Letter FOR THOSE WHO USE +/- as evidence

[tholdren]

Dear Fans with Learning Disabilities in Math,

Please note that Plus Minus, has ZERO correlation to individual performance or player.

Example:

Player A is on the court for 20 minutes. In that time player A has 6 turnovers, 0/7, 0 assists, 0 rebounds, 0 blocks. During that 20 minutes the man PLAYER A is defending is the only player to score. Player A gives up 20 points during the time he was in the game. During the same time his other teammates score 30.

PLAYER A WOULD HAVE A +/- OF = +10
...........................................
Example:

Player B is on the court for 20 minutes. In that time player B has 0 Turnovers 10/10, 6 steals, 8 rebounds, and 3 blocks. During that 20 minutes the man that Player B is defending does not score, get an assist, or a rebound. Unfortunately for Player B during that 20 minutes his team was outscored 20-30.

PLAYER B WOULD HAVE A +/- OF = -10
-------------------------------------

Some of you would look at the plus minus and claim that Player A had the better game, but as you can see, this would not be the case. Plus minus has absolutely no correlation to a SINGLE PLAYER. I have no idea why/how anyone with minimal knowledge of statistics could misconstrue this statistic. The only thing that this stat would possibly do is generate questions.

STOP USING PLUS MINUS AS SOME TYPE OF EVIDENCE.

Thank You,

Mathematicians

[SpursFan86]

So it's stupid to use regular +/- to judge a player's performance in a single game? Shocker :wow

Luckily we have adjusted +/- stats that look at the entire season. Obviously still has flaws, but certainly worth looking at.

[tholdren]

So it's stupid to use regular +/- to judge a player's performance in a single game? Shocker :wow

Luckily we have adjusted +/- stats that look at the entire season. Obviously still has flaws, but certainly worth looking at.

I can help you understand the premise behind stats if you want.

[Chinook]

Ugh. Plus-minus isn't misleading. People choose to apply it incorrectly. It reflects reality and in large sample sizes it's pretty sound. But it only says what it says. It doesn't draw conclusions.

But Jimmer sucked subjectively, too. He earned his poor plus-minus score, just like he earned his awful net rating.

[tholdren]

Ugh. Plus-minus isn't misleading. People choose to apply it incorrectly. It reflects reality and in large sample sizes it's pretty sound.

But Jimmer sucked subjectively, too. He earned his poor plus-minus score, just like he earned his awful net rating.

NO, it is.

[Chinook]

NO, it is.It's not. That's like saying PPG is misleading because of shooting percentage. It's a basic stat that has no bias. The fact that you don't like it is completely irrelevant.

[tholdren]

It's not. That's like saying PPG is misleading because of shooting percentage. It's a basic stat that has no bias. The fact that you don't like it is completely irrelevant.

No, it's not like that at all. And what a terrible analogy. The fact remains that you cannot attribute any single fact from trash data that involves AT A MINIMUM 10 players at completely separate times, conditions, and other variables, then try to formulate what impact that player had. Even real plus minus is flawed as it is basically an estimate of an average of an estimate. And that is "REAL PLUS MINUS" not this trash at the end of box score. To use this, and state that it isn't misleading is A) ignorant and B) a lie.

[SuperCam]

:cry b-b-but Kiwi is RPM mvp :cry

[Chinook]

No, it's not like that at all. And what a terrible analogy. The fact remains that you cannot attribute any single fact from trash data that involves AT A MINIMUM 10 players at completely separate times, conditions, and other variables, then try to formulate what impact that player had. Even real plus minus is flawed as it is basically an estimate of an average of an estimate. And that is "REAL PLUS MINUS" not this trash at the end of box score. To use this, and state that it isn't misleading is A) ignorant and B) a lie.

You're mixing up misleading and misusable. Plus-minus is literally the net scoring of the teams when a player is on the floor. That is not debate able. It's not biased. However, by itself, it doesn't say how good a player is. But that stat isn't called "player value rating" or anything like that. It doesn't imply a player's worth. People imply that. But that's their fault, not the stat's.

That being said, you're arguing for noise affecting the numbers. Ignoring everything I just said for the sake of discussion, most noise goes away after you get a big enough sample size. For example, Bonner literally was a valuable asset to outscoring opponents in his prime. It's just irresponsible to dismiss it after seeing it over many thousands of minutes.

