RealGM

min 500 games

1. Curry +6.53
2. Duncan +6.40
3. Kawhi +6.37
4. Klay +5.69
5. Stockton +5.61
6. Green +5.52
7. O'Neal +5.36
8. Malone +5.21
9. Ginobili +5.17
10. Lebron +5.06
11. Paul +4.99

Embiid +5.71
Tatum +5.65

Top end talent combined with some of the most winningest teams will do it.

Not that anyone is using this as a yardstick but the winningest aspect shows pretty clear when the current GSW team has 3 on the list.

Super simple estimate for APM over large samples I’ve found to work pretty well:

(On rating (per 100) + On/Off rating)/5

Divise by 5 to put it on roughly the same scale as rapm/bpm.

NbaAllDay wrote:Top end talent combined with some of the most winningest teams will do it.

Not that anyone is using this as a yardstick but the winningest aspect shows pretty clear when the current GSW team has 3 on the list.

It could be used only for teams leaders.

Klay w/o Curry 107 games -1.19
Green w/o Curry 102 games -3.21

Stockton without Malone? Malone without Stockton? (I'm assuming the Malone on that list is Karl rather than Moses)

So I like doing analyses like these two. Raw data is obviously not the entire picture, but it's easier to process and know where you need to add nuance.

The OP is doing Regular Season only, which is cool, but with bkref it's about as easy to do RS & PS combined, so with that in mind, here are the guys who rise to the top when I do a query of the Top 100 all-season +/- guys going back to '96-97, and then I divide total +/- by games:

1. Steph Curry +6.55
2. Kawhi Leonard +6.11
3. Tim Duncan +6.09
4. Jeff Hornacek +5.86
5. Joel Embiid +5.81
6. Draymond Green +5.72
7. David Robinson +5.63
8. Klay Thompson +5.41
9. Jayson Tatum +5.22
10. John Stockton +5.09

Other MVPs in that Top 100:

11. Shaquille O'Neal +5.06
12. LeBron James +5.04

14. Michael Jordan +5.00
15. Karl Malone +4.65
16. Kevin Durant +4.53

21. Nikola Jokic +3.95
22. Dirk Nowitzki +3.93
23. Steve Nash +3.93

25. James Harden +3.73

32. Kevin Garnett +3.41

36. Giannis Antetokounmpo +3.09

39. Kobe Bryant +3.01

51. Russell Westbrook +2.65

Who do you guys think would be the all-time leaders in +/- per game?

SpreeS wrote:

NbaAllDay wrote:Top end talent combined with some of the most winningest teams will do it.

Not that anyone is using this as a yardstick but the winningest aspect shows pretty clear when the current GSW team has 3 on the list.

It could be used only for teams leaders.

Klay w/o Curry 107 games -1.19
Green w/o Curry 102 games -3.21

Hum? How about Curry without Green?

dygaction wrote:

SpreeS wrote:

NbaAllDay wrote:Top end talent combined with some of the most winningest teams will do it.

Not that anyone is using this as a yardstick but the winningest aspect shows pretty clear when the current GSW team has 3 on the list.

It could be used only for teams leaders.

Klay w/o Curry 107 games -1.19
Green w/o Curry 102 games -3.21

Hum? How about Curry without Green?

Career 238 games +0.80

But this number dont tell whole story. Green was a role player first two years (18min 4.6pts 4.2reb 1.3ast .465ts) on good team and Curry was on a bad team, also he was very far from who is now. If we ignore Curry first two season and Green's first two season as role player and tanking season 2020, we have these numbers:

Green w/o Curry 57 games -2.81
Curry w/o Green 83 games +4.90

SpreeS wrote:

dygaction wrote:

SpreeS wrote:
It could be used only for teams leaders.

Klay w/o Curry 107 games -1.19
Green w/o Curry 102 games -3.21

Hum? How about Curry without Green?

Career 238 games +0.80

But this number dont tell whole story. Green was a role player first two years (18min 4.6pts 4.2reb 1.3ast .465ts) on good team and Curry was on a bad team, also he was very far from who is now. If we ignore Curry first two season and Green's first two season as role player and tanking season 2020, we have these numbers:

Green w/o Curry 57 games -2.81
Curry w/o Green 83 games +4.90

Good stuff. Were there any correlation with Klay in those 57 and 83 games? Thought Green might have played 2019 all season with both Curry and Klay out.

dygaction wrote:

SpreeS wrote:

dygaction wrote:
Hum? How about Curry without Green?

