APBRmetrics

Posted: **Sun Feb 19, 2023 10:28 pm**

Using b-r.com Advanced Stats -- BPM/VORP. Win Shares, and PER -- applied as universally as seems practical; along with some other assumptions and biases, we can rank players' careers thru all eras.

I've made closest fits between PER*Min, VORP and WS to come up with 'wins' created thru PER and BPM, for regular seasons (RS) and for postseasons (PO). Since BPM accounts for strength of competition, I've adjusted per-Wins and WS to total the 'wins' created thru BPM.
Playoff PER-wins are multiplied by 1.118, and WS are * 1.141, to make them all equal.

The list of 'candidates' is drawn from b-r.com -- top 250 Win Shares and top 250 VORP, for combined RS+PO:
https://www.basketball-reference.com/le ... eer_c.html
I also looked up numbers for top 250 Minutes Played. This is partly to include pre-VORP (<1974) players who may beat out some others, in some stat manipulations. The combined lists total 362 players.

Winning the NBA or ABA title is worth a bonus of 240 playoff minutes in this go-round. Minutes * poWinRate is shown as poW+ , the + indicating the Ring bonus.
Today's bias is to weight PER*2, WS*3, VORP*4. This is shown in the first column header.

Players with careers entirely before 1974 are given weight (perW*2 + WS*3)/5
Those who played both before and after 1973/74 are weighted by the % of minutes in each era; for RS and for PO.

The square root of RS 'wins' is multiplied by the win rate (per 484 minutes, where 1.00 is NBA avg); added to the same for PO; multiplied by 5.571 so that the totals align.
(total wins for 362 players = 34,387)
top 20:

Code: Select all

x234   player         rsW  poW+  o   rsWr   poWr
315   LeBron James    251   64   4   2.26   2.59
294   Michael Jordan  202   50   6   2.38   2.70
283  Wilt Chamberlain 238   37   2   2.41   2.22
257  Kareem A-Jabbar  247   43   6   2.08   2.03
228   Tim Duncan      191   45   5   1.95   2.06

217   Magic Johnson   145   39   5   2.11   2.17
216   Chris Paul      181   22   0   2.14   2.09
215  Shaquille O'Neal 171   38   4   1.98   2.05
212   Jerry West      155   31   1   2.05   2.26
209   David Robinson  159   20   2   2.24   2.06

x234   #11-20         rsW  poW+  o   rsWr   poWr
206   Kevin Durant    153   29   2   2.07   2.11
205   Julius Erving   179   33   3   1.92   1.95
205   Karl Malone     216   27   0   1.91   1.67
201   John Stockton   193   24   0   1.96   1.81
196  Charles Barkley  164   21   0   2.01   2.08

196   Dirk Nowitzki   191   26   1   1.80   2.03
193  Oscar Robertson  182   16   1   2.01   1.93
192   Larry Bird      141   31   3   1.99   1.95
190   Kobe Bryant     174   37   5   1.73   1.84
189  Hakeem Olajuwon  162   27   2   1.78   2.13

Kareem may be getting the shaft here. He played 24% of his RS minutes (20% of PO min) before there is vorp. So 76% of his RS wins are based on his rates after 1973. But his 1971-73 numbers were phenomenal, among the best ever.
We can guess that those rates are inflated by the rapid-expansion era; or we might further refine the process; stay tuned.

After boosting PO rates, only a few at this level are seen to be notably worse in playoffs: Wilt, DRob, Stockton- Malone, Oscar: the usual suspects.
Chris Paul has played 24% more RS minutes than Magic; but 45% less in PO.

Posted: **Sun Feb 19, 2023 11:10 pm**

Wading into it further, the general disclaimer may be that there really isn't one 'objective' way to compare, say, James Harden to Bill Russell.
If they were to go one-on-one at the same age -- or with 4 random teammates -- I could bet on either one.

Code: Select all

x234   21-30          rsW  poW+  o   rsWr   poWr
188   James Harden    145   22   0   2.05   1.97
186   Bill Russell    148   40  11   1.75   1.91
184   Kevin Garnett   190   20   1   1.83   1.76
183   Bob Pettit      137   15   1   2.16   1.94
181   Stephen Curry   121   26   4   1.98   2.10

166   Kawhi Leonard    80   25   2   2.03   2.30
164   Dwyane Wade     129   28   3   1.74   1.83
162   Artis Gilmore   168   14   1   1.72   1.80
162   Nikola Jokić     86    9   0   2.36   2.39
158   Dolph Schayes   116   13   1   1.89   2.18

x234   31-40          rsW  poW+  o   rsWr   poWr
156   Reggie Miller   156   20   0   1.58   1.83
156   Clyde Drexler   135   20   1   1.75   1.69
153   George Mikan     42   16   5   2.45   2.88
151   Giannis          93   14   1   1.96   2.24
149   Moses Malone    160   16   1   1.56   1.78

