RealGM

~Which stat produces the best GOAT career list?~

Most stats are not designed with GOAT career lists in mind. They’re often designed to quantify goodness or value in a current or future time range, using various inputs (e.g. box stats, plus minus data, WOWY data). They are often designed with an eye towards a single season or several seasons or at most a prime, not a career.

Still, it’s interesting to consider how the various stats compare to our usual GOAT list. I’ve collected a few of the more accessible stat GOAT lists. There’s a few different time ranges (e.g. shot clock era vs NBA merger vs plus minus era), and occasionally when a stat publishes their own GOAT list, they’ll have different weightings (e.g. the RAPTOR GOAT list includes extra prime weighting), so I’ve included those here.

A few discussion questions:
-Which stats produce the closest GOAT list to yours, or to the usual GOAT list? Why does that stat gets closest?
-What weightings are optimal? (e.g. should playoffs get extra weighting, should peaks/primes get extra weighting, and if so how much?)
-Are there any players or trends/archetypes where the stats diverge significantly from your GOAT list?
-When you disagree, what are you valuing that the stat doesn’t measure, or vice versa?

1. Shot Clock Era / Full NBA History (1949 NBA / 1955 shot clock era — present)

A. Win Shares: RS + PS Career Total (https://www.basketball-reference.com/leaders/ws_career_c.html) (full NBA history)

Bi. Backpicks BPM: RS + PS Career Total (https://thinkingbasketball.net) (1955–2024)

Bii. Backpicks BPM: RS + 7x PS Career Total (PS time given 7x weighting) (https://thinkingbasketball.net) (1955–2024)

Biii. Backpicks BPM: CORP (PS time given 7x weighting, championship odds follow a superlinear function so peaks are weighted over primes over non-prime years) (1955–2018)

C. Career WOWYR (‘adjusted WOWY’) (1955-2016) (https://thinkingbasketball.net/metrics/wowyr/)

2. NBA Merger Era (1974 defensive stat era / 1977 merger — present)

D. Basketball Reference BPM: RS + PS VORP (https://www.basketball-reference.com/leaders/vorp_career_c.html) (1974–2024)

E. Fiverthirtyeight’s RAPTOR: RS + PS JAWS (average between career total and 7 year prime) (https://fivethirtyeight.com/features/lebron-or-mj-raptor-picks-the-best-nba-players-of-the-past-40-years/) (1977–2023)

F. PIPM: RS + PS Career Total (in units of ‘Wins added’) (https://docs.google.com/spreadsheets/d/1EIZvj_3-9SZULWomHz54V1CPL092j70u_0vUhoEEaIk/edit?usp=share_link) (1977–first few games of 2021)

C. Plus Minus Era (1994 plus minus era / 1997 play by play era — present)

Gi. Backpicks Augmented Plus Minus: RS + PS CORP (PS games given 7x weighting like Backpicks BPM) (https://thinkingbasketball.net/2021/07/10/playoff-plus-minus-part-iii-changes-in-the-postseason/) (1994–2024 RS, 1997–2024 PS)

Gii. Backpicks Augmented Plus Minus: RS + 7x PS Career Total (PS games given 7x weighting) (1994–2024 RS, 1997–2024 PS)

H. Engelmann RAPM: RS + PS Career Value (https://docs.google.com/spreadsheets/d/1bg8KxzagN7D0O16EmUO9_kCyXwthEUjKywlrWPQUQt8/edit#gid=0) (1997–2023.5):

I. DARKO Daily Plus Minus: RS + PS Career Value (https://apanalytics.shinyapps.io/DARKO//) (1998–2024)

Within each time range:

1. The Backpicks BPM lists in general look better, would probably pick Biii. CORP if I had to choose one
2. PIPM
3. Seems like the most similarity between lists here. Tough pick, I think I lean towards DARKO DPM?

Common theme throughout these lists is that the Utah guys are higher than where I put them. Not really surprising given the volume dimension. Dirk similarly looks really good here.

Of the ones listed, I would take PIPM.

Despite its reputation, I don't think I would take any of those lists over "Playoff PER Age 21-35", which did a very good job mimicking the first 1/4 or so of the Top 100 (except of course for "no video, no impact stats so don't blame the box score either" Bill Russell). Very few "weird" results (for whatever that means), fairly good alignment with the project even without longevity being factored in (Kawhi being the notable non-longevity inclusion). Age 22-31 Box Score (3*Playoffs + 1*Regular Season) * Prime Longevity + Constant*Championships was also very good at predicting the first 1/3 of the project.

1. MJ
2. Lebron
3. Shaq
4. Hakeem
5. Kareem
6. Duncan
7. Kawhi
8. Barkley
9. Dirk
10. Durant
11. Robinson
12. CP3
13. Wilt (extended age range sunk him)
14. Magic
15. Jerry
16. Steph
17. Kobe
18. Pettit
19. Wade
20. Karl Malone
21. Harden

Even the following guys were pretty close to their rankings in the Top 100 project:
31. Reggie
35. Ewing
36. Stockton
38. Frazier

jalengreen wrote:Within each time range:

1. The Backpicks BPM lists in general look better, would probably pick Biii. CORP if I had to choose one
2. PIPM
3. Seems like the most similarity between lists here. Tough pick, I think I lean towards DARKO DPM?

Common theme throughout these lists is that the Utah guys are higher than where I put them. Not really surprising given the volume dimension. Dirk similarly looks really good here.

Good thoughts jalengreen!

1. Agreed on Backpicks BPM!
-I'm particularly impressed with it being so high (IMO appropriately) on both Magic and Russell. No other box stat does that, and that seems to jibe with its claims to try to measure creation and defense more accurately.
-In a similar trend, it also ranks Nash and Hakeem higher than other box stats.
-It seems to rank 'the Utah guys' (lol) lower than the other stats from its era, although not low enough for my taste. It ranks Malone 6-11 depending on the weighting, while WS ranks Malone 4th; it ranks Stockton 8-14 while WOWYR ranks Stockton 2nd.

-A few players shift around as you change weightings in ways that make it hard to pick a preferred version. Bird goes from 8-9 in the linear versions (RS+PS and RS+7xPS) to 14th in the CORP version; I have Bird in my top ~10 tier so I prefer Bird's linear ranking. Chris Paul and Harden are ranked 11-12 in linear career value (RS+PS), but drop in the linear playoff-weighted one, so I prefer their playoff-weighted ranking. Pippen is ranked 9th in the playoff-weighted one, which is too high IMO, so I prefer his non-playoff weighted ranking. Garnett is only ranked 19–21 in the linear version and 11th in the CORP version, so I prefer his CORP ranking.

So... does this mean my personal ranking involves inconsistent playoff/peak weightings for different players, and is that problematic if so? It could suggest I subconsciously have inconsistent playoff/peak weightings when evaluating different players, but I'm not sure this suggests a logical consistency that undermines my list's validity. After all, these stats often measure something like value (in a given role) or production (in a given role), while people's GOAT lists often care more about goodness (in general, not tied to a specific role). When using some stat like BPM as an estimate for goodness (value is likely generally proportional to goodness, even if a favorable / unfavorable role could influence the signal), the signal for goodness might require different sample weightings for different players. For example, peoples' concerns for different players are different. For Paul and Harden, people have more playoff resiliency concerns, so taking a more playoff-weighted ranking will naturally get closer to people's intuitive ranking; for Pippen, people have more concerns about the height of his peak, so taking a peak-weighted ranking will naturally get closer to people's intuitions.

There's also some inherent uncertainty range in a stat that's not reflected in a flat ranking. If 3 different sample weightings have a player generally within a certain range (e.g. Paul varies from 11th to 17th in these BPM rankings), and if it's not clear what the right weighting is a priori, you could use the changes in ranking with reasonable weightings as a way to get a sense for the uncertainty range of the stat, then use those uncertainty ranges to inform your personal ranking.

-Between the different weightings in Timespan 1, I think there's pros to both BPM (RS + 7xPS) and BPM (CORP). Like CORP, I have Jordan in my top 2 (not 4th), Kareem > Wilt, Garnett in my top 15. But like the linear playoff weighting (RS + 7xPS), I have Bird higher than 14, Kobe higher than 16, Shaq > Hakeem.