Career 238 games +0.80

But this number dont tell whole story. Green was a role player first two years (18min 4.6pts 4.2reb 1.3ast .465ts) on good team and Curry was on a bad team, also he was very far from who is now. If we ignore Curry first two season and Green's first two season as role player and tanking season 2020, we have these numbers:

Green w/o Curry 57 games -2.81
Curry w/o Green 83 games +4.90

Good stuff. Were there any correlation with Klay in those 57 and 83 games? Thought Green might have played 2019 all season with both Curry and Klay out.

This Green season (2019/2020) is ignored.

Correlation with Klay is tricky again. Curry played 26 games with Klay as rookie role player who started games from the bench in 2012 season. Also he played 23 games with Klay who was on min restriction after comeback of 2.5y absence. So Klay was very far from impacful player as we know him in 49 games with Curry.

I'd be interested to know the % of games in career where a players in game +/- was higher than the margin of victory or loss in that game

So I guess % of games where they have a positive on/off.

I think removing margin of victory will give a better picture

zimpy27 wrote:I'd be interested to know the % of games in career where a players in game +/- was higher than the margin of victory or loss in that game

So I guess % of games where they have a positive on/off.

I think removing margin of victory will give a better picture

% of games in career where a players in game positive +/- was higher than the margin of victory or loss in that game

Curry .392

SpreeS wrote:min 500 games

1. Curry +6.53
2. Duncan +6.40
3. Kawhi +6.37
4. Klay +5.69
5. Stockton +5.61
6. Green +5.52
7. O'Neal +5.36
8. Malone +5.21
9. Ginobili +5.17
10. Lebron +5.06
11. Paul +4.99

Embiid +5.71
Tatum +5.65

You should add the total minutes played next to the player I think. Remember, averages go down the longer you play

after 7 years in cleveland stint #1 and these last few years as a laker, lebron is still that high without being part of a dynasty like basically everyone else in the top 10? damn.

Klay w/o Curry 107 games -1.19
Green w/o Curry 102 games -3.21

i didn't realize klay was so much better than draymond.

f4p wrote:after 7 years in cleveland stint #1 and these last few years as a laker, lebron is still that high without being part of a dynasty like basically everyone else in the top 10? damn.

Klay w/o Curry 107 games -1.19
Green w/o Curry 102 games -3.21

i didn't realize klay was so much better than draymond.

Klay isnt better than Green. Since when one stat of 15% career played games tells you who is better???

Was Paul a part of dynasty?
Did Malone/Stockton have dynasty? Didn't know it.
Curry's first 5 years and 2.5y w/o Klay also hard to call a dynasty
O'Neal apart 5y (00-04) with Kobe hasn't dynasty too

f4p wrote:i didn't realize klay was so much better than draymond.

That isn't how this works.

Most of Draymond's minutes without Curry were also without Klay in 2020 and at the end of 2022.

ceiling raiser wrote:Who do you guys think would be the all-time leaders in +/- per game?

So, was thinking on this, and did a super-simple analysis we can use for a starting point.

What I've done is take historically noteworthy run just looking at the team's margin of victory in seasons of a player's career. This is all-season, because I find that to be a bit more interesting, but I could list out RS or PS if request.

Obvious limitation up front: Not focusing on how much the player in question actually played. You could do a better estimate based on this, but it would take more work.

Without further ado:

Russell-era Boston Celtics ('56-57 to '68-69): +6.45 (Note that Russell missed about half of that first season)
Duncan-era San Antonio Spurs +6.15
Magic-era Los Angeles Lakers ('79-80 to '90-91): +6.08 (Note that I'm not including his mid-90s comeback)
Mikan-era Minneapolis Lakers ('48-49 to '5454): +5.89 (Note that pre-NBA years and his later comeback are not included)
Jordan-era Chicago Bulls ('84-85 to '92-93 & '95-96 to '97-98): +5.00 (Note that he missed much of '85-86, I'm not including '94-95 or the Wiz years).
Curry-era Golden State Warriors ('09-10 to present): +3.52 (Note he missed the vast majority of time in '19-20)
LeBron over his entire career ('93-94 to present): +3.30

Unless I'm forgetting someone then, looks to me like the Russell's teams had the greatest domination in NBA history and thus he's a strong contender for this "+/- Per Game" championship belt.



