

At first glance
...
If you think a box-score aggregate seemingly shouldn't be spitting out some numbers ... my guess is ... assuming you broadly know the shape of the model (and I'm a long way from perfect in this regard) ... you probably haven't looked closely enough.

In this comparison ...
1) pace - it's kind of alluded to in the other direction in terms of earlier play being low-scoring (though that's going to be a balance of pace and efficiency) but look at per 100 possession numbers and
23.9 ppg to 24.1 (on similar minutes)
turns to
29.9 to 32.9
do that across the board and Paul starts to open up some advantages

2) turnovers - as you note elsewhere individual turnovers and impact on team turnovers are not the exact same thing ... still purely for a box model the slashline you gave missed out on
(per 100)
5.9 turnovers to 2.5
A pretty big gulf.

3) Parsing rebounds - offensive rebounds are arguably more valuable at an individual level or at least more creditable to the individual (defensive rebounding being a team affair) so whilst total-based rebounds looks the same (though less so after accounting for pace) ... Paul is skewed more towards the offensive glass
Nash
0.8 oreb per 100, 5.3 dreb; 1.7 oreb%, 11 dreb%
Paul
1.8 oreb per 100, 4.9 dreb; 3.7 oreb%, 9.8 dreb%

4) Steals- you allude to this elsewhere and whether it's getting things right or missing something (in Nash's case or too often in general) anyway it's another clear box difference ... 1.2 steal percentage to 3.2 (or 1.2 to 3.2 per 100).


One can make more generalized arguments as to whether box-numbers are missing something on Nash and why. You have elsewhere and I'm broadly sympathetic to the argument that they are and maybe some of the whys offered.

That said in a specific case the reasons his numbers come out different (to Paul's) is perhaps clearer when you look closer at what's going in and how it might be less similar than at an incomplete first glance.

Yeah, I mean that’s kind of my point. There’s obviously *something* in the box score that drives the numbers being really different, since these numbers are just a product of a formula using box stats. But my point was that it’s pretty clearly not really driven by the main slashline stats we think about (i.e. points, rebounds, assists, and TS%), because they’re really similar. The pace point is a good one, but Nash also shot notably more efficiently, so it feels like roughly a wash in those numbers, and definitely something that would drive anything close to the massive difference in the numbers. So then the question is what *is* driving the difference. It’s definitely something! I talked about what I think those things were. And I noted why I think Nash is being penalized too much for those things.

You bring up another potential thing—which is offensive rebounds. That seems like more of an issue just in the very specific example I used, though. CP3 happened to have a bunch of offensive rebounds in that one particular playoffs and I’m sure that did drive at least a portion of the massive chasm between their box-metric outputs for those playoff runs, but this sort of thing doesn’t really explain more generally why Nash’s box-metric numbers look much lower than we’d expect, since he didn’t generally have meaningfully fewer offensive rebounds than similar players who get higher box-metric numbers.

In any event, I don’t think we’d expect that playing at a bit higher pace and having a lower proportion of rebounds be offensive rebounds would drive anything even remotely close to CP3 having 2.5x the BPM Nash did (and especially not when there’s things on Nash’s side of the ledger too, such as meaningfully higher scoring efficiency). So I really do think that the huge difference (not just in those specific playoffs but also just with Nash in general—those particular playoffs were just an illustrative example since the slashline numbers were so similar) is a result of Nash having a low number of steals and a high number of turnovers. I think these models almost certainly conclude that Nash is a big negative in terms of turnover economy. But the thing is that that’s not actually the case, because Nash actually is a seriously positive-impact player when it comes his own team’s turnovers (since he takes an abnormal portion of his team’s risks) and isn’t a particularly negative-impact player when it comes to the opponent’s turnovers (because he took a very abnormal amount of charges).

Granted, as it applies to this specific example, Nash is still worse in terms of his impact on overall turnover economy than Chris Paul is. So a model that actually correctly assessed their impact on those things genuinely *would* ding Nash compared to Chris Paul. And with the basic slashlines looking pretty similar (with CP3 having more per-possession volume and more offensive rebounds, but worse scoring efficiency), we would expect Nash to grade out somewhat lower in that example, even if Nash wasn’t being overly penalized regarding turnover economy. But the question is how much he’s being dinged for it. When it’s appearing to ding him to the point of CP3 ending up with a BPM that is 2.5x higher (!!!), then I think that strongly suggests it’s dinging him too much. And I think that is exactly what we’d expect, because almost no players with Nash’s level of turnovers are actually such a positive-impact player in terms of his team’s turnovers, and almost no players take as many charges as Nash did. So a model that’s trying to find a line of best fit is just always going to seriously underrate Nash in those areas. Basically, the model may be right in general that that difference in turnovers and steals is typically indicative of a huge difference in impact on overall turnover economy, but the actual difference in these two players’ impact on overall turnover economy just isn’t quite as big as the model surely thinks it is.

So I think that gets to why Nash’s box-metric output looks very different than other players with similar slashlines. Some portion of it is for reasons that are fair, but a lot of the difference is almost certainly that these models think his impact on overall turnover economy is far worse than it actually is.

And then we get to the question of whether the box score in general (whether box metric or slashlines) can adequately account for Nash’s effect on TS%. Multi-factor RAPM tells us that Nash’s impact on his team’s TS% was the highest on record (and I have to say that my eye test is very consistent with that). His impact in this regard was *much* higher than players with similar numbers of assists (not to mention that it’s higher than guys who are high-volume, high-efficiency scorers *and* have a lot of assists). So the box score could never give Nash enough credit for his effect on teammates’ TS% without way overrating every other player that has a lot of assists. When it comes to box metrics, they will never do that because they’re looking for a line of best fit in general (or, for less sophisticated stuff like WS or PER, they’re at least aiming to get output that looks like it makes sense for most players). That leaves Nash severely underrated. Box models just can’t really properly account for someone with such an outlier effect on teammates’ TS% without throwing the rest of the model’s output out of whack. And, of course, just comparing slashlines won’t account for it either, because Nash with an equal number of assists as someone else is probably actually driving a far bigger increase in teammate TS% than that other player. Seeing Nash and CP3 having a similar number of assists really doesn’t mean they were having an effect on teammate TS% that is at all similar. They very likely weren’t, but the box score thinks they were!

So yeah, the bottom line is that I certainly agree that the difference in the box-metric numbers is coming from *somewhere*. My point is just that I think that Nash’s impact in various areas—particularly teammate TS% and turnover economy—are far better than any box metric could ever realize, because his impact in those areas is abnormal in ways that a line of best fit using box data would never properly account for.

[trex_8063]

Wow. Did you just ignore the context? All of the Suns left and were injured in those 2003-04 Suns, but OFC you ignore it, because it doesn't fit into your biased, completely easy to debunk agenda.