Re: WOWYR, the noise of that stat definitely limits how well it follows our prior. Stockton 2nd all time, Paul Pierce 6th all time, Russell at 18th all time, Kareem at 21 all time, Bird at 24, Erving at 83rd, Wade at 98th all time... those differ too much to beat out BPM. Still, even if it's noisy, I do wonder if there's some signal here. Consider Kobe vs Wade, who are often compared. Career Total WOWYR is a pure-impact metric that has Kobe 11th vs Wade much lower, and impact metrics that favor Kobe are hard to come by. Even if we attribute much of Wade's ranking to noise, could there still be some signal to help statistically justify how often we rank Kobe > Wade as a community? It's also a neat coincidence, given the noise levels, that the two most common GOAT candidates (Jordan/LeBron) are both in the top 4 in career WOWYR.

Era 2: Merger Era.
I think I favor PIPM in this era like you, although there are a few player rankings from RAPTOR that match my personal ranking more closely (Jordan, Bird, Magic, Wade higher). I wonder how much work is being done by the 'JAWS' approach (extra weighting for prime years by ranking players based on the average of their career value and their 7-year prime value). If we took a PIPM 'CORP' or a PIPM 'JAWS', would this do better?
--> I think one lesson from this analysis is that I don't weigh things linearly. I do give extra weighting to playoffs (unclear exactly how much ) and to peaks/primes (unclear exactly how much).

Era 3: Plus Minus Era.
I'm less sure between DPM and AuPM (especially the RS + 7xPS AuPM). Compared to DPM, AuPM is probably too high on Dirk (3 vs 4), too low on Garnett (8 vs 3), too high on Draymond (11th although DPM's 27th ranking is probably too low). Still, I like AuPM's ranking of Shaq/Jordan more (although it does have access to some pre-98 data for them which helps a ton). Having Curry and Kobe > Chris Paul > Harden is better (DPM says Chris Paul > Curry > Harden > Kobe), and having Kidd/Paul Pierce out of the top 12 is also better. Like you say, tough picks!

f4p wrote:Of the ones listed, I would take PIPM.

Despite its reputation, I don't think I would take any of those lists over "Playoff PER Age 21-35", which did a very good job mimicking the first 1/4 or so of the Top 100 (except of course for "no video, no impact stats so don't blame the box score either" Bill Russell). Very few "weird" results (for whatever that means), fairly good alignment with the project even without longevity being factored in (Kawhi being the notable non-longevity inclusion). Age 22-31 Box Score (3*Playoffs + 1*Regular Season) * Prime Longevity + Constant*Championships was also very good at predicting the first 1/3 of the project.

1. MJ
2. Lebron
3. Shaq
4. Hakeem
5. Kareem
6. Duncan
7. Kawhi
8. Barkley
9. Dirk
10. Durant
11. Robinson
12. CP3
13. Wilt (extended age range sunk him)
14. Magic
15. Jerry
16. Steph
17. Kobe
18. Pettit
19. Wade
20. Karl Malone
21. Harden

Even the following guys were pretty close to their rankings in the Top 100 project:
31. Reggie
35. Ewing
36. Stockton
38. Frazier

Thanks for the extra stat, f4p! (To all: if there's any stat you have access to that I missed, feel free to add them like f4p did -- the more, the merrier). Quick clarification: For PER ages 21-35, was this calculated manually by you, or is there a list you could link somewhere?

Good point about your analysis comparing PER to the top 100 list! I remember seeing the posts as the time and thinking it was a neat idea. I wonder if we could apply a similar methodology to the other GOAT lists here to get some sort of quantification for how similar these stats' GOAT lists are to people's opinion. We'd need appropriate time filters... e.g. there's no RAPM data for Russell, so no use 'punishing' Engelmann RAPM as a stat for not having Russell 4th all time. But if done right, it could be interesting to check!

What, in your mind, is the lesson we should take away from a stat matching our prior better than another? If one stat matches our prior GOAT list better than another, is that a sign the first stat is doing something right or something better? Or is this argument too circular? Any other lessons, or should this just be seen as a neat coincidence for the stat?

Re: PER specifically, I'm interested why you see it as producing a closer GOAT list to yours than say BPM (maybe ~7xPS weighting or CORP weighting) or PIPM or AuPM (maybe ~7xPS weighting). Are there any rankings in particular you see it doing 'better' ?

Me personally, just comparing box stat to box stat, I definitely think Backpicks BPM's ranking (with some weighting tbd) gets closer to my personal GOAT list. For example:
-Hakeem and Shaq are great, but to me, Kareem probably had the better career (more dabatable if we go to peaks tho).
-The omission of Russell out of the top 21 seems like a pretty glaring omission. PER is mostly an offensive-only stat, so it makes sense. But BPM having Russell in the top 6 regardless of the weighting seems like BPM's doing something right.
-Likewise for BPM having Magic in the top 10 (PER 14th) and Bird in the top 10/15 (Bird out of the top 21)
-Among big men, BPM having Wilt in the top 10 and garnett in the top ~20 also seems better.
-As for the 'weird results', having Kawhi and Barkley in the top 8 careers doesn't seem that much less weird than having Pippen at 9, Malone at 11, Stockton at 14. (Pippen/Malone/Stockton move around slightly with different weightings, but you get the idea).

Don't get me wrong, PER's definitely a list that matches a lot of my priors. PER's higher on Oscar and West than Backpicks BPM -- I definitely have them in my top 20. It has Jordan/LBJ top 2 unlike Backpicks BPM, and it doesn't have Jordan < Wilt like RS+PS BPM does (although BPM CORP puts Jordan over Wilt). Perhaps your GOAT rankings are slightly different from mine (e.g. you might be lower on Russell/Magic/Bird/Wilt/Garnett), or perhaps your uncertainty ranges are slightly different from mine (even if we both have Jordan top 2, maybe you see having Jordan 4th like RS+PS BPM as too much beyond your uncertainty range to be sensible, moreso than PER having Kawhi in the top 8). Different rankings are to be expected! I'm just curious which players you think have a better ranking in PER vs BPM (or your other preferred stat from the OP like PIPM or AuPM)

Biii. Backpicks BPM: CORP (PS time given 7x weighting, championship odds follow a superlinear function so peaks are weighted over primes over non-prime years).

This kind of makes sense. I have generally leaned heavily PS>RS in terms of overall career value. I've also commented on how basic career value measures are linear in nature and don't properly account for the championship odds lift given by spectacular peaks compared to just good ones.

Lists that attempt to go for career value, might underestimate Hakeem or even CP3 in very rare cases (even though maybe the latter's box-score can perhaps overrate him a tad), because they had a lot of shorter PS runs, so they didn't always accumulate as many games played in the PS as others. This perhaps lead to a discrepancy with my personal ranking, just because I tend to extrapolate from their moments of brilliance and assume they could perform similarly if given a longer leash in a different setting to really build up some PS value in their prime.

LukaTheGOAT wrote:Biii. Backpicks BPM: CORP (PS time given 7x weighting, championship odds follow a superlinear function so peaks are weighted over primes over non-prime years).

This kind of makes sense. I have generally leaned heavily PS>RS in terms of overall career value. I've also commented on how basic career value measures are linear in nature and don't properly account for the championship odds lift given by spectacular peaks compared to just good ones.

Lists that attempt to go for career value, might underestimate Hakeem or even CP3 in very rare cases (even though maybe the latter's box-score can perhaps overrate him a tad), because they had a lot of shorter PS runs, so they didn't always accumulate as many games played in the PS as others. This perhaps lead to a discrepancy with my personal ranking, just because I tend to extrapolate from their moments of brilliance and assume they could perform similarly if given a longer leash in a different setting to really build up some PS value in their prime.

Good point about postseason volume LukaTheGOAT! When doing career value, I sometimes struggle in figuring out the right way to incorporate postseason volume. We care about the postseason, and like you say, often want to weigh a postseason game more than a regular season game, so it doesn't feel right to just compare someone's regular season career value.

On the other hand, someone's postseason career volume depends a lot on whether their team is good enough to stay in the playoffs for numerous long runs. A star player can obviously influence that, but it's also highly dependent on teammates. If Player A and Player B are very similar in regular season impact metrics and perceived goodness, but Player A's in a favorable role on a stronger team making deep playoff runs while Player B's not, then they might end up looking similar in RS career value but Player A might clearly exceed Player B in RS+PS career value. Is it fair to blame Player B for not making more deeper playoff runs?

One factor that might help make this more fair is milage/fatigue. Multiple deep playoff runs shortens your offseason recovery period, and does add mileage. So after multiple deep runs, Player A might not be quite as healthy or have quite as much motor in the following regular season. Player A's body might even age more quickly due to the accumulated mileage. If this is the case, Player A looks better in PS career value, but a little worse than they would have in RS career value, so the two players might (correctly) look similar in RS+PS career value. But is this mileage effect large enough to counteract all the extra opportunities for PS career value Player A got compared to Player B? It probably varies case by case, but I'm not so sure.

