When would you generally say LeBron surpassed MJ all-time?

[twyzted]

Why would anyone ever have Jordan #1 after those Wizards seasons.

Do any of you guys actually listen to yourselves, or do you just kind-of say whatever pops into your head and immediately move on.

And you say that after watching Lebrons last 2 and third of this season?

Not including this season Lebron has averaged an IA 28.1 pts per 75 possessions on rTS% of 4.9% and has created about 12 shots per 100 possessions for teammates.

He has pretty consistently ranked in the top 10 range in one-number metrics.

If you think that is bad, then I suggest you don't check Jordan's Washington numbers.

Right but that is just stat padding on a bad team.

But thank you for introducing me to pts per 75 possesions.

Out of top 30 for regular season
Jordan has 10 of them= 33%
Lebron has 4 of them= 13%

And 10 of 16 best for playoffs.
Lbj has 1 of them.

Its not even close with the wizards years included.

[SHAQ32]

Nevuary

[AEnigma]

This isn't making any sense?

I present a point saying

1) It shouldn't be a GIVEN that the older bigs were larger outliers relative to era, based on box-score impact metrics and WOWYR regressions, and then you bring up Magic and Robinson?

I was not making a WOWYR argument. If I were, yes, I do think it reflects poorly on Jordan’s era relative “dominance” that two contemporaries fare better. Lebron really just has Curry and I guess Nash at the beginning of his prime. Russell just has Oscar. Kareem just has uhhhh Havlicek at the beginning and Magic at the end?

Based on this, thinking, Kareem has no argument over Russell in all-time ranks because Russell pretty consistently beats him in the WOWYR.

I mean if we are just talking prime WOWYR then sure? But we do not just throw out longevity like that, and in an era relative argument Kareem does not have competitors the way Russell at least has Oscar.

The point is there is good evidence that the bigs aren't automatically ahead of Jordan in terms of raw impact as it was presented.

Ignoring the era relative aspects, how is WOWYR raw impact.

And you don't want me to quote more modern impact numbers when comparing MJ to Magic/Robinson. Because then it will really get nasty in here....
Especially if we are comparing MJ to Robinson in terms of PS play. We don't want that argument.

But in our data for postseason play, Robinson actually grades out as more impactful than any Jordan stretch.

Like I am legitimately curious where this sudden confidence in “modern impact numbers” comes in for one player who does not appear in anything else (oh, sorry, cannot forget partial RAPM samples from his maybe fifth to seventh best regular seasons…) and another who was a league-dominating impact giant at his peak. Is this going to be another PIPM/BPM situation?

I won't get into the box-score metrics like BPM estimates for the older guys, Backpicks BPM, or Player Impact, that suggest, MJ is also ahead of the group; it has already been brought up that people questioning Jordan's impact more or less expect MJ to come out looking stronger in that stuff.

WOWYR is among the more reliable/objective impact signals we have for determining impact.

In what sense. How does it adjust for league SRSs? How does it adjust for variations within a prime?

Looking at some guy and saying he has a rival in impact doesn't dispute the idea, that said individual has more separation from the general pack than the others; it is possible that there can be more outliers (as the league strength and amount of players increases), and said outliers have more separation from the impact.

A.) You have not bothered to show that.

B.) If you had paid any attention to the discussion before jumping in with the WOWYR Bible, you would know that specifically era relative comparisons was the subject there, yes.

And the Robinson comment of him having more raw impact is not quite the hill to die on unless, you believe Net On/Off is all that matters.

I never said that, and if you had any memory for this sort of thing you would have figured out that I was not sincerely celebrating Robinson’s postseason here. That said, I do find it funny how not only did you just charge on through with more box score stuff (even after acknowledging it was never really in dispute or taken as overly meaningful), you also changed the frame of “look at the WOWYR ratings” to “look at how they did in the postseason”, which of course brings us right back to the usual elevators.

[VanWest82]

You have it the other way around

It would be a longevity argument if he had the most total rapm but not the best averages

He has BOTH,meaning that he is the best at rapm while playing most minutes among all his peers.

If you reduced his minutes to those of his rivals at the stat (shaq, curry, garnett, duncan) by taking off his first and last years his average would be even higher than theirs than it already is

Longevity is somethingh that punishes your averages the most non prime but still star level seasons you play

Having the most longevity and still having the best average and the most and best #1 stretches is incredibly impressive and makes him the undisputed rapm goat by any measure you want to look at (highest peak, highest career total, highest prime, best in playoffs)

The fact that you're suggesting we look at 10 year RAPM stretches already makes it a longevity argument given that most great players aren't great for nearly that length of time. This is even the case for a lot of guys with 5 year samples.

There are issues with the numbers you posted. Looking at NBA shot charts, Chris Paul had the best 5 year RAPM in both the fall 2011 -spring 2016 and fall 2012 - spring 2017 time periods. The author claims Lebron's five year stretch in that time is the GOAT five year stretch. Something's off.

Perhaps this discrepency is due to the author using full season data, so your argument is apples to oranges given you replied to my assertion that Lebron was mailing in regular seasons from his last year in Miami through first year in LA.

[falcolombardi]

You have it the other way around

It would be a longevity argument if he had the most total rapm but not the best averages

He has BOTH,meaning that he is the best at rapm while playing most minutes among all his peers.

If you reduced his minutes to those of his rivals at the stat (shaq, curry, garnett, duncan) by taking off his first and last years his average would be even higher than theirs than it already is

Longevity is somethingh that punishes your averages the most non prime but still star level seasons you play

Having the most longevity and still having the best average and the most and best #1 stretches is incredibly impressive and makes him the undisputed rapm goat by any measure you want to look at (highest peak, highest career total, highest prime, best in playoffs)

The fact that you're suggesting we look at 10 year RAPM stretches already makes it a longevity argument given that most great players aren't great for nearly that length of time. This is even the case for a lot of guys with 5 year samples.

There are issues with the numbers you posted. Looking at NBA shot charts, Chris Paul had the best 5 year RAPM in both the fall 2011 -spring 2016 and fall 2012 - spring 2017 time periods. The author claims Lebron's five year stretch in that time is the GOAT five year stretch. Something's off.

Perhaps this discrepency is due to the author using full season data, so your argument is apples to oranges given you replied to my assertion that Lebron was mailing in regular seasons from his last year in Miami through first year in LA.

No, again he has the highest average too. He is keeping a higher average rapm on top of playing more

[VanWest82]

No, again he has the highest average too. He is keeping a higher average rapm on top of playing more

Right. Average over long spans of time.

But again, we're talking specifically about that 14-19 regular season stretch, or at least that's what I've been trying to talk about, and the very large discrepancy in Bron's RS vs. PS performance. Bringing a whole bunch of season long (RS+PS) 5 year RAPM stretches into the discussion is just distracting from the point.

Edit: I guess if what you're trying to argue is even with him mailing in lengthy stretches in those seasons, he's still the best once you include playoffs and therefore he was either a) justified or b) who cares because season long data confirms his greatness, then sure.

My argument is that there's a whole other level of greatness for showing up and playing hard for all 82 that Lebron stopped achieving post 2013. And that other great players have seemingly copied this formula, and that it has hurt the product in much the same way as team stacking and load management (also things we can accuse Lebron for having a hand in popularizing). This is perhaps a very subjective issue to bring into the discussion, but I just have a really hard time squaring the idea that a player who tried to game the system that much is actually the GOAT.

[AEnigma]

There are issues with the numbers you posted. Looking at NBA shot charts, Chris Paul had the best 5 year RAPM in both the fall 2011 -spring 2016 and fall 2012 - spring 2017 time periods. The author claims Lebron's five year stretch in that time is the GOAT five year stretch. Something's off.

Perhaps this discrepency is due to the author using full season data, so your argument is apples to oranges given you replied to my assertion that Lebron was mailing in regular seasons from his last year in Miami through first year in LA.

Before going further, RAPM is not some rigidly set formula. In this case, you are almost certainly correct that incorporating playoff data gives Lebron a substantial boost in a way not true of any other player. Different regressions will produce different values, although disparities probably should not be too drastic. Even just looking at NBA Shot Charts, it gives you a “luck-adjustment” option! And if use that suddenly Lebron has the #1 sample in 2013-16 and 2014-17 even though you were saying those are his coasting years.

But with that acknowledgment made, and even without getting into lineup data and the significance of bench rotations and “loads” and possession samples and all of that… why do you see “Lebron occasionally had lower RAPM stretches than Chris Paul and/or Steph Curry” as a meaningful argument? Was Chris Paul just “trying harder”? What is the point here.

And why do you just assume something similar would not occur with Jordan and Magic? As LukaTheGOAT has very kindly and relevantly helped highlight, regressions from that era favour Magic over Jordan. Lineup data is going to prefer Magic there, with Jordan’s teams failing to break the 4-SRS mark until 1991. By any measure he was also more strictly important to his team than Jordan was. Was he also just trying harder? By your eyetest I doubt you would agree, so again, what are we actually doing here?

If Magic turns out to top Jordan in three-year samples, do you care? Does that change anything? You are trying to work backward from a presupposition here. You know Lebron coasted, and that needs to matter. But the reality is there is no objective measure of “effort relative to impact”. You are again conflating an aesthetic preference for a real reflection of a player’s value.

To your credit, you at least seem moderately aware of that, but I think it is important to call it what it is.

[OhayoKD]

^ I think if anything you are dismissing those “trailblazers” by dismissing Russell and Wilt and Kareem despite claiming to care so much about era relative dominance.

Which is time and time again the refrain when it comes to Jordan. Oh, Jordan was better relative to his era than Lebron was his own era, but also, we cannot do the same thing for the guys who were even bigger outliers because reasons.

I would greatly disagree that Russ, Wilt, and Kareem were larger outliers relative to era then MJ and I imagine that would be the case for a lot of people when evaluating them.

In terms of With or Without regressions, I would say MJ comes out looking strong, and the outlier of the group.

Yeah, this is misleading. Setting aside WOWYR's questionable utility (more on that later), you don't seem to be accounting for the fact that srs=/championship probability, and that the threshold for dominance in the 60's was vastly lower than the threshold required for dominance in the 60's. Going off the person who created the WOWRY metrics you're using here:

At the height of their dynasty, the Celtics were comically dominant. From 1962-65, their average margin-of-victory (MOV) was over 8 points per game. During the same time span, only two other teams even eclipsed 4 points per game – the ’64 Royals and the ’64 Warriors. And all of Boston’s separation was created by its historic defense, anchored by Russell:.

