Poll: Your all-time greatest players

[Crow]

If RAPM is used in this discussion, shouldn't it be 28 year RAPM? If "not", why not? Why 3-5 year?

Jokic 4th here, for those covered fully or adequately:

https://nbarapm.com/datasets/LifetimeRAPM

Jokic #1 on career BPM and by more than 10%.

Still being active can both help and hurt. Evaluate just "to date" or can project if you want, for this opinion poll.

[eminence]

I brought up mid-term because that is what BPM is based off.

A couple reasons why I personally don't use super long RAPM samples:
-The matchup matrix gets more and more sparse, leading to less mathematical certainty (and folks rarely include certainty levels)
-Plenty of players have played at significantly different levels through their careers, I don't find trying to funnel them down to 1 number particularly useful or truthful (this applies to non-APM stats as well). Eg - post injury Kobe wasn't any good, but to the best of my knowledge those seasons don't lower his career championship odds (in the style of Ben Taylor's CORP)
-Long term APMs seem to introduce dubious concepts like 'age curves' and further black box the process

[Mike G]

A player may increase his VORP by playing more minutes (or more seasons), but rarely or perhaps never increase his BPM this way.
I see b-r.com has biggest VORP seasons combining RS+PO:
https://www.basketball-reference.com/le ... son_c.html

However you may define 'elite' level over x seasons, here's a count at some arbitrary levels:

Code: Select all

top 5  13+
4 - Jordan
1 - LeBron

top 10  12+
6 - Jordan
2 - LeBron
1 - Erving, Jokic

top 25  10.8+
9 - Jordan
6 - LeBron
2 - Jokic
1 - Erving, Robinson, Garnett, Shaq,
- - Durant, Paul, Curry, Duncan,
- - Kareem

top 50  9.89+
11 - LeBron
9 - Jordan
4 - Bird 
3 - Jokic
2 - Kareem, Erving, Durant, Curry, Magic,
- - Robinson, Paul
1 - Garnett, Shaq, Duncan, Dirk, McGrady,
- - Wade, Doncic, Olajuwon, Harden

top 100  8.43+
14 - LeBron
10 - Jordan
5 - Jokic, Bird, Kareem, Harden
4 - Magic, Erving, Stockton, Robinson
3 - Garnett, Durant, Malone, Wade, Paul
2 - Duncan, Curry, Shaq, Westbrook, Kobe,
- - Olajuwon, Giannis, Dirk, Barkley
1 - McGrady, Carter, Doncic, Kawhi,
- - McGinnis, Drexler

top 250 (2 or more)  6.51+
15 - LeBron
11 - Jordan
10 - Magic, Malone, Stockton
8 - Bird, Garnett, Durant, Robinson, Paul
7 - Jokic, Kareem, Erving, Kobe
6 - Duncan, Shaq, Pippen, Harden, Giannis,
- - Dirk, Drexler
5 - Curry, Barkley
4 - McGrady, Kawhi, Wade, Olajuwon, Payton
3 - Doncic, Westbrook, Payton, Billups, Davis
2 - Carter, Porter, Marion, Shai, Ginobili
- - George, Ewing, Embiid, Butler, Kidd,

Same deal with Win Shares, RS+PO:

Code: Select all

Top 5  25.5+
3 - Wilt
2 - Kareem

Top 10  24.8+
3 - Wilt
2 - Kareem, Jordan, Mikan
1 - LeBron

Top 25  22.87+
6 - Jordan
4 - Wilt, Kareem
3 - Mikan
2 - LeBron
1 - Duncan, Shaq, Oscar, Dirk, Gilmore

Top 50  19.77+
9 - Jordan
6 - Wilt
5 - LeBron, Kareem
3 - Mikan
2 - Duncan, Jokic, Garnett, Durant, 
- - Robinson, Gilmore
1 - Curry, Shaq, Oscar, Bird, Garnett, Erving, 
- - West, Dirk, Hawkins, Paul, Russell, Groza