In my mind, Garnett may be the best example of this trend. At least more so than CP3, whose body seemed to struggle to stay healthy in deeper postseason runs. You can actually see KG's ranking drop in all the stats that give playoffs extra weighting, which is consistent with this idea. Opponents might say this is because he lacks resilience. In my mind, the bigger thing holding him back (from having a better RS+PS career value in these stats) is how often he lacked the teammates to make deeper playoff runs. He missed the playoffs in his first year, then only ever made the first round in his next 7 straight years. Then, right in the heart of his prime, he missed 3 playoffs from 2005-2007 and got injured and missed yet another playoff in 2009. We might blame him for getting injured in 09, but the rest seem at least partially on his teammates/situation. It's possible he might have had slightly less of a motor for the regular season if he was making deep playoff runs throughout these years. But I think the added playoff time would have been the greater effect, giving him more total career value from a RS+PS or RS+7xPS perspective if he had been in a better situation.

This actually might be another benefit of using 'CORP' as the unit of volume. It does consider volume gained from say total possessions per game / minutes per game played, it incorporates volume lost from say missed games, but since it's trying to estimate your championship odds (in a way that doesn't depend on your specific situation/teammates, just on how valuable you are), it doesn't penalize players for being on teams that don't make deep playoff runs!

DraymondGold wrote:

LukaTheGOAT wrote:Biii. Backpicks BPM: CORP (PS time given 7x weighting, championship odds follow a superlinear function so peaks are weighted over primes over non-prime years).

This kind of makes sense. I have generally leaned heavily PS>RS in terms of overall career value. I've also commented on how basic career value measures are linear in nature and don't properly account for the championship odds lift given by spectacular peaks compared to just good ones.

Lists that attempt to go for career value, might underestimate Hakeem or even CP3 in very rare cases (even though maybe the latter's box-score can perhaps overrate him a tad), because they had a lot of shorter PS runs, so they didn't always accumulate as many games played in the PS as others. This perhaps lead to a discrepancy with my personal ranking, just because I tend to extrapolate from their moments of brilliance and assume they could perform similarly if given a longer leash in a different setting to really build up some PS value in their prime.

Good point about postseason volume LukaTheGOAT! When doing career value, I sometimes struggle in figuring out the right way to incorporate postseason volume. We care about the postseason, and like you say, often want to weigh a postseason game more than a regular season game, so it doesn't feel right to just compare someone's regular season career value.

On the other hand, someone's postseason career volume depends a lot on whether their team is good enough to stay in the playoffs for numerous long runs. A star player can obviously influence that, but it's also highly dependent on teammates. If Player A and Player B are very similar in regular season impact metrics and perceived goodness, but Player A's in a favorable role on a stronger team making deep playoff runs while Player B's not, then they might end up looking similar in RS career value but Player A might clearly exceed Player B in RS+PS career value. Is it fair to blame Player B for not making more deeper playoff runs?

One factor that might help make this more fair is milage/fatigue. Multiple deep playoff runs shortens your offseason recovery period, and does add mileage. So after multiple deep runs, Player A might not be quite as healthy or have quite as much motor in the following regular season. Player A's body might even age more quickly due to the accumulated mileage. If this is the case, Player A looks better in PS career value, but a little worse than they would have in RS career value, so the two players might (correctly) look similar in RS+PS career value. But is this mileage effect large enough to counteract all the extra opportunities for PS career value Player A got compared to Player B? It probably varies case by case, but I'm not so sure.

In my mind, Garnett may be the best example of this trend. At least more so than CP3, whose body seemed to struggle to stay healthy in deeper postseason runs. You can actually see KG's ranking drop in all the stats that give playoffs extra weighting, which is consistent with this idea. Opponents might say this is because he lacks resilience. In my mind, the bigger thing holding him back (from having a better RS+PS career value in these stats) is how often he lacked the teammates to make deeper playoff runs. He missed the playoffs in his first year, then only ever made the first round in his next 7 straight years. Then, right in the heart of his prime, he missed 3 playoffs from 2005-2007 and got injured and missed yet another playoff in 2009. We might blame him for getting injured in 09, but the rest seem at least partially on his teammates/situation. It's possible he might have had slightly less of a motor for the regular season if he was making deep playoff runs throughout these years. But I think the added playoff time would have been the greater effect, giving him more total career value from a RS+PS or RS+7xPS perspective if he had been in a better situation.

This actually might be another benefit of using 'CORP' as the unit of volume. It does consider volume gained from say total possessions per game / minutes per game played, it incorporates volume lost from say missed games, but since it's trying to estimate your championship odds (in a way that doesn't depend on your specific situation/teammates, just on how valuable you are), it doesn't penalize players for being on teams that don't make deep playoff runs!

On the mileage making it fair ... (and not saying you're saying it does, seems more something you're throwing out as a possible partial mitigation than saying it does ... you seem to go the other way and say it doesn't really)

... you've offered, and people are going for, 7x playoff measures.

Ignoring era format differences ...

... say I'm comparing ... I'll pluck similar era guys ... Elton Brand and Derek Fisher. Similar-ish RS minutes Brand goes around 2000 minutes longer, Fisher has circa 7000 playoff minutes (a touch less than) to Brand's circa 1000 (a touch more than). Say by Ref-BPM Brand accumulates a huge VORP lead but even though Brand's good in the playoffs, 7x multiplier on Fisher's extra minutes makes it look close (48.5 to 40.4). Maybe one thinks Fisher is much better than his boxscore and it's (accidentally) right. That sort of weighting just feels as though it's playoff presence rather than player goodness. In short I'm pretty confident playoff mileage isn't 7x mileage.

Agree with KG, disagree with any implication of somehow "not Paul" (perhaps you're saying "less Paul" ... and I get Garnett's teammates were really bad for a great player) when the KG case is bad teammates and ... through 2017 Paul has two years where his team aren't below 0 when he's off the floor (I think RAPM stuff suggests this isn't a coincidence). There is something to the Paul health case but then one could argue how do we know how KG would fare. And Paul's injuries mostly aren't the sort that meant he would have been completely out, more that he missed time and the team typically didn't hold up well enough in his absence, I think (could be wrong).

I don't have his scrapes memorized but I've seen it written that injuries "knocked [Barkley] out of the playoffs" in 1994 and 1995 runs [Barry and Cohn, 1995] (he was on court but said to be less effective, taking shots to play, looking at retirement ... not necessarily viable for further rounds) plus out in 1998. If (a big if) one concurs that's injuries taking Barkley out of two of his 4 or 5 ten-or-more game runs as "the guy". And Barkley was expressly criticized for not taking conditioning/rehab seriously enough. I just think Paul takes a lot of heat on this ... IDK playoffs is hard. For instance, I'd probably rather have a great player miss round 1 than the finals if missing exactly one series because if I have any chance at all (otherwise it doesn't matter) I'm likely to be a higher seed and have a chance at beating a lower seed without the star, but less so the finalist ... but for these methods I should definitely prefer the out in finals route ... even if I don't happen to get there I'm much more likely to get 2 or 3 rounds. On this line of thinking some seem to nuke '92 Robinson but off reporting at the time of the injury he might have been back circa mid-2nd round ... as I say it's tricky and it depends what you're (trying to be) measuring. For me, I do prefer the idea of something more holistic and less circumstantial.

Sorry, bit rambly, stream-of-consciousness thoughts partially in response to the mileage idea, Paul and some areas you touched on but as I say a bit tangenty.


Fwiw, baseline will make a difference too. WS, for instance, as a cumulative stat has a very low baseline and will thus favor longevity monsters. So in this case "WS above average" (or a "replacement level" not a 0 win level) might make a more viable list.

Owly wrote:

DraymondGold wrote:

LukaTheGOAT wrote:Biii. Backpicks BPM: CORP (PS time given 7x weighting, championship odds follow a superlinear function so peaks are weighted over primes over non-prime years).

This kind of makes sense. I have generally leaned heavily PS>RS in terms of overall career value. I've also commented on how basic career value measures are linear in nature and don't properly account for the championship odds lift given by spectacular peaks compared to just good ones.

Lists that attempt to go for career value, might underestimate Hakeem or even CP3 in very rare cases (even though maybe the latter's box-score can perhaps overrate him a tad), because they had a lot of shorter PS runs, so they didn't always accumulate as many games played in the PS as others. This perhaps lead to a discrepancy with my personal ranking, just because I tend to extrapolate from their moments of brilliance and assume they could perform similarly if given a longer leash in a different setting to really build up some PS value in their prime.