149   Dan Issel       143   17   1   1.66   1.68
148   Anthony Davis    93    8   1   2.05   2.39
148   Scottie Pippen  129   32   6   1.52   1.64
148   Pau Gasol       139   19   2   1.64   1.70
146   Manu Ginóbili    98   25   4   1.76   1.75

Posted: **Mon Feb 20, 2023 1:32 am**

One fun exercise with these numbers is to see who has had a career of postseason overachievement.
Compared to their regular season win rate, the players with more playoff wins:

Code: Select all

po+W    Player      poMin   rsWr   poWr
7.4   LeBron James  11035   2.26   2.59
4.9  Michael Jordan  7474   2.38   2.70
4.2  Hakeem Olajuwon 5749   1.78   2.13
3.9   Roger Brown    4030   1.38   1.84
3.8   Isiah Thomas   4216   1.31   1.75

3.4   Robert Horry   6823   1.18   1.42
3.3   Walt Frazier   3953   1.61   2.02
3.1   Rajon Rondo    4559   1.15   1.49
3.1  Draymond Green  5080   1.28   1.57
2.8   Dirk Nowitzki  5895   1.80   2.03

po+W    Player      poMin   rsWr   poWr
2.8   Jerry West     6321   2.05   2.26
2.7   Elvin Hayes    4160   1.22   1.53
2.7   Reggie Miller  5308   1.58   1.83
2.7   Derek Fisher   6856    .90   1.08
2.6   Kawhi Leonard  4790   2.03   2.30

2.5   Ben Wallace    4524   1.34   1.61
2.4   Baron Davis    1851   1.28   1.92
2.4   John Havlicek  6860   1.33   1.51
2.4 Chauncey Billups 5321   1.58   1.80
2.4   Bill Russell   7497   1.75   1.91

po+W    Player      poMin   rsWr   poWr
2.2   Horace Grant   6172   1.31   1.49
2.2   Gus Williams   3215   1.39   1.72
2.2  Penny Hardaway  2640   1.34   1.73
2.0   Tim Duncan     9370   1.95   2.06
2.0  Tayshaun Prince 4977   1.02   1.22

2.0   Jimmy Jones    3224   1.45   1.76
2.0   James Worthy   5297   1.32   1.50
2.0   Boris Diaw     3144    .96   1.27
2.0  Scottie Pippen  8105   1.52   1.64
2.0 Richard Hamilton 5039   1.09   1.27

po+W = (poWr - rsWr)*poMin/484

Of course, it's nice when you get most of your playoff minutes during your prime years.
The opposite of this is the Cartwright Effect.

Code: Select all

po+W    Player       poMin   rsWr   poWr
-1.5   Jack Sikma     3558   1.37   1.17
-1.6   Kevin Johnson  3879   1.67   1.47
-1.6   Hedo Türkoğlu  2985   1.13    .88
-1.6  David Robinson  4221   2.24   2.06
-1.6   Kyle Lowry     3692   1.46   1.24

-1.6   Tyrone Corbin  2226   1.06    .70
-1.7   Bob Pettit     3545   2.16   1.94
-1.7   Dennis Rodman  4789   1.26   1.09
-1.7  Stephon Marbury  938   1.30    .43
-1.8   Bailey Howell  2722   1.68   1.37

po+W    Player       poMin   rsWr   poWr
-1.8  Detlef Schrempf 3338   1.40   1.14
-1.9   John Stockton  6398   1.96   1.81
-2.0   Sam Cassell    3539   1.40   1.13
-2.1  Sidney Moncrief 3226   1.64   1.32
-2.5  Bill Cartwright 3496   1.12    .77

-2.9 Wilt Chamberlain 7559   2.41   2.22
-3.0   Gary Payton    5481   1.49   1.23
-3.2   Tony Parker    7758   1.33   1.13
-3.3  Cliff Robinson  3887   1.07    .66
-4.0   Karl Malone    7907   1.91   1.67

The underachievers are ... under-represented among 'top 300+ careers'. Role players tend to do worse in postseasons, while stars tend to have their way.
And there may well be bigger playoff flops, who are not on this list. This guy comes to mind:
https://www.basketball-reference.com/pl ... s_advanced

Posted: **Mon Feb 20, 2023 2:22 am**

Never previously recall hearing of Roger Brown and Jimmy Jones. All ABA and mostly ABA respectively. Or Bailey Howell. It is possible they played in a game I saw or heard (listened to a fair amount of ABA radio in youth) but no memory of them.

Both lists draw a lot of head nods, though the underachiever list has some surprises and disappointments.