Top 100  18.09+
9 - Wilt, Jordan
8 - LeBron
6 - Kareem
4 - Magic, Bird
3 - Oscar, Malone, West, Mikan, Robinson, Paul, 
- - Russell, Gilmore
2 - Duncan, Curry, Shaq, Jokic, Garnett, Durant, 
- - Erving, Olajuwon, Dirk, Barkley, Groza
1 - P Gasol, Johnston, Garnett, Fulks, 
- - Jimmy Jones, Harden, Wade, Dwight, Issel, 
- - Hawkins, Billups, McAdoo, Feerick

Top 250  14.94+
12 - LeBron
11 - Wilt
10 - Jordan, Malone
9 - Kareem
8 - Erving, Magic, Stockton
7 - Durant, Russell
6 - Shaq, Oscar, Kobe, Mikan, Robinson,
- - West, Dirk
5 - Bird, Harden, Paul, Jokic,
- - Barkley, Garnett
4 - Frazier, Duncan, Johnston, 
- - Billups, Pettit, Gilmore
3 - Curry, Pippen, P Gasol, Moses, Wade, Dwight
2 - Reed, Gobert, Barry, Pierce, Kawhi, Olajuwon, 
- - Baylor, Macauley, McAdoo, Davis, Groza

ABA seasons are also counted.
The old-timers qualify in spite of shorter schedules and shorter postseasons.

[Crow]

Fair enough eminence. Appreciate the reply.



Lots of potential considerations to this assignment.


Such a tiny vote. 16 got 3+ votes. 12 got 5+.
Unless you really want to fight about top 5, there is a pretty high amount of similarity.


I have no real interest in fighting about a top 5 or a top 10 over a top 16 or 20.

[Mike G]

Jokic #1 on career BPM and by more than 10%.

Did you notice that thru age 29 he's well behind Jordan's BPM and barely ahead of LeBron?
Trailing even more (and probably Kareem) in VORP thru this age. Totally swamped in postseason VORP.
Here it is again:

Code: Select all

age 20-29   regular seasons        postseasons
goat?     min     BPM   VORP     min    BPM   VORP  <O>
Jokic    23062   10.3   71.4    3038   10.8    9.8   1
Jordan   25842   10.9   84.0    4645   12.1   16.5   3
LeBron   30154   10.1   91.6    6717   10.0   20.4   2
Kareem   26349    9.2*  74.3*   3023    8.7*   7.8*  1

Schtevie seemed to be saying that minutes don't matter, but in fact Jokic doesn't seem to be as strong after 30-ish minutes. In 4th quarters of games he has gone over 40 minutes, his rebound rate is about half of normal.

[Crow]

I didn't break it down by age. Not really invested.

Difference levels to discussion. Is he #1 is different than does he belong in top 10 or is he closer to 20.

[DSMok1]

I brought up mid-term because that is what BPM is based off.

A couple reasons why I personally don't use super long RAPM samples:
-The matchup matrix gets more and more sparse, leading to less mathematical certainty (and folks rarely include certainty levels)
-Plenty of players have played at significantly different levels through their careers, I don't find trying to funnel them down to 1 number particularly useful or truthful (this applies to non-APM stats as well). Eg - post injury Kobe wasn't any good, but to the best of my knowledge those seasons don't lower his career championship odds (in the style of Ben Taylor's CORP)
-Long term APMs seem to introduce dubious concepts like 'age curves' and further black box the process

I second all of these comments. My favorite version of RAPM is three or at most four year duration, with a decent non-box score prior to get it started in the right general vicinity.

The long-term rapms must include aging or they won't work at all, and if they include aging it by definition must be an assumed average aging curve for everyone... And if someone has an unusual real aging curve it can throw off a bunch of player's values.

[TeemoTeejay]

I wouldn’t call aging curves dubious but I do agree that in this type of context where we only care about really a small minority of results assuming the exact same one for everyone may not be ideal. For purely explanetory reasons getting individual aging curves based on their varied box score production changes throughout the seasons could be interesting, might be somewhat impractical though but still

Within the context of this kind of thing, perhaps having a separate set of priors for the players you wanna evaluate and the players around them make sense though, one group with a more non box score PI type prior and the other a more traditional box score one

[DSMok1]

Aging curves are not dubious in and of themselves, in terms of their use as a general tool. But as we know each person has a very unique year to year variation curve that will also be impacted by injuries and other causes.