The big question then is whether someone with a big On/Off, and a low enough MPG to make the difference.

If we just look at guys for whom we have this data for:

Duncan: Personal PM/G: +6.09, Team PM/G +6.15
Curry: Personal PM/G: +6.55, Team PM/G +3.52
LeBron: Personal PM/G +5.04, Team PM/G +3.30

Note that this isn't saying that Duncan's team literally had a better rate when he went to the bench, it's saying the team was slightly above 0 when he was out, which is a reasonable thing for a superstar on a champion.

When we consider how this looks, it seems quite plausible that Magic or Mikan or Jordan might come out on top over Russell.

Of course, I can't help but highlight Curry here. The idea that people are still thinking of Curry as a guy with more help than everyone else in history is just plain galling. Flat out, it's entirely possible that he would take the title of best PM per Game despite the fact that his team's cume PM/G really isn't that strong.

Let me end by mentioning WNBA data. In WNBA history, there are two players who have dominated the league way ahead of everyone else, Cynthia Cooper & Maya Moore.

Here's how they look by these measures:

Cooper-era Houston Comets (1997-2000): Personal PM/G: +9.84, Team PM/G +10.14 (Note that I'm not including her later comeback)
Moore-era Minnesota Lynx (2011-2018): Personal PM/G: +7.87, Team PM/G +6.66





Remember that the WNBA plays 40 minute games, so multiply these numbers by 1.2 if you want to try to apples-to-apples that. Also, there are several WNBA superstars with Personal rates in the +4.5 to +5.0 range, which when multiplied by 1.2 would put them with any of the Personal PM/G of anyone we've got data for other than Curry, Kawhi & Duncan.

What this is telling me is that Cooper & Moore were more dominant - by this sort of metric - in the WNBA than probably anyone in NBA history was in the NBA. This is something you'd expect to be possible in leagues with less talent of course, which makes us ask about how deep into the pre-NBA world you need to go to before you see teams this dominant.

I feel like Mikan in the NBL & BAA years might be there.

Earlier than that, we run into multiple issues with the most general being that we didn't have a top pro league that had all the best talent in basketball. The great teams before that made most of their money as barnstormers (Original Celtics, New York Renaissance, Harlem Globetrotters), and when you're barnstorming the average competition is much weaker.

But there's also the matter of there being much lower scoring so for these teams to have any chance in this competition we'd probably have to go with a more "times divide" approach rather than "plus minus", as in judging by ratio rather than net difference.

Doctor MJ wrote:

ceiling raiser wrote:Who do you guys think would be the all-time leaders in +/- per game?

So, was thinking on this, and did a super-simple analysis we can use for a starting point.

What I've done is take historically noteworthy run just looking at the team's margin of victory in seasons of a player's career. This is all-season, because I find that to be a bit more interesting, but I could list out RS or PS if request.

Obvious limitation up front: Not focusing on how much the player in question actually played. You could do a better estimate based on this, but it would take more work.

Without further ado:

Russell-era Boston Celtics ('56-57 to '68-69): +6.45 (Note that Russell missed about half of that first season)
Duncan-era San Antonio Spurs +6.15
Magic-era Los Angeles Lakers ('79-80 to '90-91): +6.08 (Note that I'm not including his mid-90s comeback)
Mikan-era Minneapolis Lakers ('48-49 to '5454): +5.89 (Note that pre-NBA years and his later comeback are not included)
Jordan-era Chicago Bulls ('84-85 to '92-93 & '95-96 to '97-98): +5.00 (Note that he missed much of '85-86, I'm not including '94-95 or the Wiz years).
Curry-era Golden State Warriors ('09-10 to present): +3.52 (Note he missed the vast majority of time in '19-20)
LeBron over his entire career ('93-94 to present): +3.30

Unless I'm forgetting someone then, looks to me like the Russell's teams had the greatest domination in NBA history and thus he's a strong contender for this "+/- Per Game" championship belt.