Posted: **Wed Feb 22, 2023 6:41 pm**

Any chance you could you could perform the same analysis with more prominent advanced stats like the ones below?

For past generations, there is FiveThirtyEight historical RAPTOR, which goes back to the NBA-ABA merger (i.e., 1976-77 season).

Link: https://t.co/yIznDX6Yj0

In addition, Ben Taylor’s Thinking Basketball also has a historical BPM alternative — starting as far back as 1955.

Link (paywall): https://thinkingbasketball.net/player-seasons/

You could then use WS plus your “adjusted PER” Wins for the pre-1955 years.

———————————————

Furthermore, there is an inherent flaw in using cumulative career “wins” to rank players across eras.

First of all, the league only started playing 82-game seasons in 1967-68, obviously disadvantaging those who had their most productive time mostly before that time.

Even then, some seasons since have still been shortened for all or a selection of teams because of lockouts and various disasters.

Moreover, the league just introduced a 4-round best-of-seven tournament — the longest NBA postseason iteration thus far — in 2003, thereby limiting the total available playoffs “wins” in earlier years.

To resolve the season length issues, I think a more appropriate measure for cross-chronological comparison would be looking at these win values on a per-game basis, that is, the mean “wins” that a player provided each game played to his team(s) for their career.

Posted: **Thu Feb 23, 2023 1:47 pm**

Thanks for these remarks!
There are a lot of places where one can go or not go, depending on the work-to-reward ratio. Seasons shorter than 82 games? That can be corrected without too much difficulty, at least in an estimated way. Still some work to look up the seasons everyone played in, before 1968.

But I'm trying this without a year-by-year analysis, to keep it a manageable project. In earlier rankings, I have done the annual season-length adjustment, the annual Ast/FG adjust, an opponent rebound estimate, annual ABA-to-NBA equivalents.

There are ways to estimate overall league competitiveness, too. In the time of rapid expansion (1968-71), so many teams were added (incl. ABA) that many bench players became starters. Lots of points and rebounds were secured against lesser competition.

For this reason, I don't feel too concerned about (for example) Kareem's best statistical years having no VORP. They're still covered by WS and PER, and not too shabby. However, I could see including a pre-'74 adjustment based on post-'74 ratios (of vorp to the others).

The issue of fewer playoff games in earlier eras has been raised a few times. But 14/30 teams get zero playoff games these days. It used to be only 2/8 (or 2/9) were in that situation. On avg, players in the 60s and today get just about the same % of their minutes in playoffs.

Winning a playoff series, of course leads to more playoff games, which conveys its own reward (in poW). Winning the Finals, however, does not. So for 'rings', I've just added 240 po minutes to player totals, converting to Wins at their career rates.
Sometimes this is too much, other times too little. I may yet adjust further, though it doesn't really change much in the rankings.

Posted: **Thu Feb 23, 2023 2:30 pm**

... the league just introduced a 4-round best-of-seven tournament — the longest NBA postseason iteration thus far — in 2003, thereby limiting the total available playoffs “wins” in earlier years.

Nowadays it is possible to get more playoff minutes in a postseason. But it's more likely a player will get zero. The differences pretty much average out to nil.

People have found it hard to believe, so here's a sample table of NBA games played, regular- and post-season, thru time:

Code: Select all

year   tms   sea     RS     PO     p/r
1952    10    66     330    25    .076
1957     8    72     288    19    .066
1962     9    80     360    29    .081
1967    10    81     405    31    .077
1972    17    82     697    37    .053

1977    22    82     902    51    .057
1982    23    82     943    47    .050
1987    23    82     943    71    .075
1992    27    82    1107    73    .066
1997    29    82    1189    72    .061

year   tms   sea     RS     PO     p/r
2002    29    82    1189    71    .060
2007    30    82    1230    79    .064
2012    30    82    1230    84    .068
2017    30    82    1230    79    .064
2022    30    83    1245    87    .070

The '70s do get the shaft, to be sure. Some years, not quite half of teams made the playoffs. Maybe the league was worried about truly-awful teams getting in? So there were also bad omissions (Suns, twice).
The po/rs ratio would be equal for players and for teams. My 'top 300' spreadsheet's players avg .100, but they include all the elites.
Those who played entirely before 1974 actually have a higher po/rs minutes ratio (.108) than those who played thru the transition or entirely after (both .099).

EDIT: Counting only players who are top-250 in either PER-Wins or Win Shares, the era breakdown is .110 for pre-1974 players, .100 for those spanning '73-74; and .105 post-'74.

APBRmetrics

Another attempt to rank everyone

Another attempt to rank everyone

Re: Another attempt to rank everyone

Re: Another attempt to rank everyone

Re: Another attempt to rank everyone

Re: Another attempt to rank everyone

Re: Another attempt to rank everyone

Re: Another attempt to rank everyone