Magic Johnson vs Kevin Garnett

Moderators: trex_8063, PaulieWal, Quotatious, penbeast0, Doctor MJ, Clyde Frazier

udfa
Sixth Man
Posts: 1,769
And1: 1,686
Joined: Apr 06, 2017

Re: Magic Johnson vs Kevin Garnett 

Post#41 » by udfa » Sat Aug 10, 2019 2:26 pm

1. Magic
2. Magic
3. Magic
User avatar
kendogg
Sixth Man
Posts: 1,782
And1: 203
Joined: Apr 08, 2001
Location: Cincinnati

Re: Magic Johnson vs Kevin Garnett 

Post#42 » by kendogg » Sat Aug 10, 2019 2:59 pm

eminence wrote:
kendogg wrote:
eminence wrote:When a team adds a star level player they tend to actively add better support pieces as well to turn themselves into real contenders.

I repeat, no one player is worth 30 wins, let alone 40 (the only team to make a 40+ win leap is, ironically, the '08 Celtics).


When I say these players are worth 40+ wins, I mean going to a team with no star talent whatsoever. It is a hypothetical thing really, because mega-superstars don't go to bad teams without promises they will build around them.

A better way to say it would be, a guy who is not only a mega-superstar but also a leader, like Magic, will never be on a team that gets less than 40 wins. Even if there are no other stars on the team. Jordan was on a 30 win team once and he was rather immature as a youngster as well as always being a known **** and bad teammate off the court. He doesn't lift up his teammates he takes pressure off them. That's not the same thing. Taking pressure off is just exerting your gravitas with your own skills. The other is actually finding ways to work with your teammates and make them more effective. I would argue that Jordan never really had that trait and the Bulls just lucked out in stacking the team with vets that could largely ignore Jordan's antics and work around him.

KG is a similar circumstance. He's an **** that nobody likes and it's why he has completely disappeared from the NBA scene. Nobody really wants to play with a guy like that, especially when they aren't as good as Jordan. The 2008 Celtics were a dysfunctional mess that almost blew their run multiple times and fell apart less than a year later. It's also why he had multiple 30 win seasons on the Wolves. He's a bad teammate and doesn't lift them up like the true legends do.
Timmyyy
Junior
Posts: 290
And1: 292
Joined: May 21, 2019
   

Re: Magic Johnson vs Kevin Garnett 

Post#43 » by Timmyyy » Sat Aug 10, 2019 3:47 pm

kendogg wrote:I've read most of the arguments in all these KG threads and the people who support him really all hinge their argument on KG's incredible PER's and how much he can stack a box score. Very hollow argument when he constantly loses to teams in the first round he really has no business losing to if he's the MVP and a true floor raiser. Westbrook can stack a box score too, but he can't get out of the first around either (without Durant). Some people are just worse than their box scores suggest. KG is one of those people. And in all likelihood, because he is a bad teammate.


I don't know what you read in the numerous KG threads in the last few weeks here on the PC board, but what you are saying isn't the argument the KG supporters are providing.
Their argument is that all the +/- data that is out there is having KG as a true monster that even stacks up with Lebron. Since he had this huge impact on the point differentials, him losing is only showing how bad his teams were and not that he wasn't able to elevate them.
DatAsh
Junior
Posts: 405
And1: 186
Joined: Sep 25, 2015

Re: Magic Johnson vs Kevin Garnett 

Post#44 » by DatAsh » Sat Aug 10, 2019 4:23 pm

Timmyyy wrote:
kendogg wrote:I've read most of the arguments in all these KG threads and the people who support him really all hinge their argument on KG's incredible PER's and how much he can stack a box score. Very hollow argument when he constantly loses to teams in the first round he really has no business losing to if he's the MVP and a true floor raiser. Westbrook can stack a box score too, but he can't get out of the first around either (without Durant). Some people are just worse than their box scores suggest. KG is one of those people. And in all likelihood, because he is a bad teammate.


I don't know what you read in the numerous KG threads in the last few weeks here on the PC board, but what you are saying isn't the argument the KG supporters are providing.
Their argument is that all the +/- data that is out there is having KG as a true monster that even stacks up with Lebron. Since he had this huge impact on the point differentials, him losing is only showing how bad his teams were and not that he wasn't able to elevate them.