But that will mess with long-term RAPM. For instance, if a player is elite at a very young or very old age, that will tend to cause that same player's peak years to be overestimated in the model and thus depressing their teammates. Similarly, if an elite player is injured one year, the model will assume that player is still elite and other players will get demerited for that bad season.

[schtevie]

So, a few points in reply to various comments...

First, it is refreshing for the creator of a statistic to be most concerned about its relevance and efficacy, and I look forward to the implications of the next round of BPM "de-Russell Westbrookification".

This having been said, I still am scratching my head as to how Daniel has KAJ ranked aboveJokic. Barring Jokic being ineligible for GOAT consideration, owing to the fact that only the supposed first half of his career has occurred to date or one imposes serious weight on post-season success (more on that below) if you go line by line, at a comparable stage of their careers, looking at every stat that explains player efficiency, with the exception of block percentage, Jokic dominates.

The point being that for KAJ to show up as superior in BPM terms post-whatever revision is made to the statistic's formulation, it would appear that the reweighting of block percentage will have to dominate the effect of all others (total shooting, offensive rebounding, defensive rebounding, assists, steals, and turnovers). Furthermore, this reformulation would need to need to eliminate what is now a 4.7 point differential in peak BPM.

As for Kareem's record being incomplete owing to unavailable statistics from his early career possibly overturning this concluion, I simply don't see how that dog hunts. If you look at the statistical record, his aging curve well identified; what is missing is the beginning of his career. But there is simply nothing in the record of the stats that are available in his early years to suggest that there might be a missing second peak, never mind one that might show a 4.7 points greater BPM rating than the one actually observed.

Moving on to the relevance of post-season success in GOAT debates, I simply do not understand the weight that so many folks here assign to this factor. A very nuanced argument can be made for some relevance (though this would be really difficult to objectively estimate). But to inject a pure measure of team success in what is a team sport as assert that as being relevant to the appraisal of the individual, is really kinda weird.

If we stick with BPM as the metric, Jokic has had just one year with a strong second player alongside him: Jamal Murray last year at 4.1. In the championship year, his best BPM running mate was Aaron Gordon at 2.1. I am hoping that this isn't a point that needs emphasis in this forum, but it is unlikely for a team to win a championships when the second best player clocks in with a 2.1 BPM rating. Perhaps that was a singular event in NBA history? At least I am sure that Kareem's championship teams never faced such a shortcoming in the "supporting cast" (even for the three rings when he was part of it).

P.S. PFs are a part of the BPM framework, and I hadn't mentioned these above. No matter, as it turns out, but if you compare Jokic' five peak years (what are on going) to the six best Kareem BPM seasons, the average PFs per 100 possessions differs by only 0.01 (in Kareem's favor).

[Mike G]

looking at every stat that explains player efficiency, with the exception of block percentage, Jokic dominates.

Well, Kareem had 3 years (before BPM) of .320-.340 WS/48 and did that over 42.4 min./G. Jokic has been .296-.313 now in his 5th year.

Kareem's 3 super seasons would be equivalent to BPM of ~12.2 - 13.2, while Jokic is having a 4th consecutive year at 13+, going 34.5 mpg.
So Kareem, in a possibly slower era, averaged about 6 minutes of rest to Joker's 14 min.

During ages 22-29, both have scored 34 points per 48 minutes. Jokic has shot .641 TS% to Kareem's .579.
And Jokic (Den) opponents have hit .568 TS% while Mil/LA opponents shot .478 in this interval. Kareem took more shots and hit .101 better than his opposition, and Jokic has created that differential in just one season (2023.)

Kareem outrebounded his avg opponent by 17.6 to 9.6, per 48 min. A ratio of 1.82
Joker has exactly doubled his avg counterpart, 16.6 to 8.3 per48

Jokic has had the luxury of more than twice as many rest minutes. This year he's playing the most of his career, 36.4 mpg. Kareem went to age 34 before he reduced to that level.

So the differences (thru age 29) are mostly Joker's incredible passing vs Kareem's defense; and the value of being on the floor an extra 25-30% .
When Jokic has doubled his current totals, we have a fair career comparison.