Good point about postseason volume LukaTheGOAT! When doing career value, I sometimes struggle in figuring out the right way to incorporate postseason volume. We care about the postseason, and like you say, often want to weigh a postseason game more than a regular season game, so it doesn't feel right to just compare someone's regular season career value.

On the other hand, someone's postseason career volume depends a lot on whether their team is good enough to stay in the playoffs for numerous long runs. A star player can obviously influence that, but it's also highly dependent on teammates. If Player A and Player B are very similar in regular season impact metrics and perceived goodness, but Player A's in a favorable role on a stronger team making deep playoff runs while Player B's not, then they might end up looking similar in RS career value but Player A might clearly exceed Player B in RS+PS career value. Is it fair to blame Player B for not making more deeper playoff runs?

One factor that might help make this more fair is milage/fatigue. Multiple deep playoff runs shortens your offseason recovery period, and does add mileage. So after multiple deep runs, Player A might not be quite as healthy or have quite as much motor in the following regular season. Player A's body might even age more quickly due to the accumulated mileage. If this is the case, Player A looks better in PS career value, but a little worse than they would have in RS career value, so the two players might (correctly) look similar in RS+PS career value. But is this mileage effect large enough to counteract all the extra opportunities for PS career value Player A got compared to Player B? It probably varies case by case, but I'm not so sure.

In my mind, Garnett may be the best example of this trend. At least more so than CP3, whose body seemed to struggle to stay healthy in deeper postseason runs. You can actually see KG's ranking drop in all the stats that give playoffs extra weighting, which is consistent with this idea. Opponents might say this is because he lacks resilience. In my mind, the bigger thing holding him back (from having a better RS+PS career value in these stats) is how often he lacked the teammates to make deeper playoff runs. He missed the playoffs in his first year, then only ever made the first round in his next 7 straight years. Then, right in the heart of his prime, he missed 3 playoffs from 2005-2007 and got injured and missed yet another playoff in 2009. We might blame him for getting injured in 09, but the rest seem at least partially on his teammates/situation. It's possible he might have had slightly less of a motor for the regular season if he was making deep playoff runs throughout these years. But I think the added playoff time would have been the greater effect, giving him more total career value from a RS+PS or RS+7xPS perspective if he had been in a better situation.

This actually might be another benefit of using 'CORP' as the unit of volume. It does consider volume gained from say total possessions per game / minutes per game played, it incorporates volume lost from say missed games, but since it's trying to estimate your championship odds (in a way that doesn't depend on your specific situation/teammates, just on how valuable you are), it doesn't penalize players for being on teams that don't make deep playoff runs!

On the mileage making it fair ... (and not saying you're saying it does, seems more something you're throwing out as a possible partial mitigation than saying it does ... you seem to go the other way and say it doesn't really)

Yep!

... you've offered, and people are going for, 7x playoff measures.

Ignoring era format differences ...

The era differences is an interesting aside. There's such major differences in years played, games played, minutes played, and pace as you compare different eras. If you're doing some sort of cumulative RS+PS career value, what's the fairest way to do it across eras?

Take the 60s players. They had a lot of Pros and Cons that helped and hurt their RS+PS raw career value, just from the era they played in.
Con: They played fewer years. The expectation was to play through college rather than joining the NBA early. The expectation was to retire younger, and not play until you're 40.
Con: They played fewer games per season. In the regular season, they had seasons that were slightly shorter than 82 games. But more importantly, they had much shorter playoffs, at only ~2 rounds long (lower seeds played an extra early round best of 5, but those lower seeds were rarely good enough to make the finals and those first rounds were shorter anyway).
Pro: They played far more minutes within a given game.
Pro: the pace was higher, so the possessions really added up.
And of course there's all sorts of deeper contextual factors, like the quality of the average player in the league, team fit, how well the rules/strategies maximize a player's talents, etc., that all could bias some player's measured career value in some stat in ways that are hard to quantify.

Compare that to today, where players play (1) Pro: more years (e.g. more time in the NBA than college, later retirements), Pro/Con: longer playoff runs (but more load managing in the regular season), Con: fewer minutes per game, Con: slower pace than the 60s (but Pro: higher pace than many other eras).

So which era is more favorable to accumulating some sort of RS+PS career value in some stat? I think this era, but I'm not sure how to mentally curve how much. If the oldest eras were less favorable to gaining RS+PS career value longevity, how much do I boost the old guys, if at all? It's also not clear to me if it's a linear trend, with longevity becoming progressively easier with each passing decade.

... say I'm comparing ... I'll pluck similar era guys ... Elton Brand and Derek Fisher. Similar-ish RS minutes Brand goes around 2000 minutes longer, Fisher has circa 7000 playoff minutes (a touch less than) to Brand's circa 1000 (a touch more than). Say by Ref-BPM Brand accumulates a huge VORP lead but even though Brand's good in the playoffs, 7x multiplier on Fisher's extra minutes makes it look close (48.5 to 40.4). Maybe one thinks Fisher is much better than his boxscore and it's (accidentally) right. That sort of weighting just feels as though it's playoff presence rather than player goodness. In short I'm pretty confident playoff mileage isn't 7x mileage.

Yep, taking 7x Playoff weighting definitely makes it measure more toward a players' playoff presence like you say than player goodness, and this does punish the occasional players whose playoff presence was unfairly limited by poor teammates/situation.

On the whole though, at least as far as people have discussed so far, the results qualitatively seem to improve / get closer to our prior. So what do we do? Are we just at the limits of how well cumulative statistics can measure career value? They're usually intended for comparing single seasons or several season stints after all.

Like I said before, this is one advantage of calculating a CORP from a statistic. It allows increased playoff weighting when looking at how good a player was e.g. per game, and you can penalize a player for missing games, but at the end you convert how good they were e.g. per game to some CORP %, which doesn't care about how deep the playoff run actually was.

The cons there are it's a bit more complex to calculate (and for many stats, we'd have to calculate this ourselves). And it introduces some uncertainty in exactly how much to weigh peak > prime > career when calculating CORP, even if adding that weighting would provide more accurate career value / ranking (see e.g. https://thinkingbasketball.net/2018/04/13/goat-meta-thoughts-and-longevity/ for discussion on the topic).

Agree with KG, disagree with any implication of somehow "not Paul" (perhaps you're saying "less Paul" ... and I get Garnett's teammates were really bad for a great player) when the KG case is bad teammates and ... through 2017 Paul has two years where his team aren't below 0 when he's off the floor (I think RAPM stuff suggests this isn't a coincidence). There is something to the Paul health case but then one could argue how do we know how KG would fare. And Paul's injuries mostly aren't the sort that meant he would have been completely out, more that he missed time and the team typically didn't hold up well enough in his absence, I think (could be wrong).

I don't have his scrapes memorized but I've seen it written that injuries "knocked [Barkley] out of the playoffs" in 1994 and 1995 runs [Barry and Cohn, 1995] (he was on court but said to be less effective, taking shots to play, looking at retirement ... not necessarily viable for further rounds) plus out in 1998. If (a big if) one concurs that's injuries taking Barkley out of two of his 4 or 5 ten-or-more game runs as "the guy". And Barkley was expressly criticized for not taking conditioning/rehab seriously enough. I just think Paul takes a lot of heat on this ... IDK playoffs is hard. For instance, I'd probably rather have a great player miss round 1 than the finals if missing exactly one series because if I have any chance at all (otherwise it doesn't matter) I'm likely to be a higher seed and have a chance at beating a lower seed without the star, but less so the finalist ... but for these methods I should definitely prefer the out in finals route ... even if I don't happen to get there I'm much more likely to get 2 or 3 rounds. On this line of thinking some seem to nuke '92 Robinson but off reporting at the time of the injury he might have been back circa mid-2nd round ... as I say it's tricky and it depends what you're (trying to be) measuring. For me, I do prefer the idea of something more holistic and less circumstantial.

Yep, I mean to say "more Garnett, less Paul" than "only Garnett, not Paul". Re: injuries, there's some probabilistic-ness to who/when players get injured. Like you say, when a player doesn't have the opportunity to do deeper playoff runs, they don't have the opportunity to get injured, so we'll never know. At the extremes though, among the most durable and most injury-prone players, people do usually seem comfortable projecting some confidence in what their injury chances would be if they had deeper playoff runs.

During those first ~12 years when Garnett was missing so much potential playoff time, he seemed quite durable. Compare that to Paul, who is one of the more injury-prone players among the top 20/25/whatever, getting injured in quite a few playoff runs and missing regular season time too. So just in terms of confidence, I have more confidence that Garnett would be healthy for longer playoff runs if he had the teammates to facilitate that, compared to Paul. Like you say, Paul did miss playoff time, but he was sometimes able to come back after injury and still play (albeit somewhat limited at times). Obviously everyone would benefit -- your RS+PS value would only go up if you had more time in the PS and the same time in the RS.