Yeah that struck me as kinda odd too. As a KG guy myself, kendogg's interpretation of our pro KG argument seems almost exactly opposite of the way I would characterize it. I guess I can't speak for all of the KG guys here, but generally the argument for KG is that while he has fairly weak box scores(relative to the very top guys), his impact on winning(+/-) is right up there with the very best, if not a level above everyone other than Lebron.

Other than Russell, KG is the guy who's impact far exceeds the value that his box scores would suggest.
WarriorGM
Starter
Posts: 2,447
And1: 1,176
Joined: Aug 19, 2017

Re: Magic Johnson vs Kevin Garnett 

Post#45 » by WarriorGM » Sat Aug 10, 2019 4:46 pm

DatAsh wrote:
Yeah that struck me as kinda odd too. As a KG guy myself, kendogg's interpretation of our pro KG argument seems almost exactly opposite of the way I would characterize it. I guess I can't speak for all of the KG guys here, but generally the argument for KG is that while he has fairly weak box scores(relative to the very top guys), his impact on winning(+/-) is right up there with the very best, if not a level above everyone other than Lebron.

Other than Russell, KG is the guy who's impact far exceeds the value that his box scores would suggest.


Can you calculate the +/- numbers you refer to? Can you explain each step of the calculation and what is being done? Do you know what assumptions are being made?

You can have an NBA reject join a foreign club and put up Jordan numbers. How do you know that something similar isn't what's going on with KG? Do you consider the marginal skill improvement required to shoot from a 10% to 30% free throw rate the same as from 50% to 70%? How do you know a similar phenomenon isn't at work?

The reliance on opaque statistically complicated numbers is a flashing red light to me when it comes to KG. If he was so good it should be more obvious and show up elsewhere.
User avatar
kendogg
Sixth Man
Posts: 1,782
And1: 203
Joined: Apr 08, 2001
Location: Cincinnati

Re: Magic Johnson vs Kevin Garnett 

Post#46 » by kendogg » Sat Aug 10, 2019 4:51 pm

DatAsh wrote:Yeah that struck me as kinda odd too. As a KG guy myself, kendogg's interpretation of our pro KG argument seems almost exactly opposite of the way I would characterize it. I guess I can't speak for all of the KG guys here, but generally the argument for KG is that while he has fairly weak box scores(relative to the very top guys), his impact on winning(+/-) is right up there with the very best, if not a level above everyone other than Lebron.

Other than Russell, KG is the guy who's impact far exceeds the value that his box scores would suggest.


BPM is just a rate stat. That's just PER per minute more or less (with usage factored in as well). And yes his defense doesn't show in the box score, but neither does all his negative intangibles, like not getting along with his teammates or lifting them up.

I don't see how you can possibly make the case that Garnett's impact on winning is 2nd to LeBron. Where are you getting this from?
User avatar
First Take
Ballboy
Posts: 29
And1: 24
Joined: Aug 04, 2019
 

Re: Magic Johnson vs Kevin Garnett 

Post#47 » by First Take » Sat Aug 10, 2019 8:09 pm

1993Playoffs wrote:1. Who had the better one year peak?

2. Whose extended prime was better ?

3. Who ranks higher all time?


1. Kevin Garnett
2. Kevin Garnett
3. Erving Johnson

I should add that comparing a guy who primarily played point guard to a power forward is a tricky debate but I try to answer it to the best of my abilities.
Favorite Basketball Players: LeBron James, Dwayne Wade, Dwight Howard, Kawhi Leonard
Favorite Analyst: Stephen A. Smith, Skip Bayless, Shannon Sharpe
Favorite Basketball Teams: Miami Heat, San Antonio Spurs, Toronto Raptors, Los Angeles Lakers
DatAsh
Junior
Posts: 405
And1: 186
Joined: Sep 25, 2015

Re: Magic Johnson vs Kevin Garnett 

Post#48 » by DatAsh » Sat Aug 10, 2019 9:00 pm

WarriorGM wrote:Can you calculate the +/- numbers you refer to? Can you explain each step of the calculation and what is being done? Do you know what assumptions are being made?


I don't have the base play by play data, nor do I have a program that can run the analysis. I do understand the process, though, as it's essentially what I do for a living. I run very similar analyses on large sets of heat transfer data.

Raw +/- is not really a calculation, more just an observation.