When we tie WOWYR to championships as opposed to "who has the best M.O.V", Russell(and to a lesser extent, Wilt), not Jordan is the outlier of the group you just referenced.

Boston was a 35-win team (-1.9 SRS) in 28 games he missed from 1958-69, and for the other 915 games of his career they played at a 59-win pace (6.4 SRS). This is a tiny piece of evidence – the years are spread out, teams change, and so on — but it echoes the same story as Russell’s other value signals.

WOWYR isn't saying "Russell is less valuable than Jordan", WOWYR is saying that "Russell needed far less help to win 11 Rings than Jordan needed to win 6". We can also extrapolate, based on Wilt mostly matching Russell in these metrics, that Wilt grades as a bigger contributor to championship probability than Jordan managed.

TLDR: WOWYR says Russell, and Wilt, are much bigger outliers than Jordan.

Of course, I don't think we should be putting this much stock into WOWYR, and fwiw, WOWYR's inventor seems to agree. Looking at both the Kareem and Russell "impact evaluations", that quoted portion is the only time WOWYR numbers are listed(they are linked but not stated specifically regarding 1970 Kareem) and its attached with the caveat of "tiny sample". Instead Ben focuses on unregularized data from stretches where stars miss significant time. To quote myself:

Over 10 year spans or so, wowy will typically be based on a handful of games per season with players who've probably changed signifcantly during that time span. That's not really all that useful. What you want to do is look for moments where players or teammates miss(or are absent) from an unusally high number of games so you can get the largest samples. A good starting point for this would be the year before a player joins a team or the year after a leaves a team as you get a full season sample of data(70 Celtics, 84 Bulls, 84 Rockets, 69 Bucks, ect.) Then you can track roster changes, and granular stuff to adjust or guesstimate if the team improved got worse, ect. Lookign for concentrated stretches of missed game time, or how teams do when a star's teammates go out also can be useful. Ideally you want as much of this type of data for a player in various contexts and then you can compare players in these various situations directly.

Even with RAPM, a more polished variant of what you're citing for Russell, Ben will defer to unregularized samples when dealing with outlier impact:

Like Nash, LeBron was supercharging dependent talent — finishers who disproportionately benefited from shots served to them on a silver platter. So with his talents in South Beach, Cleveland crumbled in 2011. While most teams fall off after losing a superstar, none imploded like the Lebron-less Cavs; in 21 games with a similar group of players, they played at an anemic 18-win pace (-8.9 SRS) before injuries ravaged their lineup. LeBron’s not worth 40 wins on a typical club, but no player in history has correlated more strongly with such massive, worst-to-first impact.

This is because regularization, a process that is used in generating RAPM and WOWYR sets artificial caps on player value:

In layman terms, the process of calculating RAPM involves some math that distorts the "real scale" of the numbers in favor of accuracy.

When impact exceeds this scale, value is misattributed to role players.

Luckily, we don't need to rely on a smattering of minutes to assess Russell, because we actually have an 82 game sample of "without" for Russell in 1970. The celtics, despite seeing their offense improve(second best player sees his volume skyrocket), dropped to 35 wins with Russell retiring. The previous year, Bill beat a superteam to win a championship as a player-coach on his last legs. With more help in 1990(The Bulls were a 27 win team before they drafted Jordan, incrementally improved after drafting Pippen and Grant, then saw their srs skyrocket with the introduction of the triangle) Apex Jordan managed 55 wins and a close loss in the Conference Finals.

Russell's success was also substantially less tied to his teammates. While Jordan's success correlates with Pippen as strongly as Steph's correlates with Draymond, the Celtics were able to win, win, and keep winning regardless of who left and who stayed.

Both WOWY and WOWYR suggest Russell actually had less help than Jordan. Considering that the Celtics were vastly more dominant than the Bulls, I don't think there's much of a case here for Jordan as an era-relative outlier.

I also find it questionable to use "prime-wowyr" with Kareem and Jordan when Kareem's prime was vastly longer than MJ's(averages tend to dip the more you play). When we focus on their best years with the largest(and most inclusive) samples. "raw" impact also favors Kareem:

[

About WOWY - Jordan's biggest samples don't show him as the better one than Kareem (from Ben's database):

1986 Jordan: +2.0 SRS change, 1.2 WOWY score
1995 Jordan +2.7 SRS change, 1.9 WOWY score

1975 Kareem: +7.1 SRS change, +3.6 WOWY score

I'm afraid Ben's database has an error with 1978 sample, as it shows as clear negative for Kareem, despite all the calculations I made and his own words in Kareem profile:

At the beginning of the ’78 season, Kareem cold-cocked Bucks center Kent Benson and missed substantial time with another broken hand. However, it’s hard to infer much from the injury since LA fired off two trades around that period.10 With Jabbar — and ignoring all the other lineup activity — the Lakers played like a 53-win team (4.1 SRS) in ’78. With a similar roster in ’79 (minus Charlie Scott), LA ticked along at a 50-win clip when healthy (3.1 SRS). Below, I’ve plotted the ’78 team’s performance in 21 games without Kareem, in which the Lakers played at a 36-win pace (-1.7 SRS) after a major offensive drop-off.

Which shows a +5.8 SRS change again. The biggest samples we have show Kareem having a clear advantage. We can also look at the more nuanced samples, when a player even joins or leaves his team:

1984 Bulls without MJ: -4.7 SRS, 27 wins
1985 Bulls with MJ: -0.5 SRS, 38 wins
Change: +4.2 SRS and +11 wins

1993 Bulls with MJ: +6.2 SRS, 57 wins
1994 Bulls without MJ: +2.9 SRS, 55 wins
Change: +3.3 SRS, +2 wins

1995 Bulls without MJ: +3.8 SRS, 52 wins pace
1996 Bulls with MJ: +11.8 SRS, 72 wins
Change: +8 SRS, 20 wins

I wouldn't include 1998-1999, because the whole team changed, including a coach.

1969 Bucks without Kareem: -5.1 SRS, 27 wins
1970 Bucks with Kareem: +4.3 SRS, 56 wins
Change: +9.4 SRS, 29 wins

1975 Bucks with Kareem: +2.6 SRS, 49 wins pace
1976 Bucks without Kareem: -1.6 SRS, 38 wins
Change: +4.2 SRS, 11 wins

1975 Lakers without Kareem: -3.9 SRS, 30 wins
1976 Lakers with Kareem: +0.2 SRS, 40 wins
Change: +4.1 SRS, 10 wins

The difference is that Kareem left Bucks in a trade, which means that Lakers gave a lot of value to Bucks. Jordan samples are clean, as Jordan didn't go to the Bulls in exchange.

I don't know, I don't see the case for MJ > Kareem in terms of WOWY.

As for RAPTOR:

Overall, however, RAPTOR weights the “box” component more highly than the “on-off” component. In testing RAPTOR on out-of-sample data, we found that while on-court/off-court stats provide useful information, they’re nonetheless quite noisy as compared with individual measures of player value that are used in the “box” part of RAPTOR.

We would expect it to rate guards over two-way bigs, especially considering it does not have plus-minus data pre-1997. As with most box-heavy things, RAPTOR elevates Jordan to competitiveness with Lebron, though even here, at least for a 1-year peak, Jordan doesn't quite reach the summit:

By RAPTOR (Since 77)

09 Lebron: 12.6

91 MJ: 12.3

Assuming we think individual defense is relevant, this sort of metric probably shouldn't play too big of a role in comparisons between guards and two-way bigs.

Indeed, as one might expect, things flip quickly when box-stuff is toned down:

For posterity going to list the best looking "real" apm(sample size is not even) for GOAT candidates(and players who grade similarly) and y'all can work off that I guess. (range is based on best two scoring seasons)
+9/+8: Lebron
+8/+7: Duncan, D-Rob
+7/+6: KG, MJ, Curry

Idk quite how you translate apm to wins added but I'll also post unregularized wowy(as outlier value gets mis-distributed via apm and wowyr):

(Caveat, while Russell doesn't score so high here based on wins, high championship probaility required a much lower srs in his era and Russell should probably be seen as the "most likely to win" individual relative to era, at least for his prime). Same idea where range goes off two best signals

40/30 win lift: Lebron, Hakeem, KG
30/20 win lift: Kareem, Curry, Bird, D-rob, Giannis
20/10 win lift: Jordan, Shaq, Russell

To his credit, Jordan still looks like one of the best players ever. Unprecedented outlier? No. And he's doesn't come particularly close.

[ceoofkobefans]

2016 is prolly the year I’d start to heavily consider LeBron > MJ by 2018 I’m taking LeBron more often than not by 2020 I don’t really see the argument against him

[LukaTheGOAT]

^ I think if anything you are dismissing those “trailblazers” by dismissing Russell and Wilt and Kareem despite claiming to care so much about era relative dominance.

Which is time and time again the refrain when it comes to Jordan. Oh, Jordan was better relative to his era than Lebron was his own era, but also, we cannot do the same thing for the guys who were even bigger outliers because reasons.

I would greatly disagree that Russ, Wilt, and Kareem were larger outliers relative to era then MJ and I imagine that would be the case for a lot of people when evaluating them.

In terms of With or Without regressions, I would say MJ comes out looking strong, and the outlier of the group.

Yeah, this is misleading. Setting aside WOWYR's questionable utility (more on that later), you don't seem to be accounting for the fact that srs=/championship probability, and that the threshold for dominance in the 60's was vastly lower than the threshold required for dominance in the 60's. Going off the person who created the WOWRY metrics you're using here:

At the height of their dynasty, the Celtics were comically dominant. From 1962-65, their average margin-of-victory (MOV) was over 8 points per game. During the same time span, only two other teams even eclipsed 4 points per game – the ’64 Royals and the ’64 Warriors. And all of Boston’s separation was created by its historic defense, anchored by Russell:.

When we tie WOWYR to championships as opposed to "who has the best M.O.V", Russell(and to a lesser extent, Wilt), not Jordan is the outlier of the group you just referenced.

Boston was a 35-win team (-1.9 SRS) in 28 games he missed from 1958-69, and for the other 915 games of his career they played at a 59-win pace (6.4 SRS). This is a tiny piece of evidence – the years are spread out, teams change, and so on — but it echoes the same story as Russell’s other value signals.

WOWYR isn't saying "Russell is less valuable than Jordan", WOWYR is saying that "Russell needed far less help to win 11 Rings than Jordan needed to win 6". We can also extrapolate, based on Wilt mostly matching Russell in these metrics, that Wilt grades as a bigger contributor to championship probability than Jordan managed.