I just mean I have greater confidence in Garnett to benefit more in this context (i.e. look better in some RS+7xPS metric) from playing X more playoff series due to better teammates helping him get further, since his health seemed more sustainable compared to Paul. It's a thought experiment, so we'll never know for sure of course!

Sorry, bit rambly, stream-of-consciousness thoughts partially in response to the mileage idea, Paul and some areas you touched on but as I say a bit tangenty.


Fwiw, baseline will make a difference too. WS, for instance, as a cumulative stat has a very low baseline and will thus favor longevity monsters. So in this case "WS above average" (or a "replacement level" not a 0 win level) might make a more viable list.

[/quote][/quote] All good!

And interesting point about baseline! I hadn't thought of that before.

A lot of the stats in the OP are accumulated totals so for me who doesn't have (situational) longevity as a criteria, none of those stats are frankly very close to my career GOAT list. I also don't think any of these stats even ones like RAPM or PIPM are very good at making a GOAT list on their own considering we see the likes of Karl Malone and Chris Paul near the top of almost every list. Virtually no one has those two ranked in the top 15 all time and some have them much lower. Also a huge issue with using stats that aren't available for most of NBA history.

My career GOAT list is mainly driven by three factors: peak level of play, durability (as opposed to longevity) and accolades.

Aleco wrote:Wow, Lebron dominates every list except wowyr. Simply incredible, consensus goat without a doubt.

Hey Aleco -- I definitely agree LeBron has one of the best GOAT career arguments. I will say though, I think a lot of this unanimity among the stats comes from the methodology (/underlying criteria of the methodology). Different methodologies might have a slightly more diverse set of people at #1.

Almost all of these lists here are just (linear) sums of player volume in a stat across their whole career, either with equal or extra playoff weighting. In other words, you take all their value in a stat in the RS and PS and add it together (perhaps with extra PS weighting).

This focus on Career Value (adding up value over time) linearly (so no extra weighting for peak/primes, see e.g. https://thinkingbasketball.net/2018/04/13/goat-meta-thoughts-and-longevity/) puts maximum emphasis on raw longevity.

Let's take minutes as a proxy for career time-volume/longevity (in truth we should use possessions, but minutes are more accessible and they are roughly proportional https://www.statmuse.com/nba/ask/player-in-nba-history-with-the-most-minutes-in-the-regular-season-and-playoffs).
-LeBron is 1st all time in RS+PS minute.
-He's 3% ahead of Kareem so far
-9% ahead of Karl Malone
-19.5% ahead of Kobe and Dirk
-Over 20% ahead of every other player.
If we look at playoff-only minutes, he's 26.6% ahead of Duncan who's 2nd, so adding extra weighting to the playoffs would only increase LeBron's minute advantage.

So in order for someone to eclipse LeBron in some sort of linear career value like this, their career average performance in a stat would need to be a greater advantage (by %) than LeBron's minute advantage. Given how much worse Karl Malone is than LeBron and how much fewer minutes everyone else has played, it's really only Kareem who has a chance to beat LeBron, and even then primarily just in stats in Era 1. Shot Clock Era (we can project for Era 2. NBA Merger Era, but we don't have any plus minus data for prime Kareem so Era 3 is obviously out.).

Of course, that's absolutely credit to LeBron. If we take a linear (maximum longevity weighting) approach to the GOAT, and don't do any era-relative correction, I would argue LeBron certainly does have the best case. But it's with noting that's far from the only criteria people have. Even if we just consider era-relative longevity, things start to get a lot less unanimous.

The modern era is perhaps best of any era to boost raw longevity. For example:
1. College: Modern players play minimal years in college, and often go right into the NBA. LeBron certainly did. This was unheard of in prior years.
-Bill Russell went to college for 4 years (3 years of varsity)
-Wilt had 3 years of college, and 1 year of the Harlem Globetrotters
-Kareem had 4 years of college
-Magic had 2 years in college
-Bird had 5 years after high school before the NBA
-Jordan had 3 years in college
etc.
The youngest players were actually not allowed to join the NBA until they were older than they would be in their senior year of college. Kareem is particularly noteworthy, given how close in minutes he is to LeBron, and how much he might rise if he had been allowed to spend those college years in the NBA. He was already a top 5 player in the NBA in his rookie year, and he likely had multiple college years where he was a top 10 player in the world. Russell, Wilt, and Bird also would have likely gained a lot of career longevity, if they had played in the NBA younger. Taking a linear approach to 'raw longevity' (not era-relative) punishes these players for not playing in the NBA younger, even when they weren't allowed to.

2. Retirement year: More recently in NBA history, players are expected and encouraged to retire at a much later year. There were people during LeBron's Miami years saying he could play until he was 40, and LeBron himself set the goal to play with his son in the NBA back in 2018 (age 33). Again, that's crazy! LeBron absolutely deserves credit for his longevity achievements. But he also played in an era that enabled this: much better sports medicine, sports science, science-informed development and practices, better equipment and weight training, more money available to spend on nutritionists and personal trainers, load management (which limits possessions per season but allows for more seasons / better health / more energy for playoffs), changing expectations on how long athletes can play, etc.
-Russell retired at age 34, when he was still a top 5 player
-Wilt retired at age 36, when he was still a top 5 player
-Bird had a major injury at age 32 (before which he was a top 5 player), then retired at 35 (when he was still an all-star)
-Magic retired at age 31, when he was still a top 5 player (with a brief return at age 36)
-Jordan retired from ages 30-31, again from ages 35-37 when he was still a top 5 player, and again at age 39.
Many of these retirements were considered normal, or expected. Would Russell really have retired at only age 34 if he played later? Would Bird and Magic have had better health and thus better longevity if they played later? Would Jordan's father have been murdered (causing his retirement for 1.5 years from ages 30–31) or would he have been retired when he was 'still on top' at age 35 if he had played in a later era?

Comparing across eras can be very difficult. There's all manner of contexts that can be hard to adjust for. Different play styles, different rules, different on and off-court factors, all of which could affect a players career value. Pace is another major example: although Russell and Wilt started later and retired earlier than they likely would have in a later era, they got to play in perhaps the fastest paced era ever, which helps mitigate (but not entirely negate) how much more career value they might have had in a later era.

Is it fair to take a linear (no extra weighting for peak/prime) raw (non-era relative) longevity-focused approach to career value? Sure. But other methods might be fair too! There's no right answer.
-Some people (almost all) do not value longevity linearly: almost everyone values peaks/prime years a little more.
-Some people are comfortable doing more a more era-relative approach.
-Some people are comfortable focusing more on player 'goodness' (independent of team context), others focus more on 'value' (in the team context they had).
-Some people are okay taking a 'goodness over time' (so they don't punish players for late starts or early retirements if they were driven more by historical and personal context than a limitation of the player, see e.g. Djoker's durability emphasis).
Any of these changes might shift the rankings around compared to a 'linear non-era-relative career value' approach.

Like I said above, LeBron definitely has one of the best cases for GOAT career. But I do think it's worth emphasizing that different criteria can produce different rankings, there's no clear 'right criteria' or 'right methodology' (although some criteria may be inconsistent and some methodologies may be less accurate for what they aim to measure). These statistical GOAT lists I gathered almost all share the same criteria and a highly similar methodology, so it makes sense there would be some similarities in their rankings. But that doesn't mean everyone has to share the same criteria/methodology!

Speaking of different methodologies...

Djoker wrote:A lot of the stats in the OP are accumulated totals so for me who doesn't have (situational) longevity as a criteria, none of those stats are frankly very close to my career GOAT list. I also don't think any of these stats even ones like RAPM or PIPM are very good at making a GOAT list on their own considering we see the likes of Karl Malone and Chris Paul near the top of almost every list. Virtually no one has those two ranked in the top 15 all time and some have them much lower. Also a huge issue with using stats that aren't available for most of NBA history.

My career GOAT list is mainly driven by three factors: peak level of play, durability (as opposed to longevity) and accolades.

Interesting criteria Djoker! I'm interested in hearing more about why you have these criteria (e.g. how you came up with these 3, what you're trying to "measure" if you aren't doing 'cumulative value over career', etc.).
For 1) Why peak as opposed to prime? Are you just weighing peak years > prime years > longevity (non-prime) years when evaluating how great someone is, or do you actually truncate e.g. non-prime years and not really consider them at all?