Adjusted impact metrics are like a system of equations, if you've ever taken linear algrebra.
x1 + x2 + x3 + x4 + x5 + x6 + x7 + x8 + x9 + x10 = point_diff

where you have hundreds of x values(representing the number of players) and thousands of equations, and your goal is to find the values that best fit all of the equations.

That said, as long as you understand what stats like RAPM are saying, I don't think you need to fully understand the math behind them in order to use them in your player analysis. They don't tell you who the best players are, they tell you which players most improved/worsened the winningness of their teams, and can be heavily influenced by role and situation(see guys like Korver).

WarriorGM wrote:You can have an NBA reject join a foreign club and put up Jordan numbers. How do you know that something similar isn't what's going on with KG? Do you consider the marginal skill improvement required to shoot from a 10% to 30% free throw rate the same as from 50% to 70%? How do you know a similar phenomenon isn't at work?

I'm not entirely sure what you're asking here. Garnett was playing in the NBA, and the impact metrics come entirely from that sample.
Are you asking "is it possible that there is some unknown circumstance that's causing Garnett's teams to improve more when he plays and fall more when he sits, a circumstance which is not directly related to his actual goodness?". If that's your question, I would say yes, it's definitely possible, but I personally find it unlikely.

WarriorGM wrote:The reliance on opaque statistically complicated numbers is a flashing red light to me when it comes to KG. If he was so good it should be more obvious and show up elsewhere.


I think your inclination to look elsewhere is a good one, and I agree. It's always good to not put all your eggs in one basket, and how you divide your eggs up among the impact/box-score/eye-test/accolades/outside-opinion baskets is your discretion, and what makes this so fun. Everyone is different.

I personally put a lot of eggs in the impact metrics basket, but I'm also influenced by box-score and eye-test. Looking at KG, impact metrics suggest he's basically Lebron, but I have Lebron at 1 and KG at 6(atm). I have Duncan ahead of KG, despite the fact that KG generally crushes Duncan in impact metrics, as box score and eye-test have led me to believe that Duncan's offensive skillset tends to hold up better in the postseason, and postseason impact is where the majority of championship odds come from.

Hopefully that answers some of your questions. I could be way off on KG, and most people would say that I am, so I don't blame you for thinking that I am.
DatAsh
Junior
Posts: 405
And1: 186
Joined: Sep 25, 2015

Re: Magic Johnson vs Kevin Garnett 

Post#49 » by DatAsh » Sat Aug 10, 2019 9:03 pm

kendogg wrote:
DatAsh wrote:Yeah that struck me as kinda odd too. As a KG guy myself, kendogg's interpretation of our pro KG argument seems almost exactly opposite of the way I would characterize it. I guess I can't speak for all of the KG guys here, but generally the argument for KG is that while he has fairly weak box scores(relative to the very top guys), his impact on winning(+/-) is right up there with the very best, if not a level above everyone other than Lebron.

Other than Russell, KG is the guy who's impact far exceeds the value that his box scores would suggest.


BPM is just a rate stat. That's just PER per minute more or less (with usage factored in as well). And yes his defense doesn't show in the box score, but neither does all his negative intangibles, like not getting along with his teammates or lifting them up.

I don't see how you can possibly make the case that Garnett's impact on winning is 2nd to LeBron. Where are you getting this from?


BPM and PER are just box score stats, and they don't directly measure a players impact on winning, so they're not what I'm talking about here. BPM is based off impact stats, but it's ultimately a box score metric.

Box score metrics like BPM and PER underrate KG. Impact metrics are where he shines, and I would say most of the people that rate him highly on this board are heavily influenced by impact metrics.
User avatar
kendogg
Sixth Man
Posts: 1,782
And1: 203
Joined: Apr 08, 2001
Location: Cincinnati

Re: Magic Johnson vs Kevin Garnett 

Post#50 » by kendogg » Sun Aug 11, 2019 12:00 am

DatAsh wrote:his impact on winning(+/-) is right up there with the very best, if not a level above everyone other than Lebron


I assumed you meant BPM, but RPM is really somewhat the same deal though not entirely based on box score. It's basically box score with a somewhat standardized supporting cast, but it still is based on box scores just a lot more data than normal to try and measure the average production of a possession of a player. Also, RPM wasn't around for KG's career as far as I know, which is why I assumed you meant BPM.