I feel like you overstate their injury woes.
The ONLY especially relevant injury to the '04 Suns was Amar'e missing 27 games (Bill Walton won an MVP playing only 3 more games).
Otherwise, what are we talking about? Jake Voskul missing 16? Post-injury McDyess [bench player] missed like 24? Howard Eisley missed about 14? Leandro Barbosa missing 12 (he would miss 19 in '05, fwiw)?

Those are the only other losses among their regular rotational players (and that's not exactly an unusual amount of missed time).
Marbury and Hardaway didn't miss ANY games prior to their trade [Marbury leading the team in mpg]. Joe Johnson didn't miss ANY games, Marion only missed 3 (if you're keeping track, that's only THREE missed games among their top THREE mpg players), Casey Jacobsen missed 4.


I won't deny the '05 squad had better health, but still had some lost time: Barbosa missed 19, Nash himself missed 7, QR missed 3, Stoudemire missed 2, Marion missed 1, Jim Jackson missed 1.

Their rORTG jumped by +9.9 (from -1.5 to +8.4).


And you talk about the loss of Marbury and Hardaway contributing to this [perceived] devastatingly large gap, saying it created a vacuum that Nash was able to reconnect. You imply that if those two had been around the whole year, the shift between '04 and '05 would not be near as dramatic.

Unfortunately, the record does not support that: they were 12-22 [.353] in the 34 games prior to their trade, with a -2.4 rORTG. And Amar'e Stoudemire played in 18 of 34 games [52.9%] in this sample, btw.
AFTER they were gone, the Suns went 17-31 [.354] with a -0.8 rORTG (that's right: they were +1.6 BETTER on offense AFTER Marbury/Hardaway left). Stoudemire was around a little more in this sample (playing in 37 of 48 [77.1%]), but still....

2004 Mavs with Nash were 113.2 ORTG, and then dropped to "just" 110.3 ORTG in 2005 (this is w/o context).
With context, the 2005 Mavs with Dirk were 111.7 ORTG.

So Nash' real impact, is much, much closer to a small 1.5 points worth of offense, rather than 9 or 10 points worth on offense.

Hmm.....talk about ignoring context.
'05 was the first year with the change to hand-checking rules. League-wide average ORtg jumped by a whopping +3.2 from '04 to '05 (from 102.9 to 106.1).

To reframe the numbers you just cited in light of that:
'04 Mavs with Nash were a +10.3 rORTG. Then dropped to a +4.2 rORTG in '05 (that's a shift of -6.1).
WITH Dirk, the '05 Mavs were a +5.6 rORTG (still -4.7 short of where they'd been in '04).

[lessthanjake]

You don't think the box score likes nash enough, but like I said, am I supposed to believe harden, who is one of i think 10 players ever with a box score triple crown in 2018, was less impactful than Nash but just magically created a team that practically never lost when healthy? He just got lucky that eric Gordon and PJ tucker, who had each played on exactly one +0.500 team before they joined harden, were actually the key to it all? Or that capela, who hasn't done jack since leaving harden, was the secret? The team ran basically their entire offense through a guy who wasn't impactful, and that guy wasn't helping on defense, but 65 wins just fell out of the sky. The 2019 rockets had cp3 looking like trash, had a replacement level bench, and they won 53, crushed a +5 team in the 1st round, and gave a team with 2 (maybe 3) peaks above harden, all they wanted but weirdly it wasnt the guy putting up 35/7 driving it. If Nash looks weak by the box score, I argue regular season RAPM underrates harden even more, especially since the playoff stuff tends to really like him.

As to Harden, this is the second thread that Harden has been on my ballot, so obviously I think he was a great player. But I do think some of what you’re saying is a bit contrary to what actually happened. You say Harden “magically created a team that practically never lost when healthy” and say somewhat sarcastically that “The team ran basically their entire offense through a guy who wasn’t impactful, and that guy wasn’t helping on defense, but 65 wins just fell out of the sky.” But it’s important to recognize that the 2018 Rockets had a +4.92 net rating when Harden was off the court. So yeah, while Harden was certainly impactful, I’d say that team functioned very well even without Harden, regardless of what you happen to think about the other players on the team.

And if we talk about those games with Chris Paul, where the Rockets went an insane 44-5, the Rockets were +12.14 in Harden’s minutes. That’s really good, but it’s not an on-court number that would usually result in a 74-win pace. A huge thing that drove that 44-5 record was that the Rockets had a +4.94 net rating even without Harden on the floor. This is incredible for a team to do without their best player. For reference, I posted about how great teams generally do without their best player here: https://forums.realgm.com/boards/viewtopic.php?p=119652544#p119652544. The upshot is that great players on historically great teams usually have teams that have about a -2 net rating with them off the court.

Didn't the thunder just play +6 without Shai and the 2024 Celtics play +11 without Tatum?

Yeah, the 2025 Thunder were +5.14 without Shai, and the 2024 Celtics were an absurd +12.24 without Tatum. I don’t think it should be news to anyone that those two players had extremely strong supporting casts those years!

It’s worth noting, though, that the 2025 Thunder were a massive +16.06 with SGA on the court. So the difference between how well the 2025 Thunder and 2018 Rockets did (both great teams, but the 2025 Thunder did do even better) is basically driven by the Thunder doing materially better in SGA’s minutes than the Rockets did in Harden’s minutes, rather than there being any material difference in how the teams did with those guys off the court. The same is definitely not true of Tatum, but I personally would never have Tatum above Harden (and you’ll find I’ve argued in this very thread that the 2024 Celtics supporting cast was fantastic).

That post of mine noted that Duncan was really lucky to have a team that had a positive net rating with him off the court, but even the Spurs only produced a +2.70 net rating overall in the data we have for them without Duncan. So yeah, that +4.94 net rating without Harden on the floor was abnormally good. It was probably partly just luck that they did that well, but luck or not it certainly still plays into how well the team did and isn’t something we can credit to Harden.

And, by the way, even in the 23 games Harden played without Chris Paul in that season, the Rockets still had a +3.04 net rating without Harden on the floor (compared to +6.34 with Harden on the floor). Even that is better than what we generally see from historically great teams without their best player.

Seems like James harden was just incredibly lucky to play with so many guys who were so great right when they played with him. I mean even without cp3 and harden they were playing +3 basketball? That's a 49 win team. I feel bad for all those casinos who set the rockets over under at 56. Guess they either thought harden and Paul were only worth 7 wins or they didn't realize what a juggernaut cast we had on our hands. Eric Gordon winning like 30% of his career games and then jumping to a +9.9 on/off waa great. And PJ going from loser teams and +0 on/offs right up to +6 for a season. Good for james.

I think you posted the EPM numbers that had eric Gordon at Draymond's level (3.0 for Gordon vs 2.8) and capela at KDs level ( 3.9 for capela vs 4.2). Very fortunate for James that 2 of his role player teammates were actually superstars. I guess that would help create an all time team.