The big question then is whether someone with a big On/Off, and a low enough MPG to make the difference.

If we just look at guys for whom we have this data for:

Duncan: Personal PM/G: +6.09, Team PM/G +6.15
Curry: Personal PM/G: +6.55, Team PM/G +3.52
LeBron: Personal PM/G +5.04, Team PM/G +3.30

Note that this isn't saying that Duncan's team literally had a better rate when he went to the bench, it's saying the team was slightly above 0 when he was out, which is a reasonable thing for a superstar on a champion.

When we consider how this looks, it seems quite plausible that Magic or Mikan or Jordan might come out on top over Russell.

Of course, I can't help but highlight Curry here. The idea that people are still thinking of Curry as a guy with more help than everyone else in history is just plain galling. Flat out, it's entirely possible that he would take the title of best PM per Game despite the fact that his team's cume PM/G really isn't that strong.

Let me end by mentioning WNBA data. In WNBA history, there are two players who have dominated the league way ahead of everyone else, Cynthia Cooper & Maya Moore.

Here's how they look by these measures:

Cooper-era Houston Comets (1997-2000): Personal PM/G: +9.84, Team PM/G +10.14 (Note that I'm not including her later comeback)
Moore-era Minnesota Lynx (2011-2018): Personal PM/G: +7.87, Team PM/G +6.66





Remember that the WNBA plays 40 minute games, so multiply these numbers by 1.2 if you want to try to apples-to-apples that. Also, there are several WNBA superstars with Personal rates in the +4.5 to +5.0 range, which when multiplied by 1.2 would put them with any of the Personal PM/G of anyone we've got data for other than Curry, Kawhi & Duncan.

What this is telling me is that Cooper & Moore were more dominant - by this sort of metric - in the WNBA than probably anyone in NBA history was in the NBA. This is something you'd expect to be possible in leagues with less talent of course, which makes us ask about how deep into the pre-NBA world you need to go to before you see teams this dominant.

I feel like Mikan in the NBL & BAA years might be there.

Earlier than that, we run into multiple issues with the most general being that we didn't have a top pro league that had all the best talent in basketball. The great teams before that made most of their money as barnstormers (Original Celtics, New York Renaissance, Harlem Globetrotters), and when you're barnstorming the average competition is much weaker.

But there's also the matter of there being much lower scoring so for these teams to have any chance in this competition we'd probably have to go with a more "times divide" approach rather than "plus minus", as in judging by ratio rather than net difference.

Good post doc but i have to question your specific samplings used

Like, you are using all of lebron career from 18 to 38, yeats where he had long injuries and played through them included (19, 15, 21) but for jordan you use just his bulls years, excluse his injury/inactivity related down year

You are comparing a lebron literal 20 years average to a jordan best 11 years average and a curry 12 year average. The result would be way different if you used lebron from 2009-2020 (12 years) or even 2006-2018 (13 years) and more adequate based on ages as all 3 players would be their 22-35~ periods

falcolombardi wrote:Good post doc but i have to question your specific samplings used

Like, you are using all of lebron career from 18 to 38, yeats where he had long injuries and played through them included (19, 15, 21) but for jordan you use just his bulls years, excluse his injury/inactivity related down year

You are comparing a lebron literal 20 years average to a jordan best 11 years average and a curry 12 year average. The result would be way different if you used lebron from 2009-2020 (12 years) and more adequate based on ages as all 3 players would be their 22-35~ periods

As I said, I kept it as simple as I could.

I'll note that I DID include '85-86 for Jordan. But if I include '94-95 and the Washington years, here's that updated list:

Russell's teams: +6.45
Duncan's teams: +6.15
Magic's teams: +6.08 (still excludes comeback)
Mikan's teams: +5.89 (still excludes comeback)
Jordan's teams: +4.24
Curry's teams: +3.52
LeBron's teams: +3.30

Re: different durations. Sure. The topic is a Per Game thing, so longevity is something that the analyst has to consider for themselves.

Aside from that though, to be clear, I don't think a guy's team not showing up that well here actually makes him look worse. As I said, Curry's 3-point gap between himself and his team seems like a major feather in his cap.