The thing is, nobody is arguing KG wasn't a productive player. Just like nobody argues that Westbrook is a productive player. But sometimes a productive player is doing so without really involving his teammates or lifting them up. Sometimes stats can be empty stats. He never got along with any of his star teammates and he actually had quite a few in his career. It's probably on average even worse for the journeymen on the team. Their best option is probably to never talk or make eye contact with KG. How does any team win a championship with a guy like that? Only if they get lucky and have overwhelming talent (2008 Celtics, who still almost failed in every single series).

I've seen all of KG's playoff games. I recognize he is a legit superstar and one of the best 2-way players. But he's a tier below Magic as a player primarily due to his personality more than anything. It is really what kept him from GOAT tier status.
DatAsh
Junior
Posts: 405
And1: 186
Joined: Sep 25, 2015

Re: Magic Johnson vs Kevin Garnett 

Post#51 » by DatAsh » Sun Aug 11, 2019 12:15 am

kendogg wrote:I assumed you meant BPM, but RPM is really somewhat the same deal though not entirely based on box score. It's basically box score with a somewhat standardized supporting cast, but it still is based on box scores just a lot more data than normal to try and measure the average production of a possession of a player. Also, RPM wasn't around for KG's career as far as I know, which is why I assumed you meant BPM.


I wasn't referring to BPM or RPM, I was talking about RAPM. It's more similar to RPM, but it doesn't use a box score prior. You're correct that we don't have RPM for KG's career.
DatAsh
Junior
Posts: 405
And1: 186
Joined: Sep 25, 2015

Re: Magic Johnson vs Kevin Garnett 

Post#52 » by DatAsh » Sun Aug 11, 2019 12:17 am

kendogg wrote:I've seen all of KG's playoff games. I recognize he is a legit superstar and one of the best 2-way players. But he's a tier below Magic as a player primarily due to his personality more than anything. It is really what kept him from GOAT tier status.


Important to note that this is your opinion. I have KG a tier above Magic, but that's the fun of these online forums :D
DatAsh
Junior
Posts: 405
And1: 186
Joined: Sep 25, 2015

Re: Magic Johnson vs Kevin Garnett 

Post#53 » by DatAsh » Sun Aug 11, 2019 3:12 am

kendogg wrote:
DatAsh wrote:his impact on winning(+/-) is right up there with the very best, if not a level above everyone other than Lebron


I assumed you meant BPM, but RPM is really somewhat the same deal though not entirely based on box score. It's basically box score with a somewhat standardized supporting cast, but it still is based on box scores just a lot more data than normal to try and measure the average production of a possession of a player. Also, RPM wasn't around for KG's career as far as I know, which is why I assumed you meant BPM.

The thing is, nobody is arguing KG wasn't a productive player. Just like nobody argues that Westbrook is a productive player. But sometimes a productive player is doing so without really involving his teammates or lifting them up. Sometimes stats can be empty stats. He never got along with any of his star teammates and he actually had quite a few in his career. It's probably on average even worse for the journeymen on the team. Their best option is probably to never talk or make eye contact with KG. How does any team win a championship with a guy like that? Only if they get lucky and have overwhelming talent (2008 Celtics, who still almost failed in every single series).

I've seen all of KG's playoff games. I recognize he is a legit superstar and one of the best 2-way players. But he's a tier below Magic as a player primarily due to his personality more than anything. It is really what kept him from GOAT tier status.


Something like this is what I'm referring to.

https://public.tableau.com/views/14YearRAPM/14YearRAPM?:embed=y&:showVizHome=no
Pg81
Starter
Posts: 2,437
And1: 1,042
Joined: Apr 20, 2014
 

Re: Magic Johnson vs Kevin Garnett 

Post#54 » by Pg81 » Sun Aug 11, 2019 7:09 am

eminence wrote:
kendogg wrote:
eminence wrote:No player in history is worth 30-40+ wins. The very best approach 30 on teams that are among the worst in the league without them.


Well of course the team would need to be bad to be able to raise 40 wins. Can't add 40 wins to a 40 win team that's not possible. But guys have done it if they are very developed rookies, like Larry Bird and David Robinson.


Bird got 32 on a team where a AS level guy was rounding back into form from a serious injury and another young star was continuing to develop (Tiny/Maxwell).

Robinson got 35 on a team that added Elliot and Cheeks/Strickland. (Along with the Robertson/Cummings swap).

When a team adds a star level player they tend to actively add better support pieces as well to turn themselves into real contenders.

I repeat, no one player is worth 30 wins, let alone 40 (the only team to make a 40+ win leap is, ironically, the '08 Celtics).


