+/- is so flawed

[lebomb]

Aldridge had a great game 25pts and 11 rebounds, yet he had a -8. Pau scored 6pts and had 12 rebounds and he has a +17? WTF is that bullshit. The +/- doesnt really tell the whole story at all. Lonzo Ball can have a shit game with 3-12 fg and 0-7 3pt and have a +5. :lmao

They need to drop this +/- crap. Its worthless.

[K...]

Thank you, but mat Bonner is still getting his jersey retired, nothing you can do about that

[Raven]

+- can't be flawed, the flowed bit is the analysis based on it.

[tholdren]

+- can't be flawed, the flowed bit is the analysis based on it.

Its flawd how you and virtually everyone else uses it. It is not an indicator of INDIVIDUAL perfomrance

[Raven]

Its flawd how you and virtually everyone else uses it. It is not an indicator of INDIVIDUAL perfomrance

I don't you use it as such. It is a fact, nothing more, nothing less. If you want to make analysis with it you need to have a very big sample size. But even then it doesn't tell X is better than Y, it tells X has had a positive/negative impact on team performance in it's current usage.

[tholdren]

I don't you use it as such. It is a fact, nothing more, nothing less. If you want to make analysis with it you need to have a very big sample size. But even then it doesn't tell X is better than Y, it tells X has had a positive/negative impact on team performance in it's current usage.

It doesnt even do what you said. So youve proven you use it incorrectly. U dumb

[MultiTroll]

Aldridge had a great game 25pts and 11 rebounds, yet he had a -8. Pau scored 6pts and had 12 rebounds and he has a +17? WTF is that bullshit. The +/- doesnt really tell the whole story at all. Lonzo Ball can have a shit game with 3-12 fg and 0-7 3pt and have a +5. :lmao

They need to drop this +/- crap. Its worthless.
lebomb I did not book mark it, but I recall an NBA game where someone went like 15-20 and 90% of FTs to go with 2 steals and no turnovers and had a large -

[MVPCues]

All things being equal, each player contributes to 20% of his own +/- box score. All things aren't equal, of course. Key players with the ball in their hands more will contribute a higher % to their own +/-. Guys who don't handle the ball a lot and are hid on defense will contribute less. It's a valuable statistic that is telling only over the long haul. Each individual +/- in a game can be a huge anomaly. If a very good player plays on a terrible team his entire career, his career +/- will certainly suffer. Think someone like Archie Manning, if I can switch sports for a comparison. Played his heart out, great quarterback, got beat to hell game after game for years. If a mediocre player plays on a dynasty team or 3 his entire career, his career +/- will be a little inflated, but not totally inaccurate.

The top 13 NBA box score +/- career leaders. I went to 13 for very obvious reasons.



1.
LeBron James (https://www.basketball-reference.com/players/j/jamesle01.html)
9.19


2.
Michael Jordan (https://www.basketball-reference.com/players/j/jordami01.html)*
8.09


3.
Chris Paul (https://www.basketball-reference.com/players/p/paulch01.html)
7.56


4.
Charles Barkley (https://www.basketball-reference.com/players/b/barklch01.html)*
7.44


5.
David Robinson (https://www.basketball-reference.com/players/r/robinda01.html)*
7.35


6.
Magic Johnson (https://www.basketball-reference.com/players/j/johnsma02.html)*
7.22


7.
Larry Bird (https://www.basketball-reference.com/players/b/birdla01.html)*
7.17


8.
Russell Westbrook (https://www.basketball-reference.com/players/w/westbru01.html)
6.60


9.
Stephen Curry (https://www.basketball-reference.com/players/c/curryst01.html)
6.50


10.
James Harden (https://www.basketball-reference.com/players/h/hardeja01.html)
6.49


11.
Clyde Drexler (https://www.basketball-reference.com/players/d/drexlcl01.html)*
5.98


12.
Kareem Abdul-Jabbar (https://www.basketball-reference.com/players/a/abdulka01.html)*
5.77


13.
Tim Duncan (https://www.basketball-reference.com/players/d/duncati01.html)
5.47



Huh. How flawed is a career +/- statistic? It's flawed in each and every individual game analysis, and assumptions from said analysis would be hit or miss. As the sample size gets larger, the statistic becomes less flawed.