TLDR: WOWYR says Russell, and Wilt, are much bigger outliers than Jordan.

Of course, I don't think we should be putting this much stock into WOWYR, and fwiw, WOWYR's inventor seems to agree. Looking at both the Kareem and Russell "impact evaluations", that quoted portion is the only time WOWYR numbers are listed(they are linked but not stated specifically regarding 1970 Kareem) and its attached with the caveat of "tiny sample". Instead Ben focuses on unregularized data from stretches where stars miss significant time. To quote myself:

Over 10 year spans or so, wowy will typically be based on a handful of games per season with players who've probably changed signifcantly during that time span. That's not really all that useful. What you want to do is look for moments where players or teammates miss(or are absent) from an unusally high number of games so you can get the largest samples. A good starting point for this would be the year before a player joins a team or the year after a leaves a team as you get a full season sample of data(70 Celtics, 84 Bulls, 84 Rockets, 69 Bucks, ect.) Then you can track roster changes, and granular stuff to adjust or guesstimate if the team improved got worse, ect. Lookign for concentrated stretches of missed game time, or how teams do when a star's teammates go out also can be useful. Ideally you want as much of this type of data for a player in various contexts and then you can compare players in these various situations directly.

Even with RAPM, a more polished variant of what you're citing for Russell, Ben will defer to unregularized samples when dealing with outlier impact:

Like Nash, LeBron was supercharging dependent talent — finishers who disproportionately benefited from shots served to them on a silver platter. So with his talents in South Beach, Cleveland crumbled in 2011. While most teams fall off after losing a superstar, none imploded like the Lebron-less Cavs; in 21 games with a similar group of players, they played at an anemic 18-win pace (-8.9 SRS) before injuries ravaged their lineup. LeBron’s not worth 40 wins on a typical club, but no player in history has correlated more strongly with such massive, worst-to-first impact.

This is because regularization, a process that is used in generating RAPM and WOWYR sets artificial caps on player value:

In layman terms, the process of calculating RAPM involves some math that distorts the "real scale" of the numbers in favor of accuracy.

When impact exceeds this scale, value is misattributed to role players.

Luckily, we don't need to rely on a smattering of minutes to assess Russell, because we actually have an 82 game sample of "without" for Russell in 1970. The celtics, despite seeing their offense improve(second best player sees his volume skyrocket), dropped to 35 wins with Russell retiring. The previous year, Bill beat a superteam to win a championship as a player-coach on his last legs. With more help in 1990(The Bulls were a 27 win team before they drafted Jordan, incrementally improved after drafting Pippen and Grant, then saw their srs skyrocket with the introduction of the triangle) Apex Jordan managed 55 wins and a close loss in the Conference Finals.

Russell's success was also substantially less tied to his teammates. While Jordan's success correlates with Pippen as strongly as Steph's correlates with Draymond, the Celtics were able to win, win, and keep winning regardless of who left and who stayed.

Both WOWY and WOWYR suggest Russell actually had less help than Jordan. Considering that the Celtics were vastly more dominant than the Bulls, I don't think there's much of a case here for Jordan as an era-relative outlier.

I also find it questionable to use "prime-wowyr" with Kareem and Jordan when Kareem's prime was vastly longer than MJ's(averages tend to dip the more you play). When we focus on their best years with the largest(and most inclusive) samples. "raw" impact also favors Kareem:

[

About WOWY - Jordan's biggest samples don't show him as the better one than Kareem (from Ben's database):

1986 Jordan: +2.0 SRS change, 1.2 WOWY score
1995 Jordan +2.7 SRS change, 1.9 WOWY score

1975 Kareem: +7.1 SRS change, +3.6 WOWY score

I'm afraid Ben's database has an error with 1978 sample, as it shows as clear negative for Kareem, despite all the calculations I made and his own words in Kareem profile:

At the beginning of the ’78 season, Kareem cold-cocked Bucks center Kent Benson and missed substantial time with another broken hand. However, it’s hard to infer much from the injury since LA fired off two trades around that period.10 With Jabbar — and ignoring all the other lineup activity — the Lakers played like a 53-win team (4.1 SRS) in ’78. With a similar roster in ’79 (minus Charlie Scott), LA ticked along at a 50-win clip when healthy (3.1 SRS). Below, I’ve plotted the ’78 team’s performance in 21 games without Kareem, in which the Lakers played at a 36-win pace (-1.7 SRS) after a major offensive drop-off.

Which shows a +5.8 SRS change again. The biggest samples we have show Kareem having a clear advantage. We can also look at the more nuanced samples, when a player even joins or leaves his team:

1984 Bulls without MJ: -4.7 SRS, 27 wins
1985 Bulls with MJ: -0.5 SRS, 38 wins
Change: +4.2 SRS and +11 wins

1993 Bulls with MJ: +6.2 SRS, 57 wins
1994 Bulls without MJ: +2.9 SRS, 55 wins
Change: +3.3 SRS, +2 wins

1995 Bulls without MJ: +3.8 SRS, 52 wins pace
1996 Bulls with MJ: +11.8 SRS, 72 wins
Change: +8 SRS, 20 wins

I wouldn't include 1998-1999, because the whole team changed, including a coach.

1969 Bucks without Kareem: -5.1 SRS, 27 wins
1970 Bucks with Kareem: +4.3 SRS, 56 wins
Change: +9.4 SRS, 29 wins

1975 Bucks with Kareem: +2.6 SRS, 49 wins pace
1976 Bucks without Kareem: -1.6 SRS, 38 wins
Change: +4.2 SRS, 11 wins

1975 Lakers without Kareem: -3.9 SRS, 30 wins
1976 Lakers with Kareem: +0.2 SRS, 40 wins
Change: +4.1 SRS, 10 wins

The difference is that Kareem left Bucks in a trade, which means that Lakers gave a lot of value to Bucks. Jordan samples are clean, as Jordan didn't go to the Bulls in exchange.

I don't know, I don't see the case for MJ > Kareem in terms of WOWY.

As for RAPTOR:

Overall, however, RAPTOR weights the “box” component more highly than the “on-off” component. In testing RAPTOR on out-of-sample data, we found that while on-court/off-court stats provide useful information, they’re nonetheless quite noisy as compared with individual measures of player value that are used in the “box” part of RAPTOR.

We would expect it to rate guards over two-way bigs, especially considering it does not have plus-minus data pre-1997. As with most box-heavy things, RAPTOR elevates Jordan to competitiveness with Lebron, though even here, at least for a 1-year peak, Jordan doesn't quite reach the summit:

By RAPTOR (Since 77)

09 Lebron: 12.6

91 MJ: 12.3

Assuming we think individual defense is relevant, this sort of metric probably shouldn't play too big of a role in comparisons between guards and two-way bigs.

Indeed, as one might expect, things flip quickly when box-stuff is toned down:

For posterity going to list the best looking "real" apm(sample size is not even) for GOAT candidates(and players who grade similarly) and y'all can work off that I guess. (range is based on best two scoring seasons)
+9/+8: Lebron
+8/+7: Duncan, D-Rob
+7/+6: KG, MJ, Curry

Idk quite how you translate apm to wins added but I'll also post unregularized wowy(as outlier value gets mis-distributed via apm and wowyr):

(Caveat, while Russell doesn't score so high here based on wins, high championship probaility required a much lower srs in his era and Russell should probably be seen as the "most likely to win" individual relative to era, at least for his prime). Same idea where range goes off two best signals

40/30 win lift: Lebron, Hakeem, KG
30/20 win lift: Kareem, Curry, Bird, D-rob, Giannis
20/10 win lift: Jordan, Shaq, Russell

To his credit, Jordan still looks like one of the best players ever. Unprecedented outlier? No. And he's doesn't come particularly close.

There is so much wrong with this post, but perhaps the biggest thing is using APM scores like this when we don't have full-season data for MJ's peak years.

Also I have twice in the past asked you to link me to some of the sources you are using and you never have, so I would imagine if I ask for the you are wrongly using to suggest Robinson was comparable to MJ...you likely will not share again. They do say third times the charm so maybe this time you will actually share?

[falcolombardi]

I would greatly disagree that Russ, Wilt, and Kareem were larger outliers relative to era then MJ and I imagine that would be the case for a lot of people when evaluating them.

In terms of With or Without regressions, I would say MJ comes out looking strong, and the outlier of the group.

Yeah, this is misleading. Setting aside WOWYR's questionable utility (more on that later), you don't seem to be accounting for the fact that srs=/championship probability, and that the threshold for dominance in the 60's was vastly lower than the threshold required for dominance in the 60's. Going off the person who created the WOWRY metrics you're using here:

At the height of their dynasty, the Celtics were comically dominant. From 1962-65, their average margin-of-victory (MOV) was over 8 points per game. During the same time span, only two other teams even eclipsed 4 points per game – the ’64 Royals and the ’64 Warriors. And all of Boston’s separation was created by its historic defense, anchored by Russell:.

When we tie WOWYR to championships as opposed to "who has the best M.O.V", Russell(and to a lesser extent, Wilt), not Jordan is the outlier of the group you just referenced.

Boston was a 35-win team (-1.9 SRS) in 28 games he missed from 1958-69, and for the other 915 games of his career they played at a 59-win pace (6.4 SRS). This is a tiny piece of evidence – the years are spread out, teams change, and so on — but it echoes the same story as Russell’s other value signals.

WOWYR isn't saying "Russell is less valuable than Jordan", WOWYR is saying that "Russell needed far less help to win 11 Rings than Jordan needed to win 6". We can also extrapolate, based on Wilt mostly matching Russell in these metrics, that Wilt grades as a bigger contributor to championship probability than Jordan managed.

TLDR: WOWYR says Russell, and Wilt, are much bigger outliers than Jordan.

Of course, I don't think we should be putting this much stock into WOWYR, and fwiw, WOWYR's inventor seems to agree. Looking at both the Kareem and Russell "impact evaluations", that quoted portion is the only time WOWYR numbers are listed(they are linked but not stated specifically regarding 1970 Kareem) and its attached with the caveat of "tiny sample". Instead Ben focuses on unregularized data from stretches where stars miss significant time. To quote myself:

Over 10 year spans or so, wowy will typically be based on a handful of games per season with players who've probably changed signifcantly during that time span. That's not really all that useful. What you want to do is look for moments where players or teammates miss(or are absent) from an unusally high number of games so you can get the largest samples. A good starting point for this would be the year before a player joins a team or the year after a leaves a team as you get a full season sample of data(70 Celtics, 84 Bulls, 84 Rockets, 69 Bucks, ect.) Then you can track roster changes, and granular stuff to adjust or guesstimate if the team improved got worse, ect. Lookign for concentrated stretches of missed game time, or how teams do when a star's teammates go out also can be useful. Ideally you want as much of this type of data for a player in various contexts and then you can compare players in these various situations directly.