For 2) By emphasizing durability instead of longevity, would you say my summary above is accurate? I described this as measuring 'goodness over time'. If my interpretation of your criteria is right, then in your ranking....
a. you wouldn't punish Russell for retiring at the early age 34, which is earlier than a lot of other players. He wasn't injured and would still have been quite good if he had kept playing at age 35; the retirement age was 'situational' / based on historical context (it wasn't considered early at the time), and so Russell shouldn't be punished for it. You can use the years he did play to project how good he would have been when comparing players, or just compare how good he was in the years he did play to the other top 10 candidates at the same ages, etc. Likewise,
b., you don't punish Jordan for his mid-career retirements. He could have been just as 'good' if kept playing through 1994–1995 -- It's not like he suddenly forgot how to play basketball. He happened to retire due to personal situational context (his father was murdered), and his 2nd retirement was due also to situational context: he had just finished a 2nd 3-peat, was largely considered the GOAT at the time, "he had no more worlds to conquer", his team was about to go through a tank/rebuild phase, and it was more acceptable at the time to retire when you were still near the height of your power (e.g. Russell, Wilt, Magic/Bird) instead of playing until you couldn't any more.
Let me know if I understand your criteria right here, or if I'm misinterpreting something!

For 3), Why do you value accolades, and which accolades do you consider? Personal awards like MVP/DPOY/All-nba, team awards like rings, statistical records like most points in a career, all of the above / none of the above?
If the goal is to measure something like 'goodness over time', do you not have concerns that accolades might introduce some bias / error (that's certainly a common opinion round these parts)?

Shaq's "Ringz Erneh" apporach has certainly been argued to have some limitations /biases (see e.g. someone like Garnett, who's one of the most valuable players ever but was stuck on an outlier-bad team for an all-time great player for much of his prime). There are also examples of personal accolades where the voters probably got it wrong (e.g. Curry's lack of FMVP from 2015-2019, having Cedric Maxwell get FMVP over Bird in 1981, having Duncan never get a DPOY, etc.), or examples when personal accolades were unavailable (e.g. Russell has no DPOY lol).

Do you hold it against these players for having fewer rings/accolades? Or do you try to apply accolades+context (e.g. use accolades as another measure of a player, a tangential way to measure the greatness of a player other than stats/film study, but one that requires context to apply just like stats/film-study does)? Or perhaps there's some third option for how you apply accolades into your GOAT list?

This kind of discussion of people's criteria (especially when it agrees/differs with the basic statistical methods people have used) is just the kind of thing I was interested in discussing in this thread, so thank you for bringing this up!

DraymondGold wrote: Interesting criteria Djoker! I'm interested in hearing more about why you have these criteria (e.g. how you came up with these 3, what you're trying to "measure" if you aren't doing 'cumulative value over career', etc.).

Ok here goes!

For 1) Why peak as opposed to prime? Are you just weighing peak years > prime years > longevity (non-prime) years when evaluating how great someone is, or do you actually truncate e.g. non-prime years and not really consider them at all?

There is a lot of gray area between peak and prime. People usually take peak to mean a player at his absolute best and prime to means a player near his best. But how near does it have to be to be considered a player's prime and when does prime become peak? It's not perfectly clear also how short a peak can be. Is it one season? One series? God forbid one game?

Anyways feel free to replace the word peak with prime in my criteria.

Non-prime years are basically irrelevant. They have so little value. And I think this is where people go wrong as well when assessing career value. For instance, Ben Taylor, who I respect immensely, values 1 GOAT level season (say 1993 Jordan) as equivalent to 3 All-NBA level seasons (say 1993-1995 Reggie Miller) based on his CORP model but in reality I think 1 GOAT level season is equal to 5 or even 10 All-NBA level seasons! The first reason why is the issue of scarcity. There is one MJ in the NBA (or none!) but there are many players of Reggie Miller's caliber. The second reason why is the issue of salary and the salary cap. Both Jordan and Miller would earn max contracts but Jordan being so much better than Miller makes it way easier to actually build a supporting cast good enough to win. Neither the scarcity nor the good salary value of the absolute best player is considered with a model like CORP.

For 2) By emphasizing durability instead of longevity, would you say my summary above is accurate? I described this as measuring 'goodness over time'. If my interpretation of your criteria is right, then in your ranking....
a. you wouldn't punish Russell for retiring at the early age 34, which is earlier than a lot of other players. He wasn't injured and would still have been quite good if he had kept playing at age 35; the retirement age was 'situational' / based on historical context (it wasn't considered early at the time), and so Russell shouldn't be punished for it. You can use the years he did play to project how good he would have been when comparing players, or just compare how good he was in the years he did play to the other top 10 candidates at the same ages, etc. Likewise,
b., you don't punish Jordan for his mid-career retirements. He could have been just as 'good' if kept playing through 1994–1995 -- It's not like he suddenly forgot how to play basketball. He happened to retire due to personal situational context (his father was murdered), and his 2nd retirement was due also to situational context: he had just finished a 2nd 3-peat, was largely considered the GOAT at the time, "he had no more worlds to conquer", his team was about to go through a tank/rebuild phase, and it was more acceptable at the time to retire when you were still near the height of your power (e.g. Russell, Wilt, Magic/Bird) instead of playing until you couldn't any more.
Let me know if I understand your criteria right here, or if I'm misinterpreting something!

You pretty much got it. Longevity isn't a valid criteria when judging greatness IMHO because it's largely determined by factors that have nothing to do with basketball.

Some factors that influence career longevity that have nothing to do with a player's basketball ability:
- eligibility of high school players; getting anywhere from 1-4 years head start is a big deal
- advancements in nutrition, training, medicine, footwear etc.; everyone knows modern athletes across all sports are playing longer than ever before
- exigent circumstances in a player's personal life (ex. Jordan's father's death or Magic's HIV)
- feeling of "no more worlds to conquer" as you put it so well (ex. Jordan in 1998 or Russell in 1969; if those guys lost, would they still retire? And conversely, if Lebron won 7 rings by 2018 and was proclaimed the GOAT by almost everyone, does he play 21+ years in the NBA?)

So longevity is a poor criteria to use.

Now, there is a related criteria that is important and that's durability. Having one of the greatest primes/peaks ever (ex. Bill Walton) doesn't put you anywhere near the GOAT conversation if you can't sustain it. And Walton COULDN'T sustain it. His body wouldn't let him. And injuries are a basketball reason. Penalizing players who had a lack of durability due to injuries is not only valid but imperative. We can't pretend that those who can't play are as valuable as those who can. That would be ridiculous.

For 3), Why do you value accolades, and which accolades do you consider? Personal awards like MVP/DPOY/All-nba, team awards like rings, statistical records like most points in a career, all of the above / none of the above?
If the goal is to measure something like 'goodness over time', do you not have concerns that accolades might introduce some bias / error (that's certainly a common opinion round these parts)?

Shaq's "Ringz Erneh" apporach has certainly been argued to have some limitations /biases (see e.g. someone like Garnett, who's one of the most valuable players ever but was stuck on an outlier-bad team for an all-time great player for much of his prime). There are also examples of personal accolades where the voters probably got it wrong (e.g. Curry's lack of FMVP from 2015-2019, having Cedric Maxwell get FMVP over Bird in 1981, having Duncan never get a DPOY, etc.), or examples when personal accolades were unavailable (e.g. Russell has no DPOY lol).

Do you hold it against these players for having fewer rings/accolades? Or do you try to apply accolades+context (e.g. use accolades as another measure of a player, a tangential way to measure the greatness of a player other than stats/film study, but one that requires context to apply just like stats/film-study does)? Or perhaps there's some third option for how you apply accolades into your GOAT list?

Context is everything.

The primary accolades are championships won as the best player on the team. In your Curry example, he has only 1 FMVP but he won four rings as the best player on his team and that's a big deal. The purpose of basketball is to win championships.

Every (serious) player that ever played in the NBA... if you asked them "Whose career would you choose?" they would (should...) all say "Bill Russell's". 11 rings is the greatest record in NBA history. BY FAR! And the crazy thing is that Russell probably won all 11 of those as the best player on his team.

The easiest way for Lebron or any other modern player to become "the GOAT" is to win a lot of rings. If Lebron had 7 rings as the best player on his team, you better believe that a hell of a lot more people would believe he's the best ever. Of course, the mark should be 12. Russell is underappreciated but that's a story for another day.

MVP's and other accolades are more situational and don't factor in that much but you kind of expect these accolades to follow a player with a truly great prime anyway. It would certainly be unusual for a GOAT candidate not to win a single MVP. And no I don't have a problem with mentally assigning Bill Russell with like 8 DPOY awards. Or giving Duncan a DPOY and removing one from Marcus Camby.