Although having said how great all of hardens teammates actually were, 2017 to 2021 hardens on offs do look quite suspicious. Like 2017 to 2019 is almost the worst 3 year stretch of hardens prime. Right when Mike Dantoni, an offensive genius, said I need to build an entire system around harden and make him one of the highest usage guys ever (to the point that he was basically the genesis of the heliocentric term), and the team suddenly spiked in wins and playoff success with that system, it apparently actually all happened as hardens impact was significantly decreasing. Amazing for so many other guys to simultaneously get good and pick up the slack for harden. Of course it was quite unfortunate for hardens championship chances that that +5 2018 supporting cast decided to play -13 ball in the playoffs. Eric Gordon and his career -8 playoff on/off apparently turned back into a pumpkin (-8.7).

I mean even the 2021 nets, who went 29-7 with harden and 19-17 without, for a +22 WOWY for harden, somehow resulted in a +0.2 on/off. How many primary guys with a +22 on a game level somehow ended up with 0 on/off on a lineup level? Incredible that the team became a juggernaut when harden played (80% of the time with only 1 or 0 of KD/Kyrie) but it really wasnt harden again. The team was somehow either 0.500 while playing +5 basketball in the games harden didn't play or apparently dominating non-harden minutes in the other games so hard that they were winning at a 66 win pace.

Kind of feels like harden just gets used in lineup situations that aren't necessarily great for his RAPM (at least in the regular season) but lead to his teams winning far more games than basically anyone expected in 2017, 2018, 2019, and 2021. While other players get used in ways that result in crazy on/offs that should seemingly have their team winning 75 games if the supporting cast had literally any NBA level players (like say jokic sporting +20 on/offs for several years, not that I'm putting harden up there with jokic, which means he could be playing with the -10 Process 2015 76ers and still be creating +10 2015 warriors teams). Harden significantly over over performed the over unders in 2017 and 2018 and 2021 (in games harden played) and would have in 2019 if anyone knew the rockets bench would be gerald green and James Ennis and guys like Austin rivers and Kenneth Faried would both be starters in more than 1 game. But on/off says he was just kind of "some guy" helping out where he could.

So I think the theory that Harden “just gets used in lineup situations that aren’t necessarily great for his RAPM” is interesting, but you don’t put any meat on the bones here. What were these lineup situations and why were they not great for his RAPM? What does it even mean to be in lineup situations that aren’t great for RAPM? Is it that the Harden lineups often have a bad fit? That the non-Harden lineups have a much better fit than the lineups other teams put out without their star? If so, do you have any actual specifics about why you think those things?

I have to say that, on its face, it seems a bit hard to just attribute this stuff to lineup effects, given that the 2018 & 2019 Rockets had a +2.35 net rating even just in the 14 games that Harden completely missed, and it was +5.76 in the 10 games he missed in 2018 specifically. Of course, those are pretty small sample sizes of games so I wouldn’t put much importance on the exact numbers, but if we’re going to posit that lineup effects are the reason that those teams did so well with Harden off the floor, I think we’d really want to see the Rockets doing badly in full games without Harden. And that’s not what we saw.

[Top10alltime]

Wow. Did you just ignore the context? All of the Suns left and were injured in those 2003-04 Suns, but OFC you ignore it, because it doesn't fit into your biased, completely easy to debunk agenda.

I feel like you overstate their injury woes.
The ONLY especially relevant injury to the '04 Suns was Amar'e missing 27 games (Bill Walton won an MVP playing only 3 more games).
Otherwise, what are we talking about? Jake Voskul missing 16? Post-injury McDyess [bench player] missed like 24? Howard Eisley missed about 14? Leandro Barbosa missing 12 (he would miss 19 in '05, fwiw)?

Those are the only other losses among their regular rotational players (and that's not exactly an unusual amount of missed time).
Marbury and Hardaway didn't miss ANY games prior to their trade [Marbury leading the team in mpg]. Joe Johnson didn't miss ANY games, Marion only missed 3 (if you're keeping track, that's only THREE missed games among their top THREE mpg players), Casey Jacobsen missed 4.


I won't deny the '05 squad had better health, but still had some lost time: Barbosa missed 19, Nash himself missed 7, QR missed 3, Stoudemire missed 2, Marion missed 1, Jim Jackson missed 1.

Their rORTG jumped by +9.9 (from -1.5 to +8.4).


And you talk about the loss of Marbury and Hardaway contributing to this [perceived] devastatingly large gap, saying it created a vacuum that Nash was able to reconnect. You imply that if those two had been around the whole year, the shift between '04 and '05 would not be near as dramatic.

Unfortunately, the record does not support that: they were 12-22 [.353] in the 34 games prior to their trade, with a -2.4 rORTG. And Amar'e Stoudemire played in 18 of 34 games [52.9%] in this sample, btw.
AFTER they were gone, the Suns went 17-31 [.354] with a -0.8 rORTG (that's right: they were +1.6 BETTER on offense AFTER Marbury/Hardaway left). Stoudemire was around a little more in this sample (playing in 37 of 48 [77.1%]), but still....

I can't believe I am doing this. I am NOT overstating the injuries. Amar'e missing 30 was HUGE for the team, (-2.6 without him offensively), and they were a positive team with him. That alone is terrible, and then you have guys like Marbury and Hardaway missing games too. '05 squad with FAR better health, overstates Nash' extremely overrated impact (in reality, it's at KD's impact level). Marion missed more games than he did in the '05 season.

The lost time and injuries is FAR worse, and clearly hurt the team. Nash impact is closer to a negative than 9.9 points worth. He just isn't that impactful, and the context was needed.

You yourself proved why Marbury+Hardaway numbers are misleading, thank you for letting me do less work!

2004 Mavs with Nash were 113.2 ORTG, and then dropped to "just" 110.3 ORTG in 2005 (this is w/o context).
With context, the 2005 Mavs with Dirk were 111.7 ORTG.

So Nash' real impact, is much, much closer to a small 1.5 points worth of offense, rather than 9 or 10 points worth on offense.

Hmm.....talk about ignoring context.
'05 was the first year with the change to hand-checking rules. League-wide average ORtg jumped by a whopping +3.2 from '04 to '05 (from 102.9 to 106.1).

To reframe the numbers you just cited in light of that:
'04 Mavs with Nash were a +10.3 rORTG. Then dropped to a +4.2 rORTG in '05 (that's a shift of -6.1).
WITH Dirk, the '05 Mavs were a +5.6 rORTG (still -4.7 short of where they'd been in '04).

Well, another bad comparison. You look at actual ORTG (which is actually better in 04, or worse-looking in 05), not rORtg, because there are changes to how good the ORTG was year by year. Bad comparison, trex

2004 Mavs ORTG with Nash: 113.2
2005 Mavs ORTG with Dirk: 111.7

It's an easy argument to fight against Nash at this point. I'll say you did better than the last guy though

[lessthanjake]

Wow. Did you just ignore the context? All of the Suns left and were injured in those 2003-04 Suns, but OFC you ignore it, because it doesn't fit into your biased, completely easy to debunk agenda.