Even with RAPM, a more polished variant of what you're citing for Russell, Ben will defer to unregularized samples when dealing with outlier impact:

Like Nash, LeBron was supercharging dependent talent — finishers who disproportionately benefited from shots served to them on a silver platter. So with his talents in South Beach, Cleveland crumbled in 2011. While most teams fall off after losing a superstar, none imploded like the Lebron-less Cavs; in 21 games with a similar group of players, they played at an anemic 18-win pace (-8.9 SRS) before injuries ravaged their lineup. LeBron’s not worth 40 wins on a typical club, but no player in history has correlated more strongly with such massive, worst-to-first impact.

This is because regularization, a process that is used in generating RAPM and WOWYR sets artificial caps on player value:

In layman terms, the process of calculating RAPM involves some math that distorts the "real scale" of the numbers in favor of accuracy.

When impact exceeds this scale, value is misattributed to role players.

Luckily, we don't need to rely on a smattering of minutes to assess Russell, because we actually have an 82 game sample of "without" for Russell in 1970. The celtics, despite seeing their offense improve(second best player sees his volume skyrocket), dropped to 35 wins with Russell retiring. The previous year, Bill beat a superteam to win a championship as a player-coach on his last legs. With more help in 1990(The Bulls were a 27 win team before they drafted Jordan, incrementally improved after drafting Pippen and Grant, then saw their srs skyrocket with the introduction of the triangle) Apex Jordan managed 55 wins and a close loss in the Conference Finals.

Russell's success was also substantially less tied to his teammates. While Jordan's success correlates with Pippen as strongly as Steph's correlates with Draymond, the Celtics were able to win, win, and keep winning regardless of who left and who stayed.

Both WOWY and WOWYR suggest Russell actually had less help than Jordan. Considering that the Celtics were vastly more dominant than the Bulls, I don't think there's much of a case here for Jordan as an era-relative outlier.

I also find it questionable to use "prime-wowyr" with Kareem and Jordan when Kareem's prime was vastly longer than MJ's(averages tend to dip the more you play). When we focus on their best years with the largest(and most inclusive) samples. "raw" impact also favors Kareem:

[

About WOWY - Jordan's biggest samples don't show him as the better one than Kareem (from Ben's database):

1986 Jordan: +2.0 SRS change, 1.2 WOWY score
1995 Jordan +2.7 SRS change, 1.9 WOWY score

1975 Kareem: +7.1 SRS change, +3.6 WOWY score

I'm afraid Ben's database has an error with 1978 sample, as it shows as clear negative for Kareem, despite all the calculations I made and his own words in Kareem profile:



Which shows a +5.8 SRS change again. The biggest samples we have show Kareem having a clear advantage. We can also look at the more nuanced samples, when a player even joins or leaves his team:

1984 Bulls without MJ: -4.7 SRS, 27 wins
1985 Bulls with MJ: -0.5 SRS, 38 wins
Change: +4.2 SRS and +11 wins

1993 Bulls with MJ: +6.2 SRS, 57 wins
1994 Bulls without MJ: +2.9 SRS, 55 wins
Change: +3.3 SRS, +2 wins

1995 Bulls without MJ: +3.8 SRS, 52 wins pace
1996 Bulls with MJ: +11.8 SRS, 72 wins
Change: +8 SRS, 20 wins

I wouldn't include 1998-1999, because the whole team changed, including a coach.

1969 Bucks without Kareem: -5.1 SRS, 27 wins
1970 Bucks with Kareem: +4.3 SRS, 56 wins
Change: +9.4 SRS, 29 wins

1975 Bucks with Kareem: +2.6 SRS, 49 wins pace
1976 Bucks without Kareem: -1.6 SRS, 38 wins
Change: +4.2 SRS, 11 wins

1975 Lakers without Kareem: -3.9 SRS, 30 wins
1976 Lakers with Kareem: +0.2 SRS, 40 wins
Change: +4.1 SRS, 10 wins

The difference is that Kareem left Bucks in a trade, which means that Lakers gave a lot of value to Bucks. Jordan samples are clean, as Jordan didn't go to the Bulls in exchange.

I don't know, I don't see the case for MJ > Kareem in terms of WOWY.

As for RAPTOR:

Overall, however, RAPTOR weights the “box” component more highly than the “on-off” component. In testing RAPTOR on out-of-sample data, we found that while on-court/off-court stats provide useful information, they’re nonetheless quite noisy as compared with individual measures of player value that are used in the “box” part of RAPTOR.

We would expect it to rate guards over two-way bigs, especially considering it does not have plus-minus data pre-1997. As with most box-heavy things, RAPTOR elevates Jordan to competitiveness with Lebron, though even here, at least for a 1-year peak, Jordan doesn't quite reach the summit:

By RAPTOR (Since 77)

09 Lebron: 12.6

91 MJ: 12.3

Assuming we think individual defense is relevant, this sort of metric probably shouldn't play too big of a role in comparisons between guards and two-way bigs.

Indeed, as one might expect, things flip quickly when box-stuff is toned down:

For posterity going to list the best looking "real" apm(sample size is not even) for GOAT candidates(and players who grade similarly) and y'all can work off that I guess. (range is based on best two scoring seasons)
+9/+8: Lebron
+8/+7: Duncan, D-Rob
+7/+6: KG, MJ, Curry

Idk quite how you translate apm to wins added but I'll also post unregularized wowy(as outlier value gets mis-distributed via apm and wowyr):

(Caveat, while Russell doesn't score so high here based on wins, high championship probaility required a much lower srs in his era and Russell should probably be seen as the "most likely to win" individual relative to era, at least for his prime). Same idea where range goes off two best signals

40/30 win lift: Lebron, Hakeem, KG
30/20 win lift: Kareem, Curry, Bird, D-rob, Giannis
20/10 win lift: Jordan, Shaq, Russell

To his credit, Jordan still looks like one of the best players ever. Unprecedented outlier? No. And he's doesn't come particularly close.

There is so much wrong with this post, but perhaps the biggest thing is using APM scores like this when we don't have full-season data for MJ's peak years.

Also I have twice in the past asked you to link me to some of the sources you are using and you never have, so I would imagine if I ask for the you are wrongly using to suggest Robinson was comparable to MJ...you likely will not share again. They do say third times the charm so maybe this time you will actually share?

You guys are starting to get a bit too chippy here. Lets just relax here and discuss with more chill

[LukaTheGOAT]

Yeah, this is misleading. Setting aside WOWYR's questionable utility (more on that later), you don't seem to be accounting for the fact that srs=/championship probability, and that the threshold for dominance in the 60's was vastly lower than the threshold required for dominance in the 60's. Going off the person who created the WOWRY metrics you're using here:

When we tie WOWYR to championships as opposed to "who has the best M.O.V", Russell(and to a lesser extent, Wilt), not Jordan is the outlier of the group you just referenced.


WOWYR isn't saying "Russell is less valuable than Jordan", WOWYR is saying that "Russell needed far less help to win 11 Rings than Jordan needed to win 6". We can also extrapolate, based on Wilt mostly matching Russell in these metrics, that Wilt grades as a bigger contributor to championship probability than Jordan managed.

TLDR: WOWYR says Russell, and Wilt, are much bigger outliers than Jordan.

Of course, I don't think we should be putting this much stock into WOWYR, and fwiw, WOWYR's inventor seems to agree. Looking at both the Kareem and Russell "impact evaluations", that quoted portion is the only time WOWYR numbers are listed(they are linked but not stated specifically regarding 1970 Kareem) and its attached with the caveat of "tiny sample". Instead Ben focuses on unregularized data from stretches where stars miss significant time. To quote myself:

Even with RAPM, a more polished variant of what you're citing for Russell, Ben will defer to unregularized samples when dealing with outlier impact:

This is because regularization, a process that is used in generating RAPM and WOWYR sets artificial caps on player value:

When impact exceeds this scale, value is misattributed to role players.

Luckily, we don't need to rely on a smattering of minutes to assess Russell, because we actually have an 82 game sample of "without" for Russell in 1970. The celtics, despite seeing their offense improve(second best player sees his volume skyrocket), dropped to 35 wins with Russell retiring. The previous year, Bill beat a superteam to win a championship as a player-coach on his last legs. With more help in 1990(The Bulls were a 27 win team before they drafted Jordan, incrementally improved after drafting Pippen and Grant, then saw their srs skyrocket with the introduction of the triangle) Apex Jordan managed 55 wins and a close loss in the Conference Finals.

Russell's success was also substantially less tied to his teammates. While Jordan's success correlates with Pippen as strongly as Steph's correlates with Draymond, the Celtics were able to win, win, and keep winning regardless of who left and who stayed.

Both WOWY and WOWYR suggest Russell actually had less help than Jordan. Considering that the Celtics were vastly more dominant than the Bulls, I don't think there's much of a case here for Jordan as an era-relative outlier.

I also find it questionable to use "prime-wowyr" with Kareem and Jordan when Kareem's prime was vastly longer than MJ's(averages tend to dip the more you play). When we focus on their best years with the largest(and most inclusive) samples. "raw" impact also favors Kareem:


As for RAPTOR:

We would expect it to rate guards over two-way bigs, especially considering it does not have plus-minus data pre-1997. As with most box-heavy things, RAPTOR elevates Jordan to competitiveness with Lebron, though even here, at least for a 1-year peak, Jordan doesn't quite reach the summit:

Assuming we think individual defense is relevant, this sort of metric probably shouldn't play too big of a role in comparisons between guards and two-way bigs.

Indeed, as one might expect, things flip quickly when box-stuff is toned down:

To his credit, Jordan still looks like one of the best players ever. Unprecedented outlier? No. And he's doesn't come particularly close.

There is so much wrong with this post, but perhaps the biggest thing is using APM scores like this when we don't have full-season data for MJ's peak years.

Also I have twice in the past asked you to link me to some of the sources you are using and you never have, so I would imagine if I ask for the you are wrongly using to suggest Robinson was comparable to MJ...you likely will not share again. They do say third times the charm so maybe this time you will actually share?

You guys are starting to get a bit too chippy here. Lets just relax here and discuss with more chill

I'll also play cop here, and I will continue to ask for that APM data that makes Robinson clear of Jordan in relative to era impact. He's not sliding out of this one that easily again....

Now OhayoKD, please come to the podium please and come answer for your potential crimes please.