The Celtics were the same team from the year before Bird joined. Maxwell just had a breakout season, in fact it was probably his best season in his career, yet the team was still bottom of the barrel. Also in what world is Maxwell an AS? Because of one season where he averaged 19 only to never average more than 17 again, and that was right after Bird joined. For career he averaged 12.5/6/2. That is a decent statline for a third/fourth/fifth but a far cry from a reliable AS level player and on top of that he was never selected for all star.
Tiny Archibald had injured himself in the 77/78 season, two years before Bird came back so not an argument whatsoever. He was already 31 and had declined quite a bit when Bird joined the Celtics. Sorry to burst your bubble but that team was atrocious and neither Maxwell nor Tiny were that good at that point that just mentioning their name and make some vague statements about them make them anything more. The fact is both were healthy Maxwell having his career best season the year before and they still did not manage more than bottom of the barrel, yet with just the addition of Bird they went to the WCF right away and won a title the next year with Robert Parish, the only other guy during that time who was an actual all star level player. THAT is actual impact.

kendogg wrote:
DatAsh wrote:his impact on winning(+/-) is right up there with the very best, if not a level above everyone other than Lebron


I assumed you meant BPM, but RPM is really somewhat the same deal though not entirely based on box score. It's basically box score with a somewhat standardized supporting cast, but it still is based on box scores just a lot more data than normal to try and measure the average production of a possession of a player. Also, RPM wasn't around for KG's career as far as I know, which is why I assumed you meant BPM.

The thing is, nobody is arguing KG wasn't a productive player. Just like nobody argues that Westbrook is a productive player. But sometimes a productive player is doing so without really involving his teammates or lifting them up. Sometimes stats can be empty stats. He never got along with any of his star teammates and he actually had quite a few in his career. It's probably on average even worse for the journeymen on the team. Their best option is probably to never talk or make eye contact with KG. How does any team win a championship with a guy like that? Only if they get lucky and have overwhelming talent (2008 Celtics, who still almost failed in every single series).

I've seen all of KG's playoff games. I recognize he is a legit superstar and one of the best 2-way players. But he's a tier below Magic as a player primarily due to his personality more than anything. It is really what kept him from GOAT tier status.


To me Magic is the better ball handler, better passer and better scorer. Why better scorer? Because when he wanted to he could take over games like many other all time great scorer, something KG was never able to do. Magic has multiple 30+ and 35+ point playoff games, how many does KG have? None or close to none. KG just was not that good as a scorer, Magic just opted to dish out instead of score and led probably the greatest offense of all time with the Showtime Lakers who before his arrival were languishing in mediocrity despite Kareem on the roster. They were barely making any impact in the playoffs only to win the title right off the bat once Magic joined who had an insane game 6 performance with Kareem out with injury and winning finals MvP. I can seriously not see KG even remotely reaching that level of impact or just play.
WarriorGM
Starter
Posts: 2,447
And1: 1,176
Joined: Aug 19, 2017

Re: Magic Johnson vs Kevin Garnett 

Post#55 » by WarriorGM » Sun Aug 11, 2019 7:30 am

DatAsh wrote:
WarriorGM wrote:Can you calculate the +/- numbers you refer to? Can you explain each step of the calculation and what is being done? Do you know what assumptions are being made?


That said, as long as you understand what stats like RAPM are saying, I don't think you need to fully understand the math behind them in order to use them in your player analysis.


But do you really understand what the stats like RAPM are saying? For example if an event occurs three standard deviations away from the average it could be understood to be a very rare event. Some forecasters tasked with predicting the probability of a worst case scenario in the stock market thought beforehand what happened in the 2008 market meltdown was a once in a century kind of event. Sadly for them they learned it actually happens more frequently. Their calculations I presume were correct. Their assumptions and understanding weren't. I'm sure you are acquainted with many other such examples although I do wonder if your job dealing with calculations based on heat transfers might be giving you a false sense of security on the application of statistics. You seem to be dealing with something bound to the laws of physics which are easier to predict and are not as subject to the whims of human caprice and frailty and the multiple variables that might affect them.

DatAsh wrote:They don't tell you who the best players are, they tell you which players most improved/worsened the winningness of their teams, and can be heavily influenced by role and situation(see guys like Korver).

Isn't that one way of defining who the best players are?