[lebomb]

lebomb I did not book mark it, but I recall an NBA game where someone went like 15-20 and 90% of FTs to go with 2 steals and no turnovers and had a large -


EXACTLY!!!!! You can have a killer game contributing your ass off, and be on the floor with a scrub that fucks it all up. +/- is really dumb. Just ask Lebron.

[MaNu4Tres]

Aldridge had a great game 25pts and 11 rebounds, yet he had a -8. Pau scored 6pts and had 12 rebounds and he has a +17? WTF is that bullshit. The +/- doesnt really tell the whole story at all. Lonzo Ball can have a shit game with 3-12 fg and 0-7 3pt and have a +5. :lmao

They need to drop this +/- crap. Its worthless.

Not in the long run -- especially when competition is equal.

[TheGreatYacht]

There's nothing worse than BPM. I use it sometimes just because the analytical virgins will reply to me with it if I don't. Only gullible idiots that believe in religion use that piece of fossilized dog shit.

BPM
Kyle Anderson: 3.9
Pau Gasol: 3.3
Lamarcus Aldridge: 3.1

:lol

[Raven]

It doesnt even do what you said. So youve proven you use it incorrectly. U dumb

I love your definition of "proven" :lmao

[tholdren]

I love your definition of "proven" :lmao

Deflection typical

[Chinook]

There's nothing worse than BPM. I use it sometimes just because the analytical virgins will reply to me with it if I don't. Only gullible idiots that believe in religion use that piece of fossilized dog shit.

BPM
Kyle Anderson: 3.9
Pau Gasol: 3.3
Lamarcus Aldridge: 3.1

:lol

BPM is actually a newer stat (like much newer than RAPM and about the same age as ESPN's RPM). It's not regular plus-minus, like what this thread is about.

[tholdren]

BPM is actually a newer stat (like much newer than RAPM and about the same age as ESPN's RPM). It's not regular plus-minus, like what this thread is about.

Same stupidity with different packaging

[Chinook]

Aldridge had a great game 25pts and 11 rebounds, yet he had a -8. Pau scored 6pts and had 12 rebounds and he has a +17? WTF is that bullshit. The +/- doesnt really tell the whole story at all. Lonzo Ball can have a shit game with 3-12 fg and 0-7 3pt and have a +5. :lmao

They need to drop this +/- crap. Its worthless.

The bolded is a flawed assumption. Plus-minus itself is just a fact. Acting like it's supposed to "tell the whole story" is like saying FG% is supposed to tell the whole story. It's just a dumb assumption on your part and has nothing to do with the general viability of the metric.

The Spurs outscored the Thunder by a lot when Pau was on the court and were outscored when LMA was on the court. That's true no matter which one played better. There are good explanations for it, like Pau being the only big when OKC had none of their Big Three on the court while LMA had to face OKC's starters or the team happening to hit a higher percentage of their open shots when Pau was on the floor. It's a fine critique to cite those reasons. On the aggregate, though, those things will even out.

Aldridge should be expected to have a decent plus-minus in relation to the team's margin of victory. He should not have the highest on the team (he plays too many minutes), but he should be above-average. So far, that's where he's been all year.

[Chinook]

Same stupidity with different packaging

Not really. If anything, it would be "different stupidities with the same packaging". Numerically and philosophically, they're as different as can be.

[tholdren]

Not really. If anything, it would be "different stupidities with the same packaging". Numerically and philosophically, they're as different as can be.

BPM takes box score stats from an individual and team level and tries estimate player performance relative to NBA average

[Chinook]

BPM takes box score stats from an individual and team level and tries estimate player performance relative to NBA average

Yes, and general plus-minus is just a record of the point differential of a given player when that player is on the floor. The goal of BPM is to look like general plus-minus, but they are extremely different in how they are calculated and what type of arguments they are trying to make.