[OhayoKD]

I would greatly disagree that Russ, Wilt, and Kareem were larger outliers relative to era then MJ and I imagine that would be the case for a lot of people when evaluating them.

In terms of With or Without regressions, I would say MJ comes out looking strong, and the outlier of the group.

Yeah, this is misleading. Setting aside WOWYR's questionable utility (more on that later), you don't seem to be accounting for the fact that srs=/championship probability, and that the threshold for dominance in the 60's was vastly lower than the threshold required for dominance in the 60's. Going off the person who created the WOWRY metrics you're using here:

At the height of their dynasty, the Celtics were comically dominant. From 1962-65, their average margin-of-victory (MOV) was over 8 points per game. During the same time span, only two other teams even eclipsed 4 points per game – the ’64 Royals and the ’64 Warriors. And all of Boston’s separation was created by its historic defense, anchored by Russell:.

When we tie WOWYR to championships as opposed to "who has the best M.O.V", Russell(and to a lesser extent, Wilt), not Jordan is the outlier of the group you just referenced.

Boston was a 35-win team (-1.9 SRS) in 28 games he missed from 1958-69, and for the other 915 games of his career they played at a 59-win pace (6.4 SRS). This is a tiny piece of evidence – the years are spread out, teams change, and so on — but it echoes the same story as Russell’s other value signals.

WOWYR isn't saying "Russell is less valuable than Jordan", WOWYR is saying that "Russell needed far less help to win 11 Rings than Jordan needed to win 6". We can also extrapolate, based on Wilt mostly matching Russell in these metrics, that Wilt grades as a bigger contributor to championship probability than Jordan managed.

TLDR: WOWYR says Russell, and Wilt, are much bigger outliers than Jordan.

Of course, I don't think we should be putting this much stock into WOWYR, and fwiw, WOWYR's inventor seems to agree. Looking at both the Kareem and Russell "impact evaluations", that quoted portion is the only time WOWYR numbers are listed(they are linked but not stated specifically regarding 1970 Kareem) and its attached with the caveat of "tiny sample". Instead Ben focuses on unregularized data from stretches where stars miss significant time. To quote myself:

Over 10 year spans or so, wowy will typically be based on a handful of games per season with players who've probably changed signifcantly during that time span. That's not really all that useful. What you want to do is look for moments where players or teammates miss(or are absent) from an unusally high number of games so you can get the largest samples. A good starting point for this would be the year before a player joins a team or the year after a leaves a team as you get a full season sample of data(70 Celtics, 84 Bulls, 84 Rockets, 69 Bucks, ect.) Then you can track roster changes, and granular stuff to adjust or guesstimate if the team improved got worse, ect. Lookign for concentrated stretches of missed game time, or how teams do when a star's teammates go out also can be useful. Ideally you want as much of this type of data for a player in various contexts and then you can compare players in these various situations directly.

Even with RAPM, a more polished variant of what you're citing for Russell, Ben will defer to unregularized samples when dealing with outlier impact:

Like Nash, LeBron was supercharging dependent talent — finishers who disproportionately benefited from shots served to them on a silver platter. So with his talents in South Beach, Cleveland crumbled in 2011. While most teams fall off after losing a superstar, none imploded like the Lebron-less Cavs; in 21 games with a similar group of players, they played at an anemic 18-win pace (-8.9 SRS) before injuries ravaged their lineup. LeBron’s not worth 40 wins on a typical club, but no player in history has correlated more strongly with such massive, worst-to-first impact.

This is because regularization, a process that is used in generating RAPM and WOWYR sets artificial caps on player value:

In layman terms, the process of calculating RAPM involves some math that distorts the "real scale" of the numbers in favor of accuracy.

When impact exceeds this scale, value is misattributed to role players.

Luckily, we don't need to rely on a smattering of minutes to assess Russell, because we actually have an 82 game sample of "without" for Russell in 1970. The celtics, despite seeing their offense improve(second best player sees his volume skyrocket), dropped to 35 wins with Russell retiring. The previous year, Bill beat a superteam to win a championship as a player-coach on his last legs. With more help in 1990(The Bulls were a 27 win team before they drafted Jordan, incrementally improved after drafting Pippen and Grant, then saw their srs skyrocket with the introduction of the triangle) Apex Jordan managed 55 wins and a close loss in the Conference Finals.

Russell's success was also substantially less tied to his teammates. While Jordan's success correlates with Pippen as strongly as Steph's correlates with Draymond, the Celtics were able to win, win, and keep winning regardless of who left and who stayed.

Both WOWY and WOWYR suggest Russell actually had less help than Jordan. Considering that the Celtics were vastly more dominant than the Bulls, I don't think there's much of a case here for Jordan as an era-relative outlier.

I also find it questionable to use "prime-wowyr" with Kareem and Jordan when Kareem's prime was vastly longer than MJ's(averages tend to dip the more you play). When we focus on their best years with the largest(and most inclusive) samples. "raw" impact also favors Kareem:

[

About WOWY - Jordan's biggest samples don't show him as the better one than Kareem (from Ben's database):

1986 Jordan: +2.0 SRS change, 1.2 WOWY score
1995 Jordan +2.7 SRS change, 1.9 WOWY score

1975 Kareem: +7.1 SRS change, +3.6 WOWY score

I'm afraid Ben's database has an error with 1978 sample, as it shows as clear negative for Kareem, despite all the calculations I made and his own words in Kareem profile:



Which shows a +5.8 SRS change again. The biggest samples we have show Kareem having a clear advantage. We can also look at the more nuanced samples, when a player even joins or leaves his team:

1984 Bulls without MJ: -4.7 SRS, 27 wins
1985 Bulls with MJ: -0.5 SRS, 38 wins
Change: +4.2 SRS and +11 wins

1993 Bulls with MJ: +6.2 SRS, 57 wins
1994 Bulls without MJ: +2.9 SRS, 55 wins
Change: +3.3 SRS, +2 wins

1995 Bulls without MJ: +3.8 SRS, 52 wins pace
1996 Bulls with MJ: +11.8 SRS, 72 wins
Change: +8 SRS, 20 wins

I wouldn't include 1998-1999, because the whole team changed, including a coach.

1969 Bucks without Kareem: -5.1 SRS, 27 wins
1970 Bucks with Kareem: +4.3 SRS, 56 wins
Change: +9.4 SRS, 29 wins

1975 Bucks with Kareem: +2.6 SRS, 49 wins pace
1976 Bucks without Kareem: -1.6 SRS, 38 wins
Change: +4.2 SRS, 11 wins

1975 Lakers without Kareem: -3.9 SRS, 30 wins
1976 Lakers with Kareem: +0.2 SRS, 40 wins
Change: +4.1 SRS, 10 wins

The difference is that Kareem left Bucks in a trade, which means that Lakers gave a lot of value to Bucks. Jordan samples are clean, as Jordan didn't go to the Bulls in exchange.

I don't know, I don't see the case for MJ > Kareem in terms of WOWY.

As for RAPTOR:

Overall, however, RAPTOR weights the “box” component more highly than the “on-off” component. In testing RAPTOR on out-of-sample data, we found that while on-court/off-court stats provide useful information, they’re nonetheless quite noisy as compared with individual measures of player value that are used in the “box” part of RAPTOR.

We would expect it to rate guards over two-way bigs, especially considering it does not have plus-minus data pre-1997. As with most box-heavy things, RAPTOR elevates Jordan to competitiveness with Lebron, though even here, at least for a 1-year peak, Jordan doesn't quite reach the summit:

By RAPTOR (Since 77)

09 Lebron: 12.6

91 MJ: 12.3

Assuming we think individual defense is relevant, this sort of metric probably shouldn't play too big of a role in comparisons between guards and two-way bigs.

Indeed, as one might expect, things flip quickly when box-stuff is toned down:

For posterity going to list the best looking "real" apm(sample size is not even) for GOAT candidates(and players who grade similarly) and y'all can work off that I guess. (range is based on best two scoring seasons)
+9/+8: Lebron
+8/+7: Duncan, D-Rob
+7/+6: KG, MJ, Curry

Idk quite how you translate apm to wins added but I'll also post unregularized wowy(as outlier value gets mis-distributed via apm and wowyr):

(Caveat, while Russell doesn't score so high here based on wins, high championship probaility required a much lower srs in his era and Russell should probably be seen as the "most likely to win" individual relative to era, at least for his prime). Same idea where range goes off two best signals

40/30 win lift: Lebron, Hakeem, KG
30/20 win lift: Kareem, Curry, Bird, D-rob, Giannis
20/10 win lift: Jordan, Shaq, Russell

To his credit, Jordan still looks like one of the best players ever. Unprecedented outlier? No. And he's doesn't come particularly close.

There is so much wrong with this post, but perhaps the biggest thing is using APM scores like this when we don't have full-season data for MJ's peak years.

I literally put a disclaimer the data is partial after various "full samples" though it's somewhat odd to be complaining about sample size when you were trying to argue against Russell's data based on 28 missed games over a 13 year span:

WOWYR (With or Without You, Regressed) is a game-level plus-minus measurement and a cousin of play-by-play measurements like Adjusted Plus-Minus (APM). Instead of using results from lineups within a game (play-by-play data) like traditional APM, game-level plus-minus uses final scores from game to game for the players from that game.

"prime WOWYR" has bigger sample-size issues that are exacerbated with the number of years covered(players change).

Also I have twice in the past asked you to link me to some of the sources you are using and you never have

I literally linked you a source the first time you asked:

[

All the 5 year rapm i've seen has favored duncan, for example:
https://www.reddit.com/r/nbadiscussion/comments/q50ucn/5_year_rapm_peaks_1997_to_2021_regular_season/
Also owly and squared circle's had duncan higher irrc.

Unfortunately I can't find owly(i think)'s RAPM dump. but here is some of squared-circle's(full set may have been posted in the peaks project):
https://forums.realgm.com/boards/viewtopic.php?f=64&t=2206246
And a google-doc from another realgmer whose name I can't remember:
https://docs.google.com/spreadsheets/d/e/2PACX-1vQdG8Zv84zqKEzETDjd8KPsClcw9bPETX9v_x_KEAxjv9NrFaWikOoiSaciy1jbMiygg2D-V8DUQn0O/pubhtml?gid=112475182&single=true

, so I would imagine if I ask for the you are wrongly using to suggest Robinson was comparable to MJ..

I never "wrongly suggested" anything. He's listed there for posterity, are you confusing me with enig?
Here's the full context of the last thing I quoted:

I'll come at it along two axis.