This kind of discussion of people's criteria (especially when it agrees/differs with the basic statistical methods people have used) is just the kind of thing I was interested in discussing in this thread, so thank you for bringing this up!

Yes it is interesting. Very much so!

I still use lots of stats to determine whose prime is better. But I strongly believe there is no one holy grail stat.

Here are some other GOAT lists:
RAPTOR through the 2022 season with playoffs weighted twice as much. Best season is weighted 100%. Next best season is weighted 95%. Next Season is weighted 90%. Etc


1 LeBron James
2 Michael Jordan
3 John Stockton
4 Magic Johnson
5 Chris Paul
6 Larry Bird
7 Kobe Bryant
8 Tim Duncan
9 Scottie Pippen
10 Stephen Curry
11 Kevin Garnett
12 James Harden
13 Charles Barkley
14 David Robinson
15 Jason Kidd
16 Hakeem Olajuwon
17 Clyde Drexler
18 Karl Malone
19 Shaquille O'Neal
20 Dirk Nowitzki

BPM VORP Playoffs Weighted Twice As Much


1 LeBron James
2 Michael Jordan*
3 John Stockton*
4 Tim Duncan*
5 Karl Malone*
6 Chris Paul
7 Magic Johnson*
8 Kevin Garnett*
9 Kevin Durant
10 Kobe Bryant*
11 Kareem Abdul-Jabbar*
12 Dirk Nowitzki*
13 Larry Bird*
14 Julius Erving*
15 Shaquille O'Neal*
16 James Harden
17 Charles Barkley*
18 Hakeem Olajuwon*
19 David Robinson*
20 Stephen Curry

I confess to taking a shortcut. I opened each list, looked for Bill Russell, then eliminated any that didn't have him top 10. That said, the closest to my personal list is Backpicks BPM ii.

DraymondGold wrote:~Which stat produces the best GOAT career list?~

Most stats are not designed with GOAT career lists in mind. They’re often designed to quantify goodness or value in a current or future time range, using various inputs (e.g. box stats, plus minus data, WOWY data). They are often designed with an eye towards a single season or several seasons or at most a prime, not a career.

Still, it’s interesting to consider how the various stats compare to our usual GOAT list. I’ve collected a few of the more accessible stat GOAT lists. There’s a few different time ranges (e.g. shot clock era vs NBA merger vs plus minus era), and occasionally when a stat publishes their own GOAT list, they’ll have different weightings (e.g. the RAPTOR GOAT list includes extra prime weighting), so I’ve included those here.

A few discussion questions:
-Which stats produce the closest GOAT list to yours, or to the usual GOAT list? Why does that stat gets closest?
-What weightings are optimal? (e.g. should playoffs get extra weighting, should peaks/primes get extra weighting, and if so how much?)
-Are there any players or trends/archetypes where the stats diverge significantly from your GOAT list?
-When you disagree, what are you valuing that the stat doesn’t measure, or vice versa?

1. Shot Clock Era / Full NBA History (1949 NBA / 1955 shot clock era — present)

A. Win Shares: RS + PS Career Total (https://www.basketball-reference.com/leaders/ws_career_c.html) (full NBA history)

Spoiler:

1. LeBron James
2. Kareem Abdul-Jabbar
3. Wilt Chamberlain
4. Karl Malone
5. Michael Jordan
6. Tim Duncan
7. Chris Paul
8. Dirk Nowitzki
9. John Stockton
10. Shaquille O'Neal
11. Julius Erving
12. Kevin Garnett
13. Oscar Robertson
14. Artis Gilmore
15. Kobe Bryant
16. Kevin Durant
17. Charles Barkley
18. David Robinson
19. Reggie Miller
20. Moses Malone
21. Bill Russell
22. Jerry West
23. Magic Johnson
24. James Harden
25. Hakeem Olajuwon

other notable ranks:
27. Larry Bird
32. Steph Curry
35. Dwight Howard
36. Jason Kidd
37. Clyde Drexler
42. Dwyane Wade
44. Steve Nash
46. Patrick Ewing
57. Manu Ginobili
60 Nikola Jokic
63. George Mikan
67. Anthony Davis
68. Kawhi Leonard
74. Russell Westbrook
76. Giannis Antetokounmpo
217. Draymond Green

Bi. Backpicks BPM: RS + PS Career Total (https://thinkingbasketball.net) (1955–2024)

Spoiler:

1. LeBron James
2. Wilt Chamberlain
3. Kareem Abdul-Jabbar
4. Michael Jordan
5. Bill Russell
6. Karl Malone
7. Magic
8. Stockton
9. Larry Bird
10. Duncan
11. Chris Paul
12. Harden
13. Kobe Bryant
14. Hakeem Olajuwon
15. Durant
16. Pippen
17. Dirk
18. Shaq
19. Garnett
20. Steph Curry
21. Barkley
22. Robinson
23. Oscar

Bii. Backpicks BPM: RS + 7x PS Career Total (PS time given 7x weighting) (https://thinkingbasketball.net) (1955–2024)

Spoiler:

1. LeBron James
2. Wilt Chamberlain
3. Kareem Abdul-Jabbar
4. Michael Jordan
5. Bill Russell
6. Magic
7. Duncan
8. Larry Bird
9. Pippen
10. Kobe Bryant
11. Karl Malone
12. Durant
13. Shaq
14. Stockton
15. Hakeem Olajuwon
16. Harden
17. Chris Paul
18. Steph Curry
19. Dirk
20. Barkley
21. Garnett
22. Wade
23. Robinson

Biii. Backpicks BPM: CORP (PS time given 7x weighting, championship odds follow a superlinear function so peaks are weighted over primes over non-prime years) (1955–2018)

Spoiler:

1. LeBron
2. MJ
3. Kareem
4. Wilt
5. Magic
6. Russell
7. Duncan
8. Karl
9. Stockton
10. Hakeem
11. Garnett
12. Dirk
13. Shaq
14. Bird
15. Pippen
16. Kobe

Note: missing years of players post-2018: Steph Curry, Kevin Durant, Chris Paul, Nikola Jokic, Giannis Antetokounmpo, Kawhi Leonard, etc.
(

Read on Twitter

?s=20)

C. Career WOWYR (‘adjusted WOWY’) (1955-2016) (https://thinkingbasketball.net/metrics/wowyr/)

Spoiler:

1. Michael Jordan
2. John Stockton
3. Magic Johnson
4. LeBron James
5. Oscar Robertson
6. Paul Pierce
7. Tim Duncan
8. Dikembe Mutombo
9. Dirk Nowitzki
10. Hakeem Olajuwon
11. Kobe Bryant
12. Wilt Chamberlain
13. Steve Nash
14. Jerry West
15. Shaquille O'Neal
16. David Robinson
17. Patrick Ewing
18. Bill Russell
19. Clyde Drexler
20. Kevin Garnett
21. Kareem Abdul-Jabbar
22. Bill Laimbeer
23. Rasheed Wallace
24. Larry Bird
25. Vlade Divac

other notable ranks:
26. Chris Paul (pre-2017)
32. Scottie Pippen
33. Karl Malone
36. Charles Barkley
46. Jason Kidd
52. Reggie Miller
~83. Julius Erving
~92. Moses Malone
~98. Dwyane Wade
Disclaimer: I may have missed occasional players outside the top 50, so Erving, Malone, and Wade might have slightly lower ranks if I did.

Missing/partial primes post-2016: Steph Curry, Kevin Durant, Chris Paul, Nikola Jokic, Giannis Antetokounmpo, Kawhi Leonard, etc.

2. NBA Merger Era (1974 defensive stat era / 1977 merger — present)

D. Basketball Reference BPM: RS + PS VORP (https://www.basketball-reference.com/leaders/vorp_career_c.html) (1974–2024)

Spoiler:

1. LeBron James
2. Michael Jordan
3. John Stockton
4. Karl Malone
5. Tim Duncan
6. Chris Paul
7. Kevin Garnett
8. Kevin Durant
9. Kareem Abdul-Jabbar (missing 1970–1973)
10. Magic Johnson
11. Dirk Nowitzki
12. Kobe Bryant
13. Julius Erving
14. Larry Bird
15. James Harden
16. Shaquille O'Neal
17. David Robinson
18. Charles Barkley
19. Hakeem Olajuwon
20. Jason Kidd
21. Stephen Curry
22. Clyde Drexler
23. Scottie Pippen
24. Reggie Miller
25. Dwyane Wade

other notable ranks:
27. Nikola Jokic
30. Russell Westbrook
32. Giannis Antetokounmpo
34. Kawhi Leonard
36. Manu Ginobili
37. Patrick Ewing
39. Anthony Davis
43. Steve Nash
48. Moses Malone
89. Dikembe Mutombo
101. Draymond Green

Missing primes pre-1974: Bill Russell, Wilt Chamberlain, Oscar Robertson, Jerry West.