I feel like you overstate their injury woes.
The ONLY especially relevant injury to the '04 Suns was Amar'e missing 27 games (Bill Walton won an MVP playing only 3 more games).
Otherwise, what are we talking about? Jake Voskul missing 16? Post-injury McDyess [bench player] missed like 24? Howard Eisley missed about 14? Leandro Barbosa missing 12 (he would miss 19 in '05, fwiw)?

Those are the only other losses among their regular rotational players (and that's not exactly an unusual amount of missed time).
Marbury and Hardaway didn't miss ANY games prior to their trade [Marbury leading the team in mpg]. Joe Johnson didn't miss ANY games, Marion only missed 3 (if you're keeping track, that's only THREE missed games among their top THREE mpg players), Casey Jacobsen missed 4.


I won't deny the '05 squad had better health, but still had some lost time: Barbosa missed 19, Nash himself missed 7, QR missed 3, Stoudemire missed 2, Marion missed 1, Jim Jackson missed 1.

Their rORTG jumped by +9.9 (from -1.5 to +8.4).


And you talk about the loss of Marbury and Hardaway contributing to this [perceived] devastatingly large gap, saying it created a vacuum that Nash was able to reconnect. You imply that if those two had been around the whole year, the shift between '04 and '05 would not be near as dramatic.

Unfortunately, the record does not support that: they were 12-22 [.353] in the 34 games prior to their trade, with a -2.4 rORTG. And Amar'e Stoudemire played in 18 of 34 games [52.9%] in this sample, btw.
AFTER they were gone, the Suns went 17-31 [.354] with a -0.8 rORTG (that's right: they were +1.6 BETTER on offense AFTER Marbury/Hardaway left). Stoudemire was around a little more in this sample (playing in 37 of 48 [77.1%]), but still....

I can't believe I am doing this. I am NOT overstating the injuries. Amar'e missing 30 was HUGE for the team, (-2.6 without him offensively), and they were a positive team with him.

I shouldn’t respond to this because I know you’re trolling, but just want to note one factual thing:

So this is the second time you’ve said that they were a positive rORTG team without Amare in 2004, despite me having told you that the opposite is true before you ever made this claim. As per PBPstats, the 2004 Suns had a 102.27 ORTG in games Amare played, and the league average ORTG that year was 102.54. So they had a -0.27 rORTG in games Amare played. That’s better than it was in the games Amare didn’t play, but it’s certainly nowhere near something that would justify your nonsense claim that they were the GOAT offensive supporting cast. The rORTG with Amare looks very slightly positive if we looked at BBREF data instead, but BBREF is calculating its numbers using estimated possessions, while PBPstats is actually getting the precise number of possessions from the play-by-play data, so the PBPstats data is more accurate (and, in any event, even the BBREF data really does not support your thesis).

2004 Mavs with Nash were 113.2 ORTG, and then dropped to "just" 110.3 ORTG in 2005 (this is w/o context).
With context, the 2005 Mavs with Dirk were 111.7 ORTG.

So Nash' real impact, is much, much closer to a small 1.5 points worth of offense, rather than 9 or 10 points worth on offense.

Hmm.....talk about ignoring context.
'05 was the first year with the change to hand-checking rules. League-wide average ORtg jumped by a whopping +3.2 from '04 to '05 (from 102.9 to 106.1).

To reframe the numbers you just cited in light of that:
'04 Mavs with Nash were a +10.3 rORTG. Then dropped to a +4.2 rORTG in '05 (that's a shift of -6.1).
WITH Dirk, the '05 Mavs were a +5.6 rORTG (still -4.7 short of where they'd been in '04).

Well, another bad comparison. You look at actual ORTG (which is actually better in 04, or worse-looking in 05), not rORtg, because there are changes to how good the ORTG was year by year. Bad comparison, trex

2004 Mavs ORTG with Nash: 113.2
2005 Mavs ORTG with Dirk: 111.7

It's an easy argument to fight against Nash at this point. I'll say you did better than the last guy though

Just want to note that this response is completely non-sensical. Like it’s just completely unresponsive to what you’re responding to and appears to advocate using rORTG instead of ORTG (though it’s also just hard to follow what you’re saying) but then follows that immediately by reciting the same ORTG numbers that trex had already explained ignored very important context.

[Cavsfansince84]

At this point it feels like there's a real wall of dissonance between some people who are strongly pro rapm 3-5 year samples as the ultimate baseline for player goodness and those who aren't in that camp. Personally, I'm sort of losing interest in debating any of it anymore so I think this will be my last thread until at least the 76-00 period begins. I'm just wondering what method(s) the most pro rapm voters are going to use in the other two periods of this project.

[Top10alltime]

I feel like you overstate their injury woes.
The ONLY especially relevant injury to the '04 Suns was Amar'e missing 27 games (Bill Walton won an MVP playing only 3 more games).
Otherwise, what are we talking about? Jake Voskul missing 16? Post-injury McDyess [bench player] missed like 24? Howard Eisley missed about 14? Leandro Barbosa missing 12 (he would miss 19 in '05, fwiw)?

Those are the only other losses among their regular rotational players (and that's not exactly an unusual amount of missed time).
Marbury and Hardaway didn't miss ANY games prior to their trade [Marbury leading the team in mpg]. Joe Johnson didn't miss ANY games, Marion only missed 3 (if you're keeping track, that's only THREE missed games among their top THREE mpg players), Casey Jacobsen missed 4.


I won't deny the '05 squad had better health, but still had some lost time: Barbosa missed 19, Nash himself missed 7, QR missed 3, Stoudemire missed 2, Marion missed 1, Jim Jackson missed 1.

Their rORTG jumped by +9.9 (from -1.5 to +8.4).


And you talk about the loss of Marbury and Hardaway contributing to this [perceived] devastatingly large gap, saying it created a vacuum that Nash was able to reconnect. You imply that if those two had been around the whole year, the shift between '04 and '05 would not be near as dramatic.

Unfortunately, the record does not support that: they were 12-22 [.353] in the 34 games prior to their trade, with a -2.4 rORTG. And Amar'e Stoudemire played in 18 of 34 games [52.9%] in this sample, btw.
AFTER they were gone, the Suns went 17-31 [.354] with a -0.8 rORTG (that's right: they were +1.6 BETTER on offense AFTER Marbury/Hardaway left). Stoudemire was around a little more in this sample (playing in 37 of 48 [77.1%]), but still....

I can't believe I am doing this. I am NOT overstating the injuries. Amar'e missing 30 was HUGE for the team, (-2.6 without him offensively), and they were a positive team with him.