Skill/talent/whatever you want to call it and fit.

Many levels to skill, but rough tiering here with GOAT/MVP/AllNBA/Allstar, then further down the line. Simplifying fit into Good/Okay/Poor.

Levels of duos and GOAT candidate (not including duos of GOAT candidates, I don't think one has ever happened):

Good fitting MVP duo (eg CP3/Howard)
Okay fitting MVP duo (eg CP3/KD)
GOAT Candidate (eg LeBron)
Poor fitting MVP duo (eg CP3/Nash)
Good fitting All-NBA duo (eg Billups/Gasol)
and so on...

I'd say the GOAT candidate has a lower floor than the 2 below it (eg the Wolves with KG), but a higher ceiling that is reasonably attainable through the addition of an Allstar level player (eg one Sam Cassell), overall I side with the higher upside with the GOAT candidate.

In APM type terms:

GOAT = +7
MVP = +5.5
All-NBA = +4
Allstar = +2.5
Starter = +1

I think you should presume being able to put a +1 guy in the GOAT candidates #2 spot (though the '06/'07 Wolves might have a word), and that would put you in range of the 2x All-NBA squad. A lower end Allstar partner isn't an unreasonable goal and puts you half between the 2x All-NBA/MVP levels.

For posterity going to list the best looking "real apm"(sample size is not even) for GOAT candidates(and players who grade similarly) and y'all can work off that I guess. (range is based on best two scoring seasons)
+9/+8: Lebron
+8/+7: Duncan, D-Rob
+7/+6: KG, MJ, Curry

Idk quite how you translate apm to wins added but I'll also post unregularized wowy(as outlier value gets misdistributed via apm and wowyr):

(Caveat, while Russell doesn't score so high here based on wins, high championship probaility required a much lower srs in his era and Russell should probably be seen as the "most likely to win" individual relative to era, at least for his prime). Same idea where range goes off two best signals

40/30 win lift: Lebron, Hakeem, KG
30/20 win lift: Kareem, Curry, Bird, D-rob, Giannis
20/10 win lift: Jordan, Shaq, Russell

I presume one of you smarter people can derive a +/- from that.

Sort of depends on what you consider a goat candidate, but there's more value-fluctuation for people around that tier than people acknowledge I think

I'm kind of confused though why you're focusing on the APM side of this when my post strongly argued against leaning too heavily into regularized stuff. The RAPM has mostly been an addendum to corroborate pure impact signals.

Regardless, the main focus of my post was Russell. I don't care about david robinson when it comes to the GOAT debate

[LukaTheGOAT]

Yeah, this is misleading. Setting aside WOWYR's questionable utility (more on that later), you don't seem to be accounting for the fact that srs=/championship probability, and that the threshold for dominance in the 60's was vastly lower than the threshold required for dominance in the 60's. Going off the person who created the WOWRY metrics you're using here:

When we tie WOWYR to championships as opposed to "who has the best M.O.V", Russell(and to a lesser extent, Wilt), not Jordan is the outlier of the group you just referenced.


WOWYR isn't saying "Russell is less valuable than Jordan", WOWYR is saying that "Russell needed far less help to win 11 Rings than Jordan needed to win 6". We can also extrapolate, based on Wilt mostly matching Russell in these metrics, that Wilt grades as a bigger contributor to championship probability than Jordan managed.

TLDR: WOWYR says Russell, and Wilt, are much bigger outliers than Jordan.

Of course, I don't think we should be putting this much stock into WOWYR, and fwiw, WOWYR's inventor seems to agree. Looking at both the Kareem and Russell "impact evaluations", that quoted portion is the only time WOWYR numbers are listed(they are linked but not stated specifically regarding 1970 Kareem) and its attached with the caveat of "tiny sample". Instead Ben focuses on unregularized data from stretches where stars miss significant time. To quote myself:

Even with RAPM, a more polished variant of what you're citing for Russell, Ben will defer to unregularized samples when dealing with outlier impact:

This is because regularization, a process that is used in generating RAPM and WOWYR sets artificial caps on player value:

When impact exceeds this scale, value is misattributed to role players.

Luckily, we don't need to rely on a smattering of minutes to assess Russell, because we actually have an 82 game sample of "without" for Russell in 1970. The celtics, despite seeing their offense improve(second best player sees his volume skyrocket), dropped to 35 wins with Russell retiring. The previous year, Bill beat a superteam to win a championship as a player-coach on his last legs. With more help in 1990(The Bulls were a 27 win team before they drafted Jordan, incrementally improved after drafting Pippen and Grant, then saw their srs skyrocket with the introduction of the triangle) Apex Jordan managed 55 wins and a close loss in the Conference Finals.

Russell's success was also substantially less tied to his teammates. While Jordan's success correlates with Pippen as strongly as Steph's correlates with Draymond, the Celtics were able to win, win, and keep winning regardless of who left and who stayed.

Both WOWY and WOWYR suggest Russell actually had less help than Jordan. Considering that the Celtics were vastly more dominant than the Bulls, I don't think there's much of a case here for Jordan as an era-relative outlier.

I also find it questionable to use "prime-wowyr" with Kareem and Jordan when Kareem's prime was vastly longer than MJ's(averages tend to dip the more you play). When we focus on their best years with the largest(and most inclusive) samples. "raw" impact also favors Kareem:


As for RAPTOR:

We would expect it to rate guards over two-way bigs, especially considering it does not have plus-minus data pre-1997. As with most box-heavy things, RAPTOR elevates Jordan to competitiveness with Lebron, though even here, at least for a 1-year peak, Jordan doesn't quite reach the summit:

Assuming we think individual defense is relevant, this sort of metric probably shouldn't play too big of a role in comparisons between guards and two-way bigs.

Indeed, as one might expect, things flip quickly when box-stuff is toned down:

To his credit, Jordan still looks like one of the best players ever. Unprecedented outlier? No. And he's doesn't come particularly close.

There is so much wrong with this post, but perhaps the biggest thing is using APM scores like this when we don't have full-season data for MJ's peak years.

I literally put a disclaimer the data is partial after various "full samples" though it's somewhat odd to be complaining about sample size when you were trying to argue against Russell's data based on 28 missed games over a 13 year span:

WOWYR (With or Without You, Regressed) is a game-level plus-minus measurement and a cousin of play-by-play measurements like Adjusted Plus-Minus (APM). Instead of using results from lineups within a game (play-by-play data) like traditional APM, game-level plus-minus uses final scores from game to game for the players from that game.

"prime WOWYR" has bigger sample-size issues that are exacerbated with the number of years covered(players change).

Also I have twice in the past asked you to link me to some of the sources you are using and you never have

I literally linked you a source the first time you asked:

[

All the 5 year rapm i've seen has favored duncan, for example:
https://www.reddit.com/r/nbadiscussion/comments/q50ucn/5_year_rapm_peaks_1997_to_2021_regular_season/
Also owly and squared circle's had duncan higher irrc.

Unfortunately I can't find owly(i think)'s RAPM dump. but here is some of squared-circle's(full set may have been posted in the peaks project):
https://forums.realgm.com/boards/viewtopic.php?f=64&t=2206246
And a google-doc from another realgmer whose name I can't remember:
https://docs.google.com/spreadsheets/d/e/2PACX-1vQdG8Zv84zqKEzETDjd8KPsClcw9bPETX9v_x_KEAxjv9NrFaWikOoiSaciy1jbMiygg2D-V8DUQn0O/pubhtml?gid=112475182&single=true

, so I would imagine if I ask for the you are wrongly using to suggest Robinson was comparable to MJ..

I never "wrongly suggested" anything. He's listed there for posterity, are you confusing me with enig?
Here's the full context of the last thing I quoted:

I'll come at it along two axis.

Skill/talent/whatever you want to call it and fit.

Many levels to skill, but rough tiering here with GOAT/MVP/AllNBA/Allstar, then further down the line. Simplifying fit into Good/Okay/Poor.

Levels of duos and GOAT candidate (not including duos of GOAT candidates, I don't think one has ever happened):

Good fitting MVP duo (eg CP3/Howard)
Okay fitting MVP duo (eg CP3/KD)
GOAT Candidate (eg LeBron)
Poor fitting MVP duo (eg CP3/Nash)
Good fitting All-NBA duo (eg Billups/Gasol)
and so on...

I'd say the GOAT candidate has a lower floor than the 2 below it (eg the Wolves with KG), but a higher ceiling that is reasonably attainable through the addition of an Allstar level player (eg one Sam Cassell), overall I side with the higher upside with the GOAT candidate.

In APM type terms:

GOAT = +7
MVP = +5.5
All-NBA = +4
Allstar = +2.5
Starter = +1

I think you should presume being able to put a +1 guy in the GOAT candidates #2 spot (though the '06/'07 Wolves might have a word), and that would put you in range of the 2x All-NBA squad. A lower end Allstar partner isn't an unreasonable goal and puts you half between the 2x All-NBA/MVP levels.

For posterity going to list the best looking "real apm"(sample size is not even) for GOAT candidates(and players who grade similarly) and y'all can work off that I guess. (range is based on best two scoring seasons)
+9/+8: Lebron
+8/+7: Duncan, D-Rob
+7/+6: KG, MJ, Curry

Idk quite how you translate apm to wins added but I'll also post unregularized wowy(as outlier value gets misdistributed via apm and wowyr):

(Caveat, while Russell doesn't score so high here based on wins, high championship probaility required a much lower srs in his era and Russell should probably be seen as the "most likely to win" individual relative to era, at least for his prime). Same idea where range goes off two best signals

40/30 win lift: Lebron, Hakeem, KG
30/20 win lift: Kareem, Curry, Bird, D-rob, Giannis
20/10 win lift: Jordan, Shaq, Russell

I presume one of you smarter people can derive a +/- from that.

Sort of depends on what you consider a goat candidate, but there's more value-fluctuation for people around that tier than people acknowledge I think

I'm kind of confused though why you're focusing on the APM side of this when my post strongly argued against leaning too heavily into regularized stuff. The RAPM has mostly been an addendum to corroborate pure impact signals.

Regardless, the main focus of my post was Russell. I don't care about david robinson when it comes to the GOAT debate

So as I stated, there is no full-season data suggesting that Robinson's impact exceeded Jordan (as SquareCircle only includes partial segments of the season), and you once wasted time by not even being willing to admit this. You once again ignore the argument I was making, which is that Robinson's raw impact did not exceed Peak Jordan's in the 90s (what in the world does a 98-2019 sample have to do with anything)?