E. Fiverthirtyeight’s RAPTOR: RS + PS JAWS (average between career total and 7 year prime) (https://fivethirtyeight.com/features/lebron-or-mj-raptor-picks-the-best-nba-players-of-the-past-40-years/) (1977–2023)

Spoiler:

1. LeBron James (projecting 2020–2024; #2 through 2019)
2. Michael Jordan
3. John Stockton
4. Chris Paul
5. Magic Johnson
6. Tim Duncan
7. Kevin Garnett
8. Kobe Bryant
9. Larry Bird
10. Stephen Curry (projecting 2022 ~ 2023; #19 through 2019)
11. Jason Kidd
12. Scottie Pippen
13. David Robinson
14. Charles Barkley
15. Karl Malone
16. Clyde Drexler
17. Hakeem Olajuwon
18. Dirk Nowitzki
19. Shaquille O’Neal
20. Gary Payton
21. Kevin Durant (projecting 2023 ~ 2022)
22. ? (Not included in original list)
23. Dwyane Wade
24. ? (Not included in original list)
25. Ray Allen

other notable (known) ranks:
27. Manu Ginobili
28. Paul Pierce

Missing/partial primes pre-1976: Bill Russell, Wilt Chamberlain, Oscar Robertson, Jerry West, Kareem Abdul Jabbar (1970–1976), Julius Erving (1972–1976), Moses Malone (1975–1976),

F. PIPM: RS + PS Career Total (in units of ‘Wins added’) (https://docs.google.com/spreadsheets/d/1EIZvj_3-9SZULWomHz54V1CPL092j70u_0vUhoEEaIk/edit?usp=share_link) (1977–first few games of 2021)

Spoiler:

1. LeBron
2. Kareem (projecting 1970–73. 10th discounting 1970–1973)
3. Duncan
4. Garnett
5. Jordan
6. Stockton
7. Karl Malone
8. Dirk
9. Hakeem
10. Shaq
11. Robinson
12. Jason Kidd
13. Chris Paul (projecting 2021–2023. 13th discounting 2021+)
14. Stephen Curry (projecting 2021–2023. 32nd discounting 2021+)
15. Magic
16. Barkley
17. Bird
18. Kobe
19. Paul Pierce
20. Durant (projecting 2021–2023. 25th discounting 2021+)
21. Patrick Ewing
22. Pippen
23. Drexler
24. Gary Payton
25. Vince Carter
26. Dwight

Disclaimer: I projected a few players' performance based on continued reasonable, conservative per-game prime-level performance for their primes and non-prime performance for non-prime years. But I've included their original ranks if you'd like to see them without projecting for the missing data.

other notable ranks:
27. Reggie Miller
29. Ray Allen
30. Ben Wallace
31. Moses Malone (missing years)
32. Dikembe Mutombo
33. Dwyane Wade
34. Harden (missing years)
~36. Manu Ginobili
~37. Steve Nash
~39. Russell Westbrook (missing years)
~40. Julius Erving (missing years)
~58. Kawhi Leonard (missing years)
~61. Paul George (missing years)
~78. Anthony Davis (missing years)

Missing/partial primes pre-1976: Bill Russell, Wilt Chamberlain, Oscar Robertson, Jerry West, Kareem Abdul Jabbar (1970–1976), Julius Erving (1972–1976), Moses Malone (1975–1976),

C. Plus Minus Era (1994 plus minus era / 1997 play by play era — present)

Gi. Backpicks Augmented Plus Minus: RS + PS CORP (PS games given 7x weighting like Backpicks BPM) (https://thinkingbasketball.net/2021/07/10/playoff-plus-minus-part-iii-changes-in-the-postseason/) (1994–2024 RS, 1997–2024 PS)

Spoiler:

1. LeBron James
2. Dirk Nowitzki
3. Tim Duncan
4. Kevin Garnett
5. Chris Paul
6. Stephen Curry
7. Shaquille O'Neal (missing 1993 RS, 1994–1996 PS)
8. Kobe Bryant
9. Kevin Durant
10. Steve Nash
11. Paul Pierce
12. James Harden
13. Jason Kidd
14. Vince Carter
15. Karl Malone (missing 1986–1993 RS, 1986–1996 PS)
16. Dwyane Wade
17. Dwight Howard
18. Nikola Jokic
19. Draymond Green
20. Giannis Antetokounmpo
21. Ray Allen
22. Pau Gasol
23. Manu Ginobili
24. Ben Wallace
25. Chauncey Billups
26. Rasheed Wallace
27. Michael Jordan (missing RS 1985–1993, 1985–1987 PS)
28. Kyle Lowry

Missing/partial primes pre-1994: Michael Jordan, Shaquille O’Neal, Hakeem Olajuwon, John Stockton, Karl Malone, David Robinson, etc.

Gii. Backpicks Augmented Plus Minus: RS + 7x PS Career Total (PS games given 7x weighting) (1994–2024 RS, 1997–2024 PS)

Spoiler:

1. LeBron James
2. Tim Duncan
3. Dirk Nowitzki
4. Shaquille O’Neal (missing 1993 RS, 1994–1996 PS)
5. Steph Curry
6. Kobe Bryant
7. Michael Jordan (missing RS 1985–1993, 1985–1987 PS, using PS estimated AuPM 1988–1996)
8. Kevin Garnett
9. Chris Paul
10. Kevin Durant
11. Draymond Green
12. James Harden
13. Manu Ginobili
14. Dwyane Wade
15. Jason Kidd
16. Steve Nash
17. Paul Pierce
18. Ray Allen
19. Nikola Jokic
20. Chauncey Billups
21. Karl Malone (missing 1986–1993 RS, 1986–1996 PS)
22. Pau Gasol
23. Rasheed Wallace
24. Dwight Howard
25. Kawhi Leonard
26. Ben Wallace
27. Vince Carter
28. Giannis Antetokounmpo

Missing/partial primes pre-1994: Michael Jordan, Shaquille O’Neal, Hakeem Olajuwon, John Stockton, Karl Malone, David Robinson, etc.

H. Engelmann RAPM: RS + PS Career Value (https://docs.google.com/spreadsheets/d/1bg8KxzagN7D0O16EmUO9_kCyXwthEUjKywlrWPQUQt8/edit#gid=0) (1997–2023.5):

Spoiler:

1. LeBron James
2. Kevin Garnett
3. Tim Duncan
4. Dirk Nowitzki
5. Chris Paul
6. Kevin Durant
7. Stephen Curry
8. Paul Pierce
9. Shaquille O'Neal
10. James Harden
11. Jason Kidd
12. Draymond Green
13. Manu Ginobili
14. Damian Lillard
15. Jrue Holiday
16. Vince Carter
17. Rasheed Wallace
18. Nikola Jokic
19. Steve Nash
20. Kobe Bryant
21. Ray Allen
22. Mike Conley
23. Kyle Lowry
24. Andre Iguodala
25. Paul George

Missing/partial primes pre-1997: Michael Jordan, Shaquille O’Neal, Hakeem Olajuwon, John Stockton, Karl Malone, David Robinson, etc.

I. DARKO Daily Plus Minus: RS + PS Career Value (https://apanalytics.shinyapps.io/DARKO//) (1998–2024)

Spoiler:

1. LeBron James
2. Tim Duncan
3. Garnett (missing 96–97)
4. Dirk
5. Chris Paul
6. Steph Curry
7. Kevin Durant
8. James Harden
9. Shaq (missing 94-97)
10. Paul Pierce
11. Kobe (missing 1997)
12. Kidd (missing 1995–1997)
13. Wade
14. Nash (missing 1997)
15. Kawhi Leonard
16. Rasheed Wallace
17. Nikola Jokic
18. Giannis Antetokounmpo
19. Manu
20. Kyle Lowry
21. Vince Carter
22. Damian Lillard
23. Ray Allen
24. Mike Conley
25. Andre Iguodala
26. Dwight Howard
27. Draymond Green
28. Chauncey Billups

Disclaimer: I may have missed occasional players outside the top 20, so a few of the lowest ranked players here might be ranked slightly lower.

Can you do Back Pick BPM. Playoffs Weighted Twice as Much. Best Year (In terms of WAR) weighted 100%, Second Best Year weighted 95%, Third Best Year weighted 90%, etc.