[schtevie]

Mike, color me more than a wee bit skeptical that Kareem's three years of WS/48 you mentioned would translate to a BPM of 12.2 to 13.2. How did you establish such equivalence?

If one cared, one could put in informed "upper bound" hypothetical values for the missing variables in the BPM formula. I'm not interested in going through the whole exercise, but just eyeballing this, your suggested range doesn't seem remotely plausible.

And there are a couple of other reasons to be skeptical about an earlier much higher peak to Kareem's career. Has there ever in the history of the NBA been a player, a center in particular, with such an odd BPM aging curve? I can't think of one and having poked around a bit, couldn't find one.

Second, let's consider the implications of your conjectured Kareem's crazy-good, early-career BPM stats. Like with WS, there is an approximate adding-up constraint for BPM, where the possession-weighted, contributions of individuals sum up approximately to the team's net points per 100 possessions (ORtg - DRtg). Given the minutes (possessions) KAJ played, if you stipulate that his BPM was in the 12 to 13 range, you are implicitly making a very negative appraisal of Oscar Robertson et al. Now, you may be comfortable with that, but maybe not?

[eminence]

On the level of teams the '71-'73 Bucks were (relative to their competition, and especially '71/'72) a +12 or higher BPM doesn't necessitate a particularly low assessment for Oscar and crew.

See the '16 Warriors with Steph at +11.9 and Dray at +5.5.

Was Kareem really at his best those years, ehh, I'm not sure - I do prefer his defensive activity level those seasons and so do tend to go with '71/'72 as his best 2 season stretch, but that wouldn't show up much in a BPM. Given other measures I strongly suspect they'd be his best seasons by BPM as well. I'll agree on Mike's ranges being a tad high but not by much, I'd guess at least +11 and +12 seems fully possible for nearest full point estimates.

[Mike G]

As mentioned, Jokic has consistently been in the .300-.310 WS/48 range and with higher BPM than what I estimate for Kareem with .320-.340 WS/48. So I don't get the skepticism.
BPM loves Jokic and his assists; Kareem's post-1973 BPM and WS/48 equivalencies are quite in line with the general population. Most deviation is due to team success, which is a big part of WS. So I cannot concur that 12-13 BPM in his peak "doesn't seem remotely plausible" or even "a tad high ".

Wilt Chamberlain also peaked early on, roughly at the same age as Kareem. Both later adjusted to the wear and tear of excessive minutes at a young age. There are plenty of players who did not adjust so well and did not level off, much less return to close to their age 23-26 form: Reed, McAdoo, Haywood, Hawkins, Lanier, McGinnis just from the '70s. Kemp, Coleman, ... honestly the list feels endless.

"...implicitly making a very negative appraisal of Oscar Robertson et al. Now, you may be comfortable with that, but maybe not?"
The Bucks' PER and WS/48 are there at b-r.com; that's what we've got, and Kareem was The Man, any way you cut it.
https://www.basketball-reference.com/te ... l_advanced

[Mike G]

... to the relevance of post-season success in GOAT debates, I simply do not understand the weight that so many folks here assign to this factor...

One big relevancy is that on a good team, production is shared more, so boxscore stats are a bit muted, vs being the star player on a bad team. Winning means you get into playoffs and possibly past the 1st round or even to the Finals.

If a player by luck finds himself in a dynasty or 2 (Robt. Horry) he may get more than his due share of playoff games/minutes. But if he excels in these 'big games', he's even more likely to advance and rack up the WS and VORPs.

Suppose we reduce the value of total postseason minutes/WS/VORP while maintaining the importance of per-minute Excellence.
Since Win Shares goes back to 1952, lets use that as a first pass, to be inclusive of the oldtimers. Lets just take the square root of WS and multiply by the WS/48, for regular seasons and for playoffs; and just add them together.
Here's the alternative top 50 in such a scheme.