I shouldn’t respond to this because I know you’re trolling, but just want to note one factual thing:

So this is the second time you’ve said that they were a positive rORTG team without Amare in 2004, despite me having told you that the opposite is true before you ever made this claim. As per PBPstats, the 2004 Suns had a 102.27 ORTG in games Amare played, and the league average ORTG that year was 102.54. So they had a -0.27 rORTG in games Amare played. That’s better than it was in the games Amare didn’t play, but it’s certainly nowhere near something that would justify your nonsense claim that they were the GOAT offensive supporting cast. The rORTG with Amare looks very slightly positive if we looked at BBREF data instead, but BBREF is calculating its numbers using estimated possessions, while PBPstats is actually getting the precise number of possessions from the play-by-play data, so the PBPstats data is more accurate (and, in any event, even the BBREF data really does not support your thesis).

I'm trolling, and fried you in the process? Good jokes.

Also I use databallr.com, not PBPstats. That shows Amare on-court with a +rORtg on-court

Hmm.....talk about ignoring context.
'05 was the first year with the change to hand-checking rules. League-wide average ORtg jumped by a whopping +3.2 from '04 to '05 (from 102.9 to 106.1).

To reframe the numbers you just cited in light of that:
'04 Mavs with Nash were a +10.3 rORTG. Then dropped to a +4.2 rORTG in '05 (that's a shift of -6.1).
WITH Dirk, the '05 Mavs were a +5.6 rORTG (still -4.7 short of where they'd been in '04).

Well, another bad comparison. You look at actual ORTG (which is actually better in 04, or worse-looking in 05), not rORtg, because there are changes to how good the ORTG was year by year. Bad comparison, trex

2004 Mavs ORTG with Nash: 113.2
2005 Mavs ORTG with Dirk: 111.7

It's an easy argument to fight against Nash at this point. I'll say you did better than the last guy though

Just want to note that this response is completely non-sensical. Like it’s just completely unresponsive to what you’re responding to and appears to advocate using rORTG instead of ORTG (though it’s also just hard to follow what you’re saying) but then follows that immediately by reciting the same ORTG numbers that trex had already explained ignored very important context.

That context doesn't work in that case. It just makes the '05 numbers look absolutely, completely worse. I advocated using ORTG, to not mislead anyone like Trex is doing right here to you (you have a clear biased agenda, and don't watch games, so it might not be Trex).

In any case, Trex' context works in my favour, not the other way around. So Nash' ORTG impact is less than 1.5 points! Thanks to Trex, God bless you for that.

[trex_8063]

I can't believe I am doing this. I am NOT overstating the injuries. Amar'e missing 30 was HUGE for the team, (-2.6 without him offensively), and they were a positive team with him.

I shouldn’t respond to this because I know you’re trolling, but just want to note one factual thing:

So this is the second time you’ve said that they were a positive rORTG team without Amare in 2004, despite me having told you that the opposite is true before you ever made this claim. As per PBPstats, the 2004 Suns had a 102.27 ORTG in games Amare played, and the league average ORTG that year was 102.54. So they had a -0.27 rORTG in games Amare played.

This more or less concurs with the more crude data from bbref (which is what I was sourcing): it shows the Suns as averaging a -0.45 rORTG in the 55 games Amar'e played in that year. It's a substantial upgrade from their rORTG without him (-3.53 rORTG in those 27 games, according to bbref advanced game-log data); but still negative.

Let us also not forget that Amar'e was an historically bad defensive PF. While their offense jumped by around +3 or so with him, they LOST half of that gain on the defensive end: +1.57 rDRTG without him, worsening to +3.05 rDRTG WITH him.

This reflects a net change of about +1.6 per 100 possessions overall. Losing a player like that for 27 games [27, not 30] is relevant, but it is not "huge".

[trex_8063]

Well, another bad comparison. You look at actual ORTG (which is actually better in 04, or worse-looking in 05), not rORtg, because there are changes to how good the ORTG was year by year. Bad comparison, trex

2004 Mavs ORTG with Nash: 113.2
2005 Mavs ORTG with Dirk: 111.7

Got it: use context except when it is inconvenient to your agenda. Can do!

Then I guess the Suns offense improved by an Earth-shaking +13.1 [not +9.9] from '04 to '05!!!

[Doctor MJ]

At this point it feels like there's a real wall of dissonance between some people who are strongly pro rapm 3-5 year samples as the ultimate baseline for player goodness and those who aren't in that camp. Personally, I'm sort of losing interest in debating any of it anymore so I think this will be my last thread until at least the 76-00 period begins. I'm just wondering what method(s) the most pro rapm voters are going to use in the other two periods of this project.

It'll be different won't it? And that will be interesting.

When we last did the Top 100 I used a very RPOY-style approach as my anchor, and in doing so I was particularly using an approach that allowed me to be as consistent as possible across eras. Basically, looking to give the nod to those who led the top teams.

For this 2001-25 project I thought: "Might as well really focus on the +/- data.", although I have to say, I didn't follow through with that to the extent I might have.

As we go back into the deeper past, the specific tools I'll use will have to change, but I'll still be trying to be as consistent as I can with the basketball thought process that lie beneath the specific stats I used. I can't be as consistent as I would be if I had the same stats available in all years, but I'll try my best and we'll see what happens.

[iggymcfrack]

So I’ve taken in the information in this thread from nbarapm.com that the Nash supporters put up and it has changed my opinion… on Anthony Davis. The fact that his impact numbers are consistently much lower than guys with more impressive box score seasons like Westbrook and Harden makes me think that I can’t put him ahead of them.

With Nash, I’m still not 100% sold though. It seems like there’s a couple things at play in Phoenix: not only did D’Antoni develop an ahead of its time system that perfectly suited Nash, but also it was a team full of great complementary players who weren’t necessarily very good at playmaking and creating their own offense. Considering Nash’s defensive weaknesses and his need to control the ball to be effective, it really makes you wonder how well he scales up on better teams. The fact that his impact signals drop SO sharply when he’s outside of Phoenix would suggest that there’s something to that.

In order to test this, I went and looked at the multi-year RAPMs with the minimum amount of Dallas years attached. His 4 year RAPM from 2004-2007 and his 5 year RAPM from 2004-2008 would both be behind Harden’s peak stretch. Given that the box/impact metrics would also generally have Nash toward the back of the pack of current contenders, I don’t see him making my ballot this time.

[f4p]

As to Harden, this is the second thread that Harden has been on my ballot, so obviously I think he was a great player. But I do think some of what you’re saying is a bit contrary to what actually happened. You say Harden “magically created a team that practically never lost when healthy” and say somewhat sarcastically that “The team ran basically their entire offense through a guy who wasn’t impactful, and that guy wasn’t helping on defense, but 65 wins just fell out of the sky.” But it’s important to recognize that the 2018 Rockets had a +4.92 net rating when Harden was off the court. So yeah, while Harden was certainly impactful, I’d say that team functioned very well even without Harden, regardless of what you happen to think about the other players on the team.