I didn't bring up Robinson, AEnigma did. And once again, if you couldn't answer my initial thesis that Robinson had more raw impact than MJ, it was absolutely useless trying to start an argument.

Finally as Jaivi as eloquently stated before, you can't compare different RAPM samples done by different creators that way, but you refuse to listen. Thanks for nothing.

[LukaTheGOAT]

There are issues with the numbers you posted. Looking at NBA shot charts, Chris Paul had the best 5 year RAPM in both the fall 2011 -spring 2016 and fall 2012 - spring 2017 time periods. The author claims Lebron's five year stretch in that time is the GOAT five year stretch. Something's off.

Perhaps this discrepency is due to the author using full season data, so your argument is apples to oranges given you replied to my assertion that Lebron was mailing in regular seasons from his last year in Miami through first year in LA.

Before going further, RAPM is not some rigidly set formula. In this case, you are almost certainly correct that incorporating playoff data gives Lebron a substantial boost in a way not true of any other player. Different regressions will produce different values, although disparities probably should not be too drastic. Even just looking at NBA Shot Charts, it gives you a “luck-adjustment” option! And if use that suddenly Lebron has the #1 sample in 2013-16 and 2014-17 even though you were saying those are his coasting years.

But with that acknowledgment made, and even without getting into lineup data and the significance of bench rotations and “loads” and possession samples and all of that… why do you see “Lebron occasionally had lower RAPM stretches than Chris Paul and/or Steph Curry” as a meaningful argument? Was Chris Paul just “trying harder”? What is the point here.

And why do you just assume something similar would not occur with Jordan and Magic? As LukaTheGOAT has very kindly and relevantly helped highlight, regressions from that era favour Magic over Jordan. Lineup data is going to prefer Magic there, with Jordan’s teams failing to break the 4-SRS mark until 1991. By any measure he was also more strictly important to his team than Jordan was. Was he also just trying harder? By your eyetest I doubt you would agree, so again, what are we actually doing here?

If Magic turns out to top Jordan in three-year samples, do you care? Does that change anything? You are trying to work backward from a presupposition here. You know Lebron coasted, and that needs to matter. But the reality is there is no objective measure of “effort relative to impact”. You are again conflating an aesthetic preference for a real reflection of a player’s value.

To your credit, you at least seem moderately aware of that, but I think it is important to call it what it is.

"Relevantly,"....so you admit the WOWYR does have some validity?

Btw, Oscar Robertson and Jerry West, both contemporaries of Russell, rate out as better than him in such regressions. Know you claimed that Russell was indisputably a bigger outlier relative to era than MJ, yet once again we have an example of where a top 4 guy ever, was surpassed in impact per the regressions. Just a counter to the Robinson/Magic being better than MJ in WOWYR, and thus he must automatically be a fraud.

[falcolombardi]

There are issues with the numbers you posted. Looking at NBA shot charts, Chris Paul had the best 5 year RAPM in both the fall 2011 -spring 2016 and fall 2012 - spring 2017 time periods. The author claims Lebron's five year stretch in that time is the GOAT five year stretch. Something's off.

Perhaps this discrepency is due to the author using full season data, so your argument is apples to oranges given you replied to my assertion that Lebron was mailing in regular seasons from his last year in Miami through first year in LA.

Before going further, RAPM is not some rigidly set formula. In this case, you are almost certainly correct that incorporating playoff data gives Lebron a substantial boost in a way not true of any other player. Different regressions will produce different values, although disparities probably should not be too drastic. Even just looking at NBA Shot Charts, it gives you a “luck-adjustment” option! And if use that suddenly Lebron has the #1 sample in 2013-16 and 2014-17 even though you were saying those are his coasting years.

But with that acknowledgment made, and even without getting into lineup data and the significance of bench rotations and “loads” and possession samples and all of that… why do you see “Lebron occasionally had lower RAPM stretches than Chris Paul and/or Steph Curry” as a meaningful argument? Was Chris Paul just “trying harder”? What is the point here.

And why do you just assume something similar would not occur with Jordan and Magic? As LukaTheGOAT has very kindly and relevantly helped highlight, regressions from that era favour Magic over Jordan. Lineup data is going to prefer Magic there, with Jordan’s teams failing to break the 4-SRS mark until 1991. By any measure he was also more strictly important to his team than Jordan was. Was he also just trying harder? By your eyetest I doubt you would agree, so again, what are we actually doing here?

If Magic turns out to top Jordan in three-year samples, do you care? Does that change anything? You are trying to work backward from a presupposition here. You know Lebron coasted, and that needs to matter. But the reality is there is no objective measure of “effort relative to impact”. You are again conflating an aesthetic preference for a real reflection of a player’s value.

To your credit, you at least seem moderately aware of that, but I think it is important to call it what it is.

"Relevantly,"....so you admit the WOWYR does have some validity?

Btw, Oscar Robertson and Jerry West, both contemporaries of Russell, rate out as better than him in such regressions. Know you claimed that Russell was indisputably a bigger outlier relative to era than MJ, yet once again we have an example of where a top 4 guy ever, was surpassed in impact per the regressions. Just a counter to the Robinson/Magic being better than MJ in WOWYR, and thus he must automatically be a fraud.

I dont think anyone argued this, less so with that wording

Is more than wowyr is not a data point that favors jordan over russel (or the other way around necessrrily)

With such a small and volatile sanple size stat like wowyr is better to just use it in a loose sense

Both players looking good in the stat is great, one of then looking bad in it would be worrisome (not the case here for either) so then since both llook great in that stat any jordan vs russel comparision needs to move to different angles and data points than wowyr

Wowyr is not a data point that agrees with jordan goat case, it would do more so for russel goat case if anythingh

[AEnigma]

There are issues with the numbers you posted. Looking at NBA shot charts, Chris Paul had the best 5 year RAPM in both the fall 2011 -spring 2016 and fall 2012 - spring 2017 time periods. The author claims Lebron's five year stretch in that time is the GOAT five year stretch. Something's off.

Perhaps this discrepency is due to the author using full season data, so your argument is apples to oranges given you replied to my assertion that Lebron was mailing in regular seasons from his last year in Miami through first year in LA.

Before going further, RAPM is not some rigidly set formula. In this case, you are almost certainly correct that incorporating playoff data gives Lebron a substantial boost in a way not true of any other player. Different regressions will produce different values, although disparities probably should not be too drastic. Even just looking at NBA Shot Charts, it gives you a “luck-adjustment” option! And if use that suddenly Lebron has the #1 sample in 2013-16 and 2014-17 even though you were saying those are his coasting years.

But with that acknowledgment made, and even without getting into lineup data and the significance of bench rotations and “loads” and possession samples and all of that… why do you see “Lebron occasionally had lower RAPM stretches than Chris Paul and/or Steph Curry” as a meaningful argument? Was Chris Paul just “trying harder”? What is the point here.

And why do you just assume something similar would not occur with Jordan and Magic? As LukaTheGOAT has very kindly and relevantly helped highlight, regressions from that era favour Magic over Jordan. Lineup data is going to prefer Magic there, with Jordan’s teams failing to break the 4-SRS mark until 1991. By any measure he was also more strictly important to his team than Jordan was. Was he also just trying harder? By your eyetest I doubt you would agree, so again, what are we actually doing here?

If Magic turns out to top Jordan in three-year samples, do you care? Does that change anything? You are trying to work backward from a presupposition here. You know Lebron coasted, and that needs to matter. But the reality is there is no objective measure of “effort relative to impact”. You are again conflating an aesthetic preference for a real reflection of a player’s value.

To your credit, you at least seem moderately aware of that, but I think it is important to call it what it is.

"Relevantly,"....so you admit the WOWYR does have some validity?

Well I was being sarcastic, but I never said it was outright invalid, no; merely limited (which you have yet to bother to address). Just seems like a weird thing to try to use to champion Jordan specifically.

Btw, Oscar Robertson and Jerry West, both contemporaries of Russell, rate out as better than him in such regressions. Know you claimed that Russell was indisputably a bigger outlier relative to era than MJ, yet once again we have an example of where a top 4 guy ever, was surpassed in impact per the regressions. Just a counter to the Robinson/Magic being better than MJ in WOWYR, and thus he must automatically be a fraud.

I continue to not see where you ever thought that I was somehow citing WOWYR as an element of that assertion, or where I said WOWYR proved Jordan was a “fraud”.

I am sure there must be better uses of your time than picking poorly conceived fights with no real goal in mind.

[LukaTheGOAT]

Before going further, RAPM is not some rigidly set formula. In this case, you are almost certainly correct that incorporating playoff data gives Lebron a substantial boost in a way not true of any other player. Different regressions will produce different values, although disparities probably should not be too drastic. Even just looking at NBA Shot Charts, it gives you a “luck-adjustment” option! And if use that suddenly Lebron has the #1 sample in 2013-16 and 2014-17 even though you were saying those are his coasting years.

But with that acknowledgment made, and even without getting into lineup data and the significance of bench rotations and “loads” and possession samples and all of that… why do you see “Lebron occasionally had lower RAPM stretches than Chris Paul and/or Steph Curry” as a meaningful argument? Was Chris Paul just “trying harder”? What is the point here.

And why do you just assume something similar would not occur with Jordan and Magic? As LukaTheGOAT has very kindly and relevantly helped highlight, regressions from that era favour Magic over Jordan. Lineup data is going to prefer Magic there, with Jordan’s teams failing to break the 4-SRS mark until 1991. By any measure he was also more strictly important to his team than Jordan was. Was he also just trying harder? By your eyetest I doubt you would agree, so again, what are we actually doing here?

If Magic turns out to top Jordan in three-year samples, do you care? Does that change anything? You are trying to work backward from a presupposition here. You know Lebron coasted, and that needs to matter. But the reality is there is no objective measure of “effort relative to impact”. You are again conflating an aesthetic preference for a real reflection of a player’s value.

To your credit, you at least seem moderately aware of that, but I think it is important to call it what it is.

"Relevantly,"....so you admit the WOWYR does have some validity?

Btw, Oscar Robertson and Jerry West, both contemporaries of Russell, rate out as better than him in such regressions. Know you claimed that Russell was indisputably a bigger outlier relative to era than MJ, yet once again we have an example of where a top 4 guy ever, was surpassed in impact per the regressions. Just a counter to the Robinson/Magic being better than MJ in WOWYR, and thus he must automatically be a fraud.