[tholdren]

Not really. If anything, it would be "different stupidities with the same packaging". Numerically and philosophically, they're as different as can be.
the name suggests, real plus-minus shares a family resemblance with the +/- stat in*the box score, which merely registers the net change in score (plus or minus) while each player is on the court.

RPM is inspired by the same underlying +/- logic: If a team outscores its opponents when a player is on the court, he's probably doing something to help his team, whether or not he's putting up big numbers.

[tholdren]

Yes, and general plus-minus is just a record of the point differential of a given player when that player is on the floor. The goal of BPM is to look like general plus-minus, but they are extremely different in how they are calculated and what type of arguments they are trying to make.

Trying to estimate the value of individual players perfromance through averages and net score is philosophically the same, correct?

[Chinook]

the name suggests, real plus-minus shares a family resemblance with the +/- stat in*the box score, which merely registers the net change in score (plus or minus) while each player is on the court.

RPM is inspired by the same underlying +/- logic: If a team outscores its opponents when a player is on the court, he's probably doing something to help his team, whether or not he's putting up big numbers.


Trying to estimate the value of individual players perfromance through averages and net score is philosophically the same, correct?

RAPM, RPM, BPM and a number of other stats like them are derivative metrics. They aren't trying to estimate general plus-minus; they are trying to "correct" it. Essentially, someone like the OP decided plus-minus was no informative enough and created a metric that would hopefully align better with our subjective ideas of who the best players were. That additional value argument is what makes those stats philosophically different from general plus-minus. The numerical different comes from the fact that general plus minus is an exact piece of data, while the other stats are estimates.

To put it another way, if in a game Curry shot 3/10 from three while Patty shot 4/6, the box score would say Curry's 3P% was .300 while Mills' was .667. No one in their right mind would say, "Man 3P% is so flawed. Curry's a way better shooter than Mills is." No one would even think about assigning a value argument or a value component to a simple descriptive metric. For some reason, however, people make this confusion with plus-minus, even though it's philosophically and numerically akin to 3P%. Neither stat is trying to 'say' anything. They are just factual records of what happened.

BPM and the like are akin to PER, win-shares and individual ratings. They are not merely records of what happened, they are mathematical arguments created by a statistician, assigning values based on how much weight the creator felt each number they get should have. They are not objective and can be wrong in a way regular stats can't be. In that way, RPM saying something like LMA is barely a net positive can be criticized for being wrong. It doesn't conform to reality, and unlike the regular plus-minus, it doesn't carry the inherent power to truth. The goal of those stats is to get a ranking of players, with the best (/ most impactful) players at the top. Therefore, it failing to do that would call the whole metric into question.

[rjv]

i always think of bruce bowen when it came to the +/- metric

[lebomb]

The bolded is a flawed assumption. Plus-minus itself is just a fact. Acting like it's supposed to "tell the whole story" is like saying FG% is supposed to tell the whole story. It's just a dumb assumption on your part and has nothing to do with the general viability of the metric.

The Spurs outscored the Thunder by a lot when Pau was on the court and were outscored when LMA was on the court. That's true no matter which one played better. There are good explanations for it, like Pau being the only big when OKC had none of their Big Three on the court while LMA had to face OKC's starters or the team happening to hit a higher percentage of their open shots when Pau was on the floor. It's a fine critique to cite those reasons. On the aggregate, though, those things will even out.

Aldridge should be expected to have a decent plus-minus in relation to the team's margin of victory. He should not have the highest on the team (he plays too many minutes), but he should be above-average. So far, that's where he's been all year.

Your explanation is exactly what makes it stupid. Who even cares. Especially when Pau is on the floor and their midgets are in.