DatAsh wrote:
WarriorGM wrote:You can have an NBA reject join a foreign club and put up Jordan numbers. How do you know that something similar isn't what's going on with KG? Do you consider the marginal skill improvement required to shoot from a 10% to 30% free throw rate the same as from 50% to 70%? How do you know a similar phenomenon isn't at work?

I'm not entirely sure what you're asking here. Garnett was playing in the NBA, and the impact metrics come entirely from that sample.
Are you asking "is it possible that there is some unknown circumstance that's causing Garnett's teams to improve more when he plays and fall more when he sits, a circumstance which is not directly related to his actual goodness?". If that's your question, I would say yes, it's definitely possible, but I personally find it unlikely.


No, my question is directed to whether we know if playing on a weak team inflates RAPM relative to playing on a strong team. The reason always given for Garnett not accomplishing as much as his supporters believe he could is that he played with bad teammates. He may have played in the NBA but if we accept this argument one could translate it to mean that he played with an NBA B-Team. Extend the idea further and you have something like the example I gave. This comes back to understanding of RAPM and derivative numbers like it. Does anyone really understand it?

It may be the above factor is taken into account by RAPM. But I wouldn't be the least bit surprised if there are probably other variables that can distort or confound RAPM results and if one isn't aware of them make inaccurate conclusions based on RAPM.

DatAsh wrote:
WarriorGM wrote:The reliance on opaque statistically complicated numbers is a flashing red light to me when it comes to KG. If he was so good it should be more obvious and show up elsewhere.


I think your inclination to look elsewhere is a good one, and I agree. It's always good to not put all your eggs in one basket, and how you divide your eggs up among the impact/box-score/eye-test/accolades/outside-opinion baskets is your discretion, and what makes this so fun. Everyone is different.

I personally put a lot of eggs in the impact metrics basket, but I'm also influenced by box-score and eye-test. Looking at KG, impact metrics suggest he's basically Lebron, but I have Lebron at 1 and KG at 6(atm). I have Duncan ahead of KG, despite the fact that KG generally crushes Duncan in impact metrics, as box score and eye-test have led me to believe that Duncan's offensive skillset tends to hold up better in the postseason, and postseason impact is where the majority of championship odds come from.

Hopefully that answers some of your questions. I could be way off on KG, and most people would say that I am, so I don't blame you for thinking that I am.


My apologies, my question was rhetorical and I wasn't expecting a long reply. I am sorry if that caused you to write something longer than was strictly necessary to address me in particular regarding the calculation of adjusted plus-minus. But thank you for the thought put into it and it seems it may have helped others. Your explanation for your use of RAPM is understandable and I would agree it certainly seems to point in the right direction generally but I find it too imprecise to make sweeping conclusions on it alone. That it is not transparent and easily arrived at and understood on one's own only makes me even more skeptical of it. As a derivative it is used to tease out associations that might not be obvious in the raw numbers but if the raw numbers make an obvious association I tend to find those signals more reliable.
Timmyyy
Junior
Posts: 290
And1: 292
Joined: May 21, 2019
   

Re: Magic Johnson vs Kevin Garnett 

Post#56 » by Timmyyy » Sun Aug 11, 2019 8:56 am

WarriorGM wrote:
DatAsh wrote:That said, as long as you understand what stats like RAPM are saying, I don't think you need to fully understand the math behind them in order to use them in your player analysis.


But do you really understand what the stats like RAPM are saying? For example if an event occurs three standard deviations away from the average it could be understood to be a very rare event. Some forecasters tasked with predicting the probability of a worst case scenario in the stock market thought beforehand what happened in the 2008 market meltdown was a once in a century kind of event. Sadly for them they learned it actually happens more frequently. Their calculations I presume were correct. Their assumptions and understanding weren't. I'm sure you are acquainted with many other such examples although I do wonder if your job dealing with calculations based on heat transfers might be giving you a false sense of security on the application of statistics. You seem to be dealing with something bound to the laws of physics which are easier to predict and are not as subject to the whims of human caprice and frailty and the multiple variables that might affect them.


If you understand what happens in the regression and the adjustments and priors being used, yes, you have a solid understanding what is happening there. The stock market is also a completely different situation as right now, we are not talking about about predicting the future as in the stock market (even if RAPM is actually a tool for predicting), we try to say how precise the correlation between players being on the court and their teams point differentials measured in the RAPM numbers is to the real world impact these guys have and when you have the needed knowledge to see where the strength and weaknesses of the stat are you have a way more precise picture than looking at team wins.