I dont think anyone argued this, less so with that wording

Is more than wowyr is not a data point that favors jordan over russel (or the other way around necessrrily)

With such a small and volatile sanple size stat like wowyr is better to just use it in a loose sense

Both players looking good in the stat is great, one of then looking bad in it would be worrisome (not the case here for either) so then since both llook great in that stat any jordan vs russel comparision needs to move to different angles and data points than wowyr

Wowyr is not a data point that agrees with jordan goat case, it would do more so for russel goat case if anythingh

No, WOWYR clearly favors Jordan. I prefer to use other metrics stated like Player Impact, BPM estimates, Backpicks BPM, etc. for evaluation, however since the people arguing do not appreciate anything with the box-score, I went the route of WOWYR.

And once again, my point was that Jordan does have an argument for being more of an outlier relative to era than Russell, not necessarily that I disagree. It was presented as a foregone conclusion, that Russell was better relative to era, yet even 10-year prime samples suggest, that is far from the truth.
Per the variability scores, Russell's high end woils still put him below Jordan in the regression.

And once again, Jordan clearly looks to be in a different tier based on WOWYR which isn't everything but suggests Jordan has more of an argument than suggested. You aren't answering what I am arguing.

[falcolombardi]

"Relevantly,"....so you admit the WOWYR does have some validity?

Btw, Oscar Robertson and Jerry West, both contemporaries of Russell, rate out as better than him in such regressions. Know you claimed that Russell was indisputably a bigger outlier relative to era than MJ, yet once again we have an example of where a top 4 guy ever, was surpassed in impact per the regressions. Just a counter to the Robinson/Magic being better than MJ in WOWYR, and thus he must automatically be a fraud.

I dont think anyone argued this, less so with that wording

Is more than wowyr is not a data point that favors jordan over russel (or the other way around necessrrily)

With such a small and volatile sanple size stat like wowyr is better to just use it in a loose sense

Both players looking good in the stat is great, one of then looking bad in it would be worrisome (not the case here for either) so then since both llook great in that stat any jordan vs russel comparision needs to move to different angles and data points than wowyr

Wowyr is not a data point that agrees with jordan goat case, it would do more so for russel goat case if anythingh

No, WOWYR clearly favors Jordan. I prefer to use other metrics stated like Player Impact, BPM estimates, Backpicks BPM, etc. for evaluation, however since the people arguing do not appreciate anything with the box-score, I went the route of WOWYR.

And once again, my point was that Jordan does have an argument for being more of an outlier relative to era than Russell, not necessarily that I disagree. It was presented as a foregone conclusion, that Russell was better relative to era, yet even 10-year prime samples suggest, that is far from the truth.

And once again, Jordan clearly looks to be in a different tier based on WOWYR which isn't everything but suggests Jordan has more of an argument than suggested. You aren't answering what I am arguing.

wowyr doesnt have jordan as #1 so i wouldnt use it as neither evidence for or against his goat case beyond both passing the smell test of being all time level in wowyr

going bychampionships as a ON russell looks incredible since by wowyr he had 35 win help with him off

Margins of victory were lower in the 60's with less teams so the net rating advantages needed to gain the same separation that a 80's or 90's team. Is why a direct conparision between points margins favors jordan era

[OhayoKD]

Maybe I'm off, but it feels like you're intentionally trying to avoid the WOWYR/Russell/Kareem stuff with this d-rob tangent. I'l address the more personal stuff now, but after that Imma be sticking to Russell, MJ, and WOWYR.

You were given an RAPM sample upon request in the "50 greatest players" thread:

All the 5 year rapm i've seen has favored duncan, for example:
https://www.reddit.com/r/nbadiscussion/comments/q50ucn/5_year_rapm_peaks_1997_to_2021_regular_season/
Also owly and squared circle's had duncan higher irrc.

You, claimed i never gave you a sample anyway:

Also I have twice in the past asked you to link me to some of the sources you are using and you never have

I gave you two more:

Unfortunately I can't find owly(i think)'s RAPM dump. but here is some of squared-circle's(full set may have been posted in the peaks project):
viewtopic.php?f=64&t=2206246
And a google-doc from another realgmer whose name I can't remember:
https://docs.google.com/spreadsheets/d/e/2PACX-1vQdG8Zv84zqKEzETDjd8KPsClcw9bPETX9v_x_KEAxjv9NrFaWikOoiSaciy1jbMiygg2D-V8DUQn0O/pubhtml?gid=112475182&single=true

I don't know if you're intentionally lying or just missing ****, but from this point forward I'm going to ask you to leave me out of whatever fight you're having with Enigma. This is what you argued:

I would greatly disagree that Russ, Wilt, and Kareem were larger outliers relative to era then MJ and I imagine that would be the case for a lot of people when evaluating them.

In terms of With or Without regressions, I would say MJ comes out looking strong, and the outlier of the group.

This is me directly addressing what you argued:

Yeah, this is misleading. Setting aside WOWYR's questionable utility (more on that later), you don't seem to be accounting for the fact that srs=/championship probability, and that the threshold for dominance in the 60's was vastly lower than the threshold required for dominance in the 60's. Going off the person who created the WOWRY metrics you're using here:

At the height of their dynasty, the Celtics were comically dominant. From 1962-65, their average margin-of-victory (MOV) was over 8 points per game. During the same time span, only two other teams even eclipsed 4 points per game – the ’64 Royals and the ’64 Warriors. And all of Boston’s separation was created by its historic defense, anchored by Russell:.

When we tie WOWYR to championships as opposed to "who has the best M.O.V", Russell(and to a lesser extent, Wilt), not Jordan is the outlier of the group you just referenced.

Boston was a 35-win team (-1.9 SRS) in 28 games he missed from 1958-69, and for the other 915 games of his career they played at a 59-win pace (6.4 SRS). This is a tiny piece of evidence – the years are spread out, teams change, and so on — but it echoes the same story as Russell’s other value signals.

WOWYR isn't saying "Russell is less valuable than Jordan", WOWYR is saying that "Russell needed far less help to win 11 Rings than Jordan needed to win 6". We can also extrapolate, based on Wilt mostly matching Russell in these metrics, that Wilt grades as a bigger contributor to championship probability than Jordan managed.

TLDR: WOWYR says Russell, and Wilt, are much bigger outliers than Jordan.

Of course, I don't think we should be putting this much stock into WOWYR, and fwiw, WOWYR's inventor seems to agree. Looking at both the Kareem and Russell "impact evaluations", that quoted portion is the only time WOWYR numbers are listed(they are linked but not stated specifically regarding 1970 Kareem) and its attached with the caveat of "tiny sample". Instead Ben focuses on unregularized data from stretches where stars miss significant time. To quote myself:

Over 10 year spans or so, wowy will typically be based on a handful of games per season with players who've probably changed signifcantly during that time span. That's not really all that useful. What you want to do is look for moments where players or teammates miss(or are absent) from an unusally high number of games so you can get the largest samples. A good starting point for this would be the year before a player joins a team or the year after a leaves a team as you get a full season sample of data(70 Celtics, 84 Bulls, 84 Rockets, 69 Bucks, ect.) Then you can track roster changes, and granular stuff to adjust or guesstimate if the team improved got worse, ect. Lookign for concentrated stretches of missed game time, or how teams do when a star's teammates go out also can be useful. Ideally you want as much of this type of data for a player in various contexts and then you can compare players in these various situations directly.

Even with RAPM, a more polished variant of what you're citing for Russell, Ben will defer to unregularized samples when dealing with outlier impact:

Like Nash, LeBron was supercharging dependent talent — finishers who disproportionately benefited from shots served to them on a silver platter. So with his talents in South Beach, Cleveland crumbled in 2011. While most teams fall off after losing a superstar, none imploded like the Lebron-less Cavs; in 21 games with a similar group of players, they played at an anemic 18-win pace (-8.9 SRS) before injuries ravaged their lineup. LeBron’s not worth 40 wins on a typical club, but no player in history has correlated more strongly with such massive, worst-to-first impact.

This is because regularization, a process that is used in generating RAPM and WOWYR sets artificial caps on player value:

In layman terms, the process of calculating RAPM involves some math that distorts the "real scale" of the numbers in favor of accuracy.

When impact exceeds this scale, value is misattributed to role players.

Luckily, we don't need to rely on a smattering of minutes to assess Russell, because we actually have an 82 game sample of "without" for Russell in 1970. The celtics, despite seeing their offense improve(second best player sees his volume skyrocket), dropped to 35 wins with Russell retiring. The previous year, Bill beat a superteam to win a championship as a player-coach on his last legs. With more help in 1990(The Bulls were a 27 win team before they drafted Jordan, incrementally improved after drafting Pippen and Grant, then saw their srs skyrocket with the introduction of the triangle) Apex Jordan managed 55 wins and a close loss in the Conference Finals.

Russell's success was also substantially less tied to his teammates. While Jordan's success correlates with Pippen as strongly as Steph's correlates with Draymond, the Celtics were able to win, win, and keep winning regardless of who left and who stayed.

Both WOWY and WOWYR suggest Russell actually had less help than Jordan. Considering that the Celtics were vastly more dominant than the Bulls, I don't think there's much of a case here for Jordan as an era-relative outlier.

And this is you being logically inconsistent:

So as I stated, there is no full-season data suggesting that Robinson's impact exceeded Jordan (as SquareCircle only includes partial segments of the season), and you once wasted time by not even being willing to admit this

Setting aside that I "admitted this" in the orginal post a week ago...

(sample size is not even)

how are you going to complain about sample size while trying to estimate russell's impact from 28 games over 13 years?

WOWYR (With or Without You, Regressed) is a game-level plus-minus measurement and a cousin of play-by-play measurements like Adjusted Plus-Minus (APM). Instead of using results from lineups within a game (play-by-play data) like traditional APM, game-level plus-minus uses final scores from game to game for the players from that game.

WOWYR has a larger sample-size issue than the partial samples for peak MJ I threw in as "supplements", yet you've made it the centerpiece of your case.

Jaivi said "seasons" not "creators" and the reason why he's not comfortable with that comparison is because regularization "distorts" actual impact, an issue with is even more prevalent with WOWYR.

And off course, even if you were to ignore all of that, we're left with Russell leading a vastly more dominant team than the Bulls with 35-win help throughout his prime(at least per wowyr). Russell is winning way more, with significantly less. WOWYR actually supports the 82 game sample(as opposed to 28 games from 13 seasons) we have in 1970. In this case the regularized and unregularized data tell us the same story. Russell was a much, much bigger era-relative outlier. But you just chose to ignore that and go on some tangent about d-rob and enigma.

I(well mostly 70's) also addressed the Kareem-MJ comparison, and you ignored that too, but for the sake of clarity, I think we should streamline this conversation to Jordan and Bill.