[tholdren]

RAPM, RPM, BPM and a number of other stats like them are derivative metrics. They aren't trying to estimate general plus-minus; they are trying to "correct" it. Essentially, someone like the OP decided plus-minus was no informative enough and created a metric that would hopefully align better with our subjective ideas of who the best players were. That additional value argument is what makes those stats philosophically different

Same family of thinking, philosophocally, as you just stated, they are trying to adjust plus minus. You wouldnt adjust anything that wasnt on similar philosphical planes. Trying to judge individual performance on a net value that takes 10 players minimum to contribute to is stupid. Which i think, after years of debate, you just confirmed

[Raven]

Same family of thinking, philosophocally, as you just stated, they are trying to adjust plus minus. You wouldnt adjust anything that wasnt on similar philosphical planes. Trying to judge individual performance on a net value that takes 10 players minimum to contribute to is stupid. Which i think, after years of debate, you just confirmed

... you just don't get it.

[MVPCues]

BPM is actually a newer stat (like much newer than RAPM and about the same age as ESPN's RPM). It's not regular plus-minus, like what this thread is about.

I didn't realize that. Thanks for the correction on what I posted.

[tholdren]

... you just don't get it.

Says the guy who claims plus minus tells who had an effect... lol dumb

[Chinook]

Your explanation is exactly what makes it stupid. Who even cares. Especially when Pau is on the floor and their midgets are in.

But that's not any flaw in the stat, though. It's just a record

[Chinook]

Same family of thinking, philosophocally, as you just stated, they are trying to adjust plus minus. You wouldnt adjust anything that wasnt on similar philosphical planes. Trying to judge individual performance on a net value that takes 10 players minimum to contribute to is stupid. Which i think, after years of debate, you just confirmed

No. They are very different. Plus-minus is an objective record of what happened during a game. The other stats are arguments about which players are best. They try to use the same simple format of plus-minus to make their point, but they aren't the same. I'd actually go so far as to say that of that list, only RAPM actually tries to be a statement on net scoring. It's actually very different from BPM and RPM numerically, but I lumped them together for convenience. Anyway, none of those stats are just tweaks to the general plus-minus format. They are rewrites of the whole formula that go far beyond the scope of the original metric. I haven't confirmed, but I don't think those stats really even look like general plus-minus when compared side by side.

As far as your last point, I have never really defended using plus-minus for an individual game. At best, it can just shed some light onto a guy's performance leading to runs from one side or the other. It's not ironclad. However, the great thing about stats is that they get stronger the more data that gets collected. That's why I said LMA's total plus-minus is actually where you'd expect it to be and is thusly more representative of his overall impact. A lot of things come out in the wash.

[tholdren]

No. They are very different. Plus-minus is an objective record of what happened during a game. The other stats are arguments about which players are best. They try to use the same simple format of plus-minus to make their point, but they aren't the same. I'd actually go so far as to say that of that list, only RAPM actually tries to be a statement on net scoring. It's actually very different from BPM and RPM numerically, but I lumped them together for convenience. Anyway, none of those stats are just tweaks to the general plus-minus format. They are rewrites of the whole formula that go far beyond the scope of the original metric. I haven't confirmed, but I don't think those stats really even look like general plus-minus when compared side by side.

I copy and pasted below from the site that explains what the differences are between the two.

the name suggests, real plus-minus shares a family resemblance with the +/- stat in*the box score, which merely registers the net change in score (plus or minus) while each player is on the court.

RPM is inspired by the same underlying +/- logic: If a team outscores its opponents when a player is on the court, he's probably doing something to help his team, whether or not he's putting up big numbers

[Ed Helicopter Jones]

It's not a good single game metric to evaluate a particular player, but when used with other data, like starters v. starters, the impact of certain lineups, long-term analysis, I think it can be useful. Just like any piece of information it can become horribly skewed in the wrong hands.

But, yeah, a scrub who plays in garbage time on a particular night can look amazing in the +/-, vs. a guy who carried the load all night and breaks even because he's busting it against the other team's all-stars.

[Chinook]

I copy and pasted below from the site that explains what the differences are between the two.

the name suggests, real plus-minus shares a family resemblance with the +/- stat in*the box score, which merely registers the net change in score (plus or minus) while each player is on the court.