WarriorGM wrote:
DatAsh wrote:They don't tell you who the best players are, they tell you which players most improved/worsened the winningness of their teams, and can be heavily influenced by role and situation(see guys like Korver).

Isn't that one way of defining who the best players are?


When you adjust for role and other circumstances it is the nearest you can get, exactly. That is why people like DatAsh are trusting these numbers heavily while adjusting for things that do not get caught in the numbers.

WarriorGM wrote:
DatAsh wrote:I'm not entirely sure what you're asking here. Garnett was playing in the NBA, and the impact metrics come entirely from that sample.
Are you asking "is it possible that there is some unknown circumstance that's causing Garnett's teams to improve more when he plays and fall more when he sits, a circumstance which is not directly related to his actual goodness?". If that's your question, I would say yes, it's definitely possible, but I personally find it unlikely.


No, my question is directed to whether we know if playing on a weak team inflates RAPM relative to playing on a strong team. The reason always given for Garnett not accomplishing as much as his supporters believe he could is that he played with bad teammates. He may have played in the NBA but if we accept this argument one could translate it to mean that he played with an NBA B-Team. Extend the idea further and you have something like the example I gave. This comes back to understanding of RAPM and derivative numbers like it. Does anyone really understand it?

It may be the above factor is taken into account by RAPM. But I wouldn't be the least bit surprised if there are probably other variables that can distort or confound RAPM results and if one isn't aware of them make inaccurate conclusions based on RAPM.


The regression itself adjusts for players playing on bad teams having huge on/off numbers AND guys playing on good teams having huge on court net ratings (yes, it could go both ways, not just one). In smaller samples these effects aren't perfectly accounted for but the larger the sample size is the less the error gets by the nature of the regression. SInce KG ranks great in single years, in multiyears in 5 year spans in 10 year spans and for his whole career (no matter if Boston years or MIN) the probability of him being overrated by the data is close to 0, because of the fact that different sample sizes, different data sets and so on all paint the same picture.

WarriorGM wrote:
DatAsh wrote:I think your inclination to look elsewhere is a good one, and I agree. It's always good to not put all your eggs in one basket, and how you divide your eggs up among the impact/box-score/eye-test/accolades/outside-opinion baskets is your discretion, and what makes this so fun. Everyone is different.

I personally put a lot of eggs in the impact metrics basket, but I'm also influenced by box-score and eye-test. Looking at KG, impact metrics suggest he's basically Lebron, but I have Lebron at 1 and KG at 6(atm). I have Duncan ahead of KG, despite the fact that KG generally crushes Duncan in impact metrics, as box score and eye-test have led me to believe that Duncan's offensive skillset tends to hold up better in the postseason, and postseason impact is where the majority of championship odds come from.

Hopefully that answers some of your questions. I could be way off on KG, and most people would say that I am, so I don't blame you for thinking that I am.


My apologies, my question was rhetorical and I wasn't expecting a long reply. I am sorry if that caused you to write something longer than was strictly necessary to address me in particular regarding the calculation of adjusted plus-minus. But thank you for the thought put into it and it seems it may have helped others. Your explanation for your use of RAPM is understandable and I would agree it certainly seems to point in the right direction generally but I find it too imprecise to make sweeping conclusions on it alone. That it is not transparent and easily arrived at and understood on one's own only makes me even more skeptical of it. As a derivative it is used to tease out associations that might not be obvious in the raw numbers but if the raw numbers make an obvious association I tend to find those signals more reliable.


It definitely isn't a tool to base conclusions on without further context. Nobody that uses these numbers is doing it that way (somehow everybody that doesn't like what he sees in RAPM is using this argument when nobody is approaching the data that way).
Yet you draw conclusions from team wins and use it more heavily in your analysis than any 'RAPM user' uses RAPM. Additionally you are pointing to raw numbers, which are the input for RAPM, being more reliable, when it is just the unadjusted version of RAPM (so by virtue of making adjustments RAPM IS more precise).
It just isn't clear why we should use the least adjusted data there is (team wins) and the raw input for something, above the thing itself when the thing is adjusting for correlation issues between the team and the player in question. We want the impact of the player not the team.
User avatar
kendogg
Sixth Man
Posts: 1,782
And1: 203
Joined: Apr 08, 2001
Location: Cincinnati

Re: Magic Johnson vs Kevin Garnett 

Post#57 » by kendogg » Sun Aug 11, 2019 1:40 pm

RAPM is still based on box scores, it is just using a lot more data to try and average out what the average competition is. And yes, KG's RAPM is high, mainly because of his DRAPM.