Code: Select all

xx      Win Shares:      RS   WS/48   RSx      PO  WS/48   POx
5.46   LeBron James      270  .221   3.64      59  .238   1.82
5.27   Michael Jordan    214  .251   3.66      40  .255   1.61
5.02   Wilt Chamberlain  247  .248   3.90      31  .200   1.12
4.93 Kareem Abdul-Jabbar 273  .228   3.78      36  .193   1.15
4.20   Chris Paul        214  .228   3.34      21  .187    .86

4.20   Tim Duncan        206  .209   3.00      38  .194   1.19
4.18   David Robinson    179  .250   3.34      18  .199    .83
3.99   Magic Johnson     156  .225   2.81      33  .208   1.19
3.85   Nikola Jokić      126  .259   2.91      15  .241    .94
3.83   Shaquille O'Neal  182  .208   2.81      31  .184   1.03

xx      Win Shares:      RS   WS/48   RSx      PO  WS/48   POx
3.81   Karl Malone       235  .205   3.14      23  .140    .67
3.77   Jerry West        163  .213   2.72      27  .203   1.05
3.75   John Stockton     208  .209   3.01      21  .160    .74
3.73   Charles Barkley   177  .216   2.88      20  .193    .85
3.68   Dirk Nowitzki     206  .193   2.77      23  .188    .91

3.67   Kevin Durant      175  .205   2.71      27  .185    .96
3.64   George Mikan      109  .249   2.60      17  .254   1.05
3.54   James Harden      172  .209   2.74      22  .172    .80
3.50   Oscar Robertson   189  .207   2.85      14  .178    .66
3.50   Julius Erving     181  .192   2.59      27  .176    .91

xx      Win Shares:      RS   WS/48   RSx      PO  WS/48   POx
3.40   Bill Russell      164  .193   2.46      28  .178    .94
3.32   Larry Bird        146  .203   2.45      25  .173    .86
3.29   Artis Gilmore     190  .193   2.66      13  .174    .63
3.22   Stephen Curry     141  .197   2.34      22  .190    .88
3.19   Kawhi Leonard     100  .211   2.11      23  .224   1.07

3.15   Hakeem Olajuwon   163  .177   2.25      23  .189    .90
3.13   Kevin Garnett     191  .182   2.52      16  .149    .60
3.12   Reggie Miller     174  .176   2.32      20  .180    .80
3.08   Kobe Bryant       173  .171   2.24      28  .157    .83
3.05   Anthony Davis     119  .212   2.31      11  .226    .74

xx      Win Shares:      RS   WS/48   RSx      PO  WS/48   POx
3.03   Bob Pettit        136  .213   2.48      12  .159    .54
3.01   Moses Malone      179  .174   2.33      15  .176    .68
3.00   Dolph Schayes     142  .192   2.29      14  .189    .71
2.99   Jimmy Butler      121  .205   2.25      17  .179    .74
2.92   Dan Issel         158  .181   2.28      16  .161    .64

2.84   Giannis A         118  .203   2.21      11  .189    .63
2.83   Rudy Gobert       113  .218   2.32       8  .174    .50
2.77   Chauncey Billups  121  .176   1.93      21  .186    .84
2.71   Manu Ginóbili     106  .190   1.96      21  .164    .75
2.71   Kevin McHale      113  .180   1.92      21  .174    .79

xx      Win Shares:      RS   WS/48   RSx      PO  WS/48   POx
2.70   Adrian Dantley    134  .189   2.19       9  .172    .51
2.67   Dwight Howard     142  .172   2.05      14  .167    .62
2.65   Walt Frazier      114  .176   1.88      16  .193    .77
2.63   Pau Gasol         144  .169   2.02      16  .154    .61
2.56   Neil Johnston      92  .241   2.31       2  .159    .24

2.54   Clyde Drexler     136  .173   2.02      16  .134    .53
2.50   Dwyane Wade       121  .162   1.78      22  .155    .72
2.43   Paul Pierce       150  .157   1.92      16  .126    .51
2.42   Ray Allen         145  .150   1.81      18  .143    .61
2.41   Bob Lanier        117  .175   1.89       9  .176    .52

LeBron's postseasons count half as much as his regular seasons. He has been a beast.
Chris Paul with the same RegSea WS as Jordan!

If an NBA-avg player can expect to have 1/10 of his minutes in postseasons, that's a 9:1 ratio of RS to PO.
Taking the square root of each, that becomes a 3:1 ratio in this summation.

Reggie over Kobe. Ginobili over Wade. I'm not a big fan of WinShares.