RPM is inspired by the same underlying +/- logic: If a team outscores its opponents when a player is on the court, he's probably doing something to help his team, whether or not he's putting up big numbers

Plus-minus doesn't have "logic" though, any more than offensive rebounds do. That second sentence does about as good of a job at explaining the difference as anything I've read. The bolded part is exactly why the two metrics are philosophically incompatible.

[tholdren]

Plus-minus doesn't have "logic" though, any more than offensive rebounds do. That second sentence does about as good of a job at explaining the difference as anything I've read. The bolded part is exactly why the two metrics are philosophically incompatible.

Totally disagree with the bold. Its a terrible assumption.

[Chinook]

Totally disagree with the bold. Its a terrible assumption.

And you can if you want. I'm just saying that regular plus-minus doesn't make that assumption. It doesn't make any assumptions.

[SASdynasty!]

Lol the stat that proved Boban was our best player. And these “advanced” stats guys still use it, lol.

[tholdren]

And you can if you want. I'm just saying that regular plus-minus doesn't make that assumption. It doesn't make any assumptions.

Lol dancing. Look you said the stats werent related. I showed you they are, from the creators nonetheless. You say the stats dont make assumptions, technically you are correct but you know very well that stats are not only given to identify a point in time but to predict, rank, and sort. Hence the media hype bullshit with unintelligent fans. Arguing the semantics is like arguing the stats themselves, its for lazy, unintelligent people who dont understand the game. Watch the game and its easy to see the stats negate the actual detail of play.

The one thing you cant dance around is that plus minus is not a stat that can determine individual performance. Which you argued at length was not the case several years ago

[Raven]

Totally disagree with the bold. Its a terrible assumption.

it's not an assumption, it's a fact.

[Chinook]

Lol dancing. Look you said the stats werent related. I showed you they are, from the creators nonetheless.

I actually said they were different things packaged the same way. The quote from the "creators" said the same thing. RPM is a stat that's made to look like regular plus-minus. But it's completely different in pretty much every possible way. I have no problem saying that the quote is wrong in its interpretation of both stats.


You say the stats dont make assumptions,

No. I said regulars, general plus-minus doesn't make assumptions. I specifically said RPM and BPM DO make assumptions, which is why they are philosophically different from regular plus-minus.


stats are not only given to identify a point in time but to predict, rank, and sort

No. I'd go as far as to say no raw stats predict, rank or sort. Doing that comes from models, like RPM and its ilk. There are a ton of rules about the procedure and risk of creating models with data and the power of conclusions drawn from those. Again, the fundamental philosophical difference between RPM and plus-minus is that the former cares that there's a rhyme or reason to its results, while the latter doesn't.


Arguing the semantics is like arguing the stats themselves, its for lazy, unintelligent people who dont understand the game. Watch the game and its easy to see the stats negate the actual detail of play.

This isn't just a "semantic" argument. A person can't consistently misuse evidence and them accuse the evidence of being wrong. Again, no one seems to have this problem with the other raw stats like points, or turnovers. People understand those capture what they capture and aren't supposed to explain the "detail of play". But for plus-minus, people turn stupid, both for an against it.


The one thing you cant dance around is that plus minus is not a stat that can determine individual performance

NO stat can "determine" that. Arguing it can or bashing it because it can't are equally pointless. Stats are merely evidence for an argument, not the argument themselves. To that end, plus-minus CAN be used as evidence for individual value, and with a large enough sample size, it tracks reasonably well in that regard. Even when stats are backed up by large samples, it's not advisable to use them as end-all/be-alls in an argument. Keeping them in context is important.

For example, Matt Bonner's great plus-minus numbers are not an example of the stat being flawed. But they also aren't evidence for how good Bonner was compared to better players. Rather, they fit nicely in an argument of how stretch-bigs changed the NBA with Bonner being one of the the first role-player bigs who shot threes at a high level.

[Chinook]

it's not an assumption, it's a fact.