And if we talk about those games with Chris Paul, where the Rockets went an insane 44-5, the Rockets were +12.14 in Harden’s minutes. That’s really good, but it’s not an on-court number that would usually result in a 74-win pace. A huge thing that drove that 44-5 record was that the Rockets had a +4.94 net rating even without Harden on the floor. This is incredible for a team to do without their best player. For reference, I posted about how great teams generally do without their best player here: https://forums.realgm.com/boards/viewtopic.php?p=119652544#p119652544. The upshot is that great players on historically great teams usually have teams that have about a -2 net rating with them off the court.

Didn't the thunder just play +6 without Shai and the 2024 Celtics play +11 without Tatum?

Yeah, the 2025 Thunder were +5.14 without Shai, and the 2024 Celtics were an absurd +12.24 without Tatum. I don’t think it should be news to anyone that those two players had extremely strong supporting casts those years!

It’s worth noting, though, that the 2025 Thunder were a massive +16.06 with SGA on the court. So the difference between how well the 2025 Thunder and 2018 Rockets did (both great teams, but the 2025 Thunder did do even better) is basically driven by the Thunder doing materially better in SGA’s minutes than the Rockets did in Harden’s minutes, rather than there being any material difference in how the teams did with those guys off the court. The same is definitely not true of Tatum, but I personally would never have Tatum above Harden (and you’ll find I’ve argued in this very thread that the 2024 Celtics supporting cast was fantastic).

That post of mine noted that Duncan was really lucky to have a team that had a positive net rating with him off the court, but even the Spurs only produced a +2.70 net rating overall in the data we have for them without Duncan. So yeah, that +4.94 net rating without Harden on the floor was abnormally good. It was probably partly just luck that they did that well, but luck or not it certainly still plays into how well the team did and isn’t something we can credit to Harden.

And, by the way, even in the 23 games Harden played without Chris Paul in that season, the Rockets still had a +3.04 net rating without Harden on the floor (compared to +6.34 with Harden on the floor). Even that is better than what we generally see from historically great teams without their best player.

Seems like James harden was just incredibly lucky to play with so many guys who were so great right when they played with him. I mean even without cp3 and harden they were playing +3 basketball? That's a 49 win team. I feel bad for all those casinos who set the rockets over under at 56. Guess they either thought harden and Paul were only worth 7 wins or they didn't realize what a juggernaut cast we had on our hands. Eric Gordon winning like 30% of his career games and then jumping to a +9.9 on/off waa great. And PJ going from loser teams and +0 on/offs right up to +6 for a season. Good for james.

I think you posted the EPM numbers that had eric Gordon at Draymond's level (3.0 for Gordon vs 2.8) and capela at KDs level ( 3.9 for capela vs 4.2). Very fortunate for James that 2 of his role player teammates were actually superstars. I guess that would help create an all time team.

Although having said how great all of hardens teammates actually were, 2017 to 2021 hardens on offs do look quite suspicious. Like 2017 to 2019 is almost the worst 3 year stretch of hardens prime. Right when Mike Dantoni, an offensive genius, said I need to build an entire system around harden and make him one of the highest usage guys ever (to the point that he was basically the genesis of the heliocentric term), and the team suddenly spiked in wins and playoff success with that system, it apparently actually all happened as hardens impact was significantly decreasing. Amazing for so many other guys to simultaneously get good and pick up the slack for harden. Of course it was quite unfortunate for hardens championship chances that that +5 2018 supporting cast decided to play -13 ball in the playoffs. Eric Gordon and his career -8 playoff on/off apparently turned back into a pumpkin (-8.7).

I mean even the 2021 nets, who went 29-7 with harden and 19-17 without, for a +22 WOWY for harden, somehow resulted in a +0.2 on/off. How many primary guys with a +22 on a game level somehow ended up with 0 on/off on a lineup level? Incredible that the team became a juggernaut when harden played (80% of the time with only 1 or 0 of KD/Kyrie) but it really wasnt harden again. The team was somehow either 0.500 while playing +5 basketball in the games harden didn't play or apparently dominating non-harden minutes in the other games so hard that they were winning at a 66 win pace.

Kind of feels like harden just gets used in lineup situations that aren't necessarily great for his RAPM (at least in the regular season) but lead to his teams winning far more games than basically anyone expected in 2017, 2018, 2019, and 2021. While other players get used in ways that result in crazy on/offs that should seemingly have their team winning 75 games if the supporting cast had literally any NBA level players (like say jokic sporting +20 on/offs for several years, not that I'm putting harden up there with jokic, which means he could be playing with the -10 Process 2015 76ers and still be creating +10 2015 warriors teams). Harden significantly over over performed the over unders in 2017 and 2018 and 2021 (in games harden played) and would have in 2019 if anyone knew the rockets bench would be gerald green and James Ennis and guys like Austin rivers and Kenneth Faried would both be starters in more than 1 game. But on/off says he was just kind of "some guy" helping out where he could.

So I think the theory that Harden “just gets used in lineup situations that aren’t necessarily great for his RAPM” is interesting, but you don’t put any meat on the bones here.

I mean do I need to? Unless your next statement is "I Jake, only care about on off numbers and look at nothing else" then what do I need to explain? Now to be fair, that does seem to be your position in most of our conversations (unless its the playoffs). 2014 to 2016 harden has a better on off every year than every year from 2017 to 2019. Does that make sense? Also 2022 on Philly, when harden basically sucked, and 2024 with the clippers. Does that make sense?

I mean hardens on offs almost seem to be inversely correlated with his career arc, practically bottoming out in 2017 to 2019. While something like your stat BPM (and the rest of the box score and team success) basically follows the career arc we all thought we watched, peaking in 2017 to 2020 (and 2015), when he got all his MVP votes and his teams did their best. Maybe the on off is just kind of obviously wrong? Like i said, how exactly do we explain +0.2 harden going 29-7 in Brooklyn? Joe harris? Literally everyone on the team just magically played well in the exact 50% of the games harden played?

Mike Dantoni certainly seemed to think harden was impactful or I'm guessing he wouldn't have given harden the ball as much as humanly possible. Again, for the rockets to be so good while harden was declining, not only would people have to have been way wrong about basically all of his teammates from 2017 to 2021 (very different teams), but those guys would have basically had to produce all that impact while working around a teammate who was using all the possessions. Massive impact on the possession scraps harden and Dantoni were giving them. Or...impact numbers are wrong in this case and everything else is right.