Thing is, we already know KG is a top notch help defender and versatile man defender. Most would not argue that he is probably top 5 all-time in that category. But defensively, the thing is, that the most important aspects of defense is not help defense, but rim protection and rebounding. Garnett is a top level rebounder, but rim protector he's just very good. There are a lot of players who are worse help defenders and man defenders than KG but better rim protectors and as good or better rebounders.

For example, of guys that COULD have played PF full time, who are better rim protectors and equal or better rebounders to KG:

Bill Russell
Hakeem Olajuwon
David Robinson
Ben Wallace
Tim Duncan

So KG might not crack my top 5 rim protecting PF's. And that's not including all of the true Centers who are better rim protectors either. Considering KG is known as a defensive specialist, that doesn't bode well my feelings toward his legacy, especially once I factor in that he's a bad teammate and chemistry problem.
User avatar
Zeitgeister
Lead Assistant
Posts: 5,692
And1: 2,144
Joined: Nov 11, 2008
   

Re: Magic Johnson vs Kevin Garnett 

Post#58 » by Zeitgeister » Sun Aug 11, 2019 1:50 pm

kendogg wrote:RAPM is still based on box scores, it is just using a lot more data to try and average out what the average competition is. And yes, KG's RAPM is high, mainly because of his DRAPM.

Thing is, we already know KG is a top notch help defender and versatile man defender. Most would not argue that he is probably top 5 all-time in that category. But defensively, the thing is, that the most important aspects of defense is not help defense, but rim protection and rebounding. Garnett is a top level rebounder, but rim protector he's just very good. There are a lot of players who are worse help defenders and man defenders than KG but better rim protectors and as good or better rebounders.

For example, of guys that COULD have played PF full time, who are better rim protectors and equal or better rebounders to KG:

Bill Russell
Hakeem Olajuwon
David Robinson
Ben Wallace
Tim Duncan

So KG might not crack my top 5 rim protecting PF's. And that's not including all of the true Centers who are better rim protectors either. Considering KG is known as a defensive specialist, that doesn't bode well my feelings toward his legacy, especially once I factor in that he's a bad teammate and chemistry problem.


RAPM does not use box score data, you may be thinking of RPM.
User avatar
Jaivl
Assistant Coach
Posts: 4,434
And1: 3,572
Joined: Jan 28, 2014
Location: A Coruña, Spain
Contact:
   

Re: Magic Johnson vs Kevin Garnett 

Post#59 » by Jaivl » Sun Aug 11, 2019 2:04 pm

WarriorGM wrote:
DatAsh wrote:
Yeah that struck me as kinda odd too. As a KG guy myself, kendogg's interpretation of our pro KG argument seems almost exactly opposite of the way I would characterize it. I guess I can't speak for all of the KG guys here, but generally the argument for KG is that while he has fairly weak box scores(relative to the very top guys), his impact on winning(+/-) is right up there with the very best, if not a level above everyone other than Lebron.

Other than Russell, KG is the guy who's impact far exceeds the value that his box scores would suggest.


Can you calculate the +/- numbers you refer to? Can you explain each step of the calculation and what is being done? Do you know what assumptions are being made?

Yes

KG
KG
KG
Maf wrote:I'd undestand if anyone had KG outside top ten PF's. Having him top five all-time? Often I jokingly rank Kyle Korver as the GOAT but I never try to fake serious discussion about it.

ShawnKemp96 wrote:Infact he made a lot more steals than the statisticians think.
User avatar
kendogg
Sixth Man
Posts: 1,782
And1: 203
Joined: Apr 08, 2001
Location: Cincinnati

Re: Magic Johnson vs Kevin Garnett 

Post#60 » by kendogg » Sun Aug 11, 2019 2:08 pm

Zeitgeister wrote:RAPM does not use box score data, you may be thinking of RPM.


Both are actually regressive analysis based on box-score data. Which is why the stat doesn't really work in the playoffs because there isn't enough data.

BPM is the most simplistic one that is basically just straight off box score data (and usage)

Return to Player Comparisons