It's not a fact; it's a hypothesis, meaning it's pretty much an assumption. But depending on the sample size, it could be a really strong assumption.

[Raven]

I actually said they were different things packaged the same way. The quote from the "creators" said the same thing. RPM is a stat that's made to look like regular plus-minus. But it's completely different in pretty much every possible way. I have no problem saying that the quote is wrong in its interpretation of both stats.



No. I said regulars, general plus-minus doesn't make assumptions. I specifically said RPM and BPM DO make assumptions, which is why they are philosophically different from regular plus-minus.



No. I'd go as far as to say no raw stats predict, rank or sort. Doing that comes from models, like RPM and its ilk. There are a ton of rules about the procedure and risk of creating models with data and the power of conclusions drawn from those. Again, the fundamental philosophical difference between RPM and plus-minus is that the former cares that there's a rhyme or reason to its results, while the latter doesn't.



This isn't just a "semantic" argument. A person can't consistently misuse evidence and them accuse the evidence of being wrong. Again, no one seems to have this problem with the other raw stats like points, or turnovers. People understand those capture what they capture and aren't supposed to explain the "detail of play". But for plus-minus, people turn stupid, both for an against it.



NO stat can "determine" that. Arguing it can or bashing it because it can't are equally pointless. Stats are merely evidence for an argument, not the argument themselves. To that end, plus-minus CAN be used as evidence for individual value, and with a large enough sample size, it tracks reasonably well in that regard. Even when stats are backed up by large samples, it's not advisable to use them as end-all/be-alls in an argument. Keeping them in context is important.

For example, Matt Bonner's great plus-minus numbers are not an example of the stat being flawed. But they also aren't evidence for how good Bonner was compared to better players. Rather, they fit nicely in an argument of how stretch-bigs changed the NBA with Bonner being one of the the first role-player bigs who shot threes at a high level.

AMEN.

[Raven]

It's not a fact; it's a hypothesis, meaning it's pretty much an assumption. But depending on the sample size, it could be a really strong assumption.

i meant the quoted message, not the bold quoted in the msg. But yeah..

[tholdren]

i meant the quoted message, not the bold quoted in the msg. But yeah..

Lol u so dumb

[tholdren]

I actually said they were different things packaged the same way. The quote from the "creators" said the same thing. RPM is a stat that's made to look like regular plus-minus. But it's completely different in pretty much every possible way. I have no problem saying that the quote is wrong in its interpretation of both stats.



No. I said regulars, general plus-minus doesn't make assumptions. I specifically said RPM and BPM DO make assumptions, which is why they are philosophically different from regular plus-minus.



No. I'd go as far as to say no raw stats predict, rank or sort. Doing that comes from models, like RPM and its ilk. There are a ton of rules about the procedure and risk of creating models with data and the power of conclusions drawn from those. Again, the fundamental philosophical difference between RPM and plus-minus is that the former cares that there's a rhyme or reason to its results, while the latter doesn't.



This isn't just a "semantic" argument. A person can't consistently misuse evidence and them accuse the evidence of being wrong. Again, no one seems to have this problem with the other raw stats like points, or turnovers. People understand those capture what they capture and aren't supposed to explain the "detail of play". But for plus-minus, people turn stupid, both for an against it.



NO stat can "determine" that. Arguing it can or bashing it because it can't are equally pointless. Stats are merely evidence for an argument, not the argument themselves. To that end, plus-minus CAN be used as evidence for individual value, and with a large enough sample size, it tracks reasonably well in that regard. Even when stats are backed up by large samples, it's not advisable to use them as end-all/be-alls in an argument. Keeping them in context is important.

For example, Matt Bonner's great plus-minus numbers are not an example of the stat being flawed. But they also aren't evidence for how good Bonner was compared to better players. Rather, they fit nicely in an argument of how stretch-bigs changed the NBA with Bonner being one of the the first role-player bigs who shot threes at a high level.

No it cannot be used in any sample size to determine individual worth.