I think I looked at the harden/no CP3 and cp3/no harden stuff from 2018 once. In cp3's 9 games, the rockets played about half their games against terrible teams and bludgeoned them by like 20 ppg and then played about half against average teams and played +0 ball. So like +15 SRS against bottom feeders and 0 SRS against mid teams. Which averaged out to about +7 (probably the +5.8 you mention when you factor in the one game both harden and Paul missed). Meanwhile for harden, he didn't play quite as many bottom feeders but it was a much different story. +10 against the bottoms feeders but +5 against the average teams. For a +5 SRS against bad teams and the same +5 SRS against mid teams. Which kind of lines up with my impression of harden. Kind of opponent agnostic. Kind of like LeBron seemed to be. Never the guy who showed up against the wizards and went up 30 in the 3rd quarter and sat out the 4th. He and LeBron loved coasting against those teams and then trying hard in the 2nd half. But they also didn't seem to care if they were facing tough teams. They could just throw a team on their back and beat them by themselves. Just the same performance either way. Meanwhile Paul seems exactly like the kind of surgeon who carves up the wizards by 20 because they make too many mistakes he can take advantage of.

[Djoker]

Pop quiz: match the following 3 year RAPM peak to LeBron's teammates (Wade/Bosh, Kyrie/Love, Davis)

3rd
5th
11th
25th
26th

Spoiler:

Answer from best to worst: 10 Wade, 24 Irving, 10 Bosh, 18 Love, 20 Davis

Kyrie that high is definitely surprising. The man is 5th in 5-year RAPM from 2021-2025.

[iggymcfrack]

At this point it feels like there's a real wall of dissonance between some people who are strongly pro rapm 3-5 year samples as the ultimate baseline for player goodness and those who aren't in that camp. Personally, I'm sort of losing interest in debating any of it anymore so I think this will be my last thread until at least the 76-00 period begins. I'm just wondering what method(s) the most pro rapm voters are going to use in the other two periods of this project.

It'll be different won't it? And that will be interesting.

When we last did the Top 100 I used a very RPOY-style approach as my anchor, and in doing so I was particularly using an approach that allowed me to be as consistent as possible across eras. Basically, looking to give the nod to those who led the top teams.

For this 2001-25 project I thought: "Might as well really focus on the +/- data.", although I have to say, I didn't follow through with that to the extent I might have.

As we go back into the deeper past, the specific tools I'll use will have to change, but I'll still be trying to be as consistent as I can with the basketball thought process that lie beneath the specific stats I used. I can't be as consistent as I would be if I had the same stats available in all years, but I'll try my best and we'll see what happens.

When you think about it, I feel like doing the recent peaks first where we have the most data is actually very instructive. By really digging in on the modern era in great detail, I feel like it’s given me even more insight into which player types and profiles are kept valuable.

Someone like Anthony Davis I would have previously thought had to be massively impactful between his great box numbers and defense, but the fact that so many of his baskets are created by someone else really caps his value. That will be a good thing to remember working backwards to the era when we had less information to work with.

When you think about this, it’s really the best way we could have done this project. First we have the play-by-play era evaluating players with the full complement of metrics. Then we go back and do the BPM era where we don’t have impact stats but we had a full complement of box data. Then finally, once we finis that we can go back in time to the point we have the least data, using everything we’ve learned about value and player types. I feel like we’re really maxing out the project for learning.

[lessthanjake]

My Vote

1. 2005 Steve Nash

2. 2005 Manu Ginobili

3. 2018 James Harden

4. 2016 Draymond Green

The first three on this list were the first three for me in the last thread, so I’d just refer back to my explanation there (as well as all the discussion I’ve had about these players in these threads).

The last pick is very difficult. I waffled between various players, including Anthony Davis and Draymond Green. To me, Anthony Davis definitely had the better/greater playoff run, both because he played incredibly well, but also because it actually ended in a title. But I do think Draymond was generally more impactful, and that he had had a really good performance overall in his own playoff run.

Ultimately, I went with Draymond Green. A big thing with him is that I do genuinely think that his 2016 year was a step above any other year of his career. In that one particular year, he actually wasn’t a bad shooter. Indeed, he was a pretty good one! And I think that makes a massive difference in terms of his offensive value. I think this is why you see RAPM telling us Draymond was pretty neutral as an offensive player for the vast majority of his career but was genuinely a serious positive offensively in timeframes the more we focus down on 2016. This makes sense, because we’d expect a big who is a great passer *and* can space the floor to be a significant positive offensively, even without otherwise being a dynamic offensive player. Meanwhile, of course, he was a fantastic defender, and I think 2016 was during his peak defensive years. So it really all came together for Draymond in this one particular year IMO, such that I have 2016 Draymond as a tier or two above any other year of his career.

As for playoff performance specifically, he was really good! He was the primary driver of the Warriors’s easy first-round win against the Rockets. It wasn’t against a great team, but it was still an impressive performance from him. He was also fantastic against Portland with Steph missing three games. Draymond averaged 22/11/7 in that series with a 62% TS%, while being Draymond on defense. Again, not against a great team, but it was great stuff from Draymond. Draymond did actually struggle a good bit in the conference finals—in that series his shot did leave him, and he only averaged 11/9/4. It’s not all bad with Draymond when his shot leaves him, since he’s a great defender, but that was a weak series from him for purposes of this project, and it’s a significant reason the Warriors struggled in that series. But, importantly, he bounced back in the Finals, having another really good series. He was particularly great in Game 7. I do think we have to ding him for missing a game in the Finals though. So, overall, I think he was great in the playoffs, but there were some real lows, such that I think Anthony Davis was better in the 2020 playoffs. That said, Draymond’s 2016 playoffs was still impressive.

A big thing that Anthony Davis has over him is actually winning the title. So there’s a team achievement gap. But I think winning 73 regular season games is a massive achievement as well, and that has to factor in almost as significantly to the greatness of Draymond’s year as winning a title would.

So I think Anthony Davis was better in the playoffs and had more team achievement, though I don’t think the gaps there are insurmountable massive. But if Anthony Davis was better in the playoffs and achieved more, why do I have Draymond Green ahead of him? Well, it comes down to the multi-season RAPM stuff—where Draymond is just on another tier from Anthony Davis. If Draymond hadn’t played really well in the playoffs and/or the Warriors hadn’t set the regular-season wins record, I’d probably have voted for Anthony Davis. But, in this context, the large-sample impact stuff is enough to overcome what AD did in the playoffs.

I also considered Jimmy Butler here, but he suffers from a bit of an issue where it’s hard to pinpoint an exact year to use for him, because his best regular season and best playoff run weren’t the same year, and he also missed lots of regular season games in his better years. So, while I think he was a high-impact player who had great playoff runs and I think taking the teams he had to the Finals was really impressive, I just couldn’t really find a year that could get me there for him in this thread.

[Djoker]

15. 2020 Anthony Davis
16. 2007 Steve Nash
17. 2023 Joel Embiid
18. 2024 Luka Doncic

HM: 2018/2019 James Harden

I've gone at length about AD's case in prior threads quoted here.

Spoiler:

The last spot was also tough. It's between Davis and Dirk for me. The thing is... in the 2020 playoffs, AD was shooting so well that he was Dirk-like. 55% from long midrange and 38% from 3pt land is elite shooting. And compared with DIrk, he was also deadly near the basket and more importantly, an elite player on the other end of the court too. I do consider Dirk's intangibles to be better and his shooting gravity cannot be understated but I still feel like 2020 Davis is better than any version of Nowitzki as a basketball player. Certainly not a lot better but he's at least reasonably close offensively while being an elite defensive big man. That combo is tough to beat. He co-led the 2020 Lakers with Lebron so he flies a bit under the radar but he played like a first option, no doubt about it regardless of if you think Lebron was better than him that year. And I don't think his shooting is a total fluke. In the 2023 playoffs, he again shot 59% from long midrange and 33% from 3pt land over 16 games. Ben and Cody discussed on the podcast how AD somehow shoots orders of magnitude better in the PS and that it might be noise but this project is about 1-year peaks. I can't pretend like Davis shot poorly when he in fact shot the lights out. For his entire Lakers' run, the man shot 51% from midrange and 33% from 3pt in the playoffs over 47 games. What I also don't like about Dirk is that he's a 7 footer than isn't a rim protector. Basically you have to pair him up with a great defensive C and that is tough in terms of roster construction. Basically, as good as Dirk is, I do feel like his prototype of player is limiting.

I also considered AD over guys like KD but ultimately I don't think he has the same advantage on defense over those guys that he does on Dirk. And more importantly, those KD's defensive weaknesses are IMO easier to shore up in a team concept compared to DIrk's.

With that said, I don't love his impact stat portfolio being so pedestrian. It does give me some pause although I understand AD to be a huge playoff riser. His elite shooting combined with elite defense in the 2020 playoffs is the most value I can get out of any remaining player. Is it sustainable? I mean I think so. His length, size, and athleticism are elite so his defensive dominance is there. And if we look at another lengthy PS run in 2023, he also shoots the lights out so we can't dismiss his hot shooting as being on a heater. If a heater is 40 games long, at that point it's not a heater anymore. It's sustained great shooting.

Nash is an offensive dynamo and being unproven in terms of leading a team to a championship is probably the only reason he isn't in the offensive GOAT discussion for me. Just extremely impressive offensive signals from whichever way you slice it... ORtg ON Court, impact data, box score, eye test in terms of superb vision and just destabilizing the defense to an insane degree. His defense is a major weakness but he is a PG and as such he can be hidden easily. The Suns' problems defensively weren't due to Nash's deficiency but that they gave a lion's share of big minutes to the defensive sieve Amare Stoudemire.

Next is Embiid for me. His RS resume is spectacular to the point that he's definitely in top 10 discussions for RS only. Problem is his body frequently breaks down by the time the playoffs roll around. I chose 2023 as his peak despite the PS injury because judging by his play in surrounding relatively healthy years like 2021 and 2024, he could have continued his MVP form into the PS. He's the toughest player to place but around here feels right.

And the last spot comes down to Luka vs. Harden. Both heliocentric, methodical weapons of destructions. Luka is more resilient in the PS and I consider him a better player as a result.

[Cavsfansince84]

I'm going to add one or two more things which I don't like about rapm which is that the playoffs in many way can be a different thing than the rs. You have different levels of planning and intensity which go into series. So preferring a 4 or 5 year rapm rs sample over what a guy actually did in a particular year/playoff run is a faulty way of thinking imo. Guys who are consistently leading teams deep into the playoffs just might be better than guys who don't with similar team results in the rs or better rapm.

[Doctor MJ]

At this point it feels like there's a real wall of dissonance between some people who are strongly pro rapm 3-5 year samples as the ultimate baseline for player goodness and those who aren't in that camp. Personally, I'm sort of losing interest in debating any of it anymore so I think this will be my last thread until at least the 76-00 period begins. I'm just wondering what method(s) the most pro rapm voters are going to use in the other two periods of this project.

It'll be different won't it? And that will be interesting.

When we last did the Top 100 I used a very RPOY-style approach as my anchor, and in doing so I was particularly using an approach that allowed me to be as consistent as possible across eras. Basically, looking to give the nod to those who led the top teams.

For this 2001-25 project I thought: "Might as well really focus on the +/- data.", although I have to say, I didn't follow through with that to the extent I might have.

As we go back into the deeper past, the specific tools I'll use will have to change, but I'll still be trying to be as consistent as I can with the basketball thought process that lie beneath the specific stats I used. I can't be as consistent as I would be if I had the same stats available in all years, but I'll try my best and we'll see what happens.

When you think about it, I feel like doing the recent peaks first where we have the most data is actually very instructive. By really digging in on the modern era in great detail, I feel like it’s given me even more insight into which player types and profiles are kept valuable.

Someone like Anthony Davis I would have previously thought had to be massively impactful between his great box numbers and defense, but the fact that so many of his baskets are created by someone else really caps his value. That will be a good thing to remember working backwards to the era when we had less information to work with.

When you think about this, it’s really the best way we could have done this project. First we have the play-by-play era evaluating players with the full complement of metrics. Then we go back and do the BPM era where we don’t have impact stats but we had a full complement of box data. Then finally, once we finis that we can go back in time to the point we have the least data, using everything we’ve learned about value and player types. I feel like we’re really maxing out the project for learning.

Good thoughts iggy!

Back when we did the original Retro POY, I chose to do it in reverse chronological order because I felt the community would need this - both to interest them, and to allow them a more gradual way of trying to gain comfort analyzing the deep past. I think it worked basically like I hoped, and I think it makes a lot of sense to consider generally.

I should note that after that I was always advocating that if an RPOY was done again, it should be with forward chronological order (which was how it was done). Why? Because the nature of season-by-season play means you miss information by going backwards in time. During the original RPOY I frequently had the thought "Ooh, I wish I'd really realized this thing I just learned when we did the RPOY for the year after."

But for a quarter-century block like this project is trying to do, I'd say this is considerably less of a concern.

One note regarding Davis:

A thing I think that goes on with Davis is that his intensity and focus really seems to wax and wane over the course of a season with a variance greater than most other players. I don't want to excuse this as something that shouldn't be held against him, but it does mean that I don't think RAPM generally is going to be able to capture how devastating he is at his best.

This to say that while I don't know where I'll put AD in this project, I can't dismiss him lightly. The furthest I can go really is to say that I'm uncomfortable treating his shooting in the Bubble as if it represents "peak AD" for our purposes. In one sense, it was clearly his peak, but it's also a peak in a context that most other players in history never got a chance to play in.

I reject the idea of asterisking the Lakers' Bubble championship or AD's role in that ring, but when comparing AD to players in this project, saying anything like "and AD shot the 3 too, so no biggie there" just doesn't seem right. The fact that we just never saw AD look like a 3-point shooter to fear in any normally attended season seems significant.