Plus-minus is measured, not determined. It reflects reality. It can only be flawed if someone measures it incorrectly. Keep it in its context, and you're golden. If you don't it loses power, but that's true of every stat.

[K...]

Plus minus alone is a great tool to detect hidden performance. It can mean an average player is reaping the benefits of a better player, or that a player has non statistic superlatives. Like boban....modest points and rebounds but good plus score, and his shot altering size is the reason. There is no "made the sg put up weak shit in fear"stat. There is no "boxed out the best rebounder" stat. Just like Tony Parker never got credit for assists when we swung it around or Duncan took down low. Doesn't mean Parker was useless.



And that's also why I think people love Kawhi but not green. Green's defensive hustle doesn't result in stats (although his blocks are usually good). Meanwhile Kawhi gets blocks and steals. But Kawhi isn't any more important than green. (Kawhi is special b/c of his size, if there weren't other 3&d sg green would get the max too).

[tholdren]

You're mixing up misleading and misusable. Plus-minus is literally the net scoring of the teams when a player is on the floor. That is not debate able. It's not biased. However, by itself, it doesn't say how good a player is. But that stat isn't called "player value rating" or anything like that. It doesn't imply a player's worth. People imply that. But that's their fault, not the stat's.

That being said, you're arguing for noise affecting the numbers. Ignoring everything I just said for the sake of discussion, most noise goes away after you get a big enough sample size. For example, Bonner literally was a valuable asset to outscoring opponents in his prime. It's just irresponsible to dismiss it after seeing it over many thousands of minutes.

Plus-minus is measured, not determined. It reflects reality. It can only be flawed if someone measures it incorrectly. Keep it in its context, and you're golden. If you don't it loses power, but that's true of every stat.

Im not mixing anything up. I stated plus minus cannot be used as evidence to state individual player performance. Which you just admitted to above.

Second, no sample size can ever take "noise" completely from stats based on the definition of stats itself.

Plus minus is measured, but as you state, cannot be used to determine individual performance.

I am uncertain why you even felt compelled to respond that it isn't misleading when many posters on here, when reflecting on performance, try to use this stat as a way to suggest one player performed better than another. Ironically, NBA/ESPN/etc. Throw plus minus in the box score, like it is to be used as some easily transferrable/analyzed piece of data.

[RD2191]

Plus minus alone is a great tool to detect hidden performance. It can mean an average player is reaping the benefits of a better player, or that a player has non statistic superlatives. Like boban....modest points and rebounds but good plus score, and his shot altering size is the reason. There is no "made the sg put up weak shit in fear"stat. There is no "boxed out the best rebounder" stat. Just like Tony Parker never got credit for assists when we swung it around or Duncan took down low. Doesn't mean Parker was useless.



And that's also why I think people love Kawhi but not green. Green's defensive hustle doesn't result in stats (although his blocks are usually good). Meanwhile Kawhi gets blocks and steals. But Kawhi isn't any more important than green. (Kawhi is special b/c of his size, if there weren't other 3&d sg green would get the max too).

Lol. Faggot.

[tholdren]

Plus minus alone is a great tool to detect hidden performance. It can mean an average player is reaping the benefits of a better player, or that a player has non statistic superlatives. Like boban....modest points and rebounds but good plus score, and his shot altering size is the reason. There is no "made the sg put up weak shit in fear"stat. There is no "boxed out the best rebounder" stat. Just like Tony Parker never got credit for assists when we swung it around or Duncan took down low. Doesn't mean Parker was useless.



And that's also why I think people love Kawhi but not green. Green's defensive hustle doesn't result in stats (although his blocks are usually good). Meanwhile Kawhi gets blocks and steals. But Kawhi isn't any more important than green. (Kawhi is special b/c of his size, if there weren't other 3&d sg green would get the max too).

And here is another terrible explanation/use of plus minus. Shot alteration has nothing to do with plus minus.
Boban alters a shot and opponent scores off rebound, boban is still -2.

[Kawhitstorm]

Score plus-minus is the most useless stat in sports but all the Kawhi haters cling on to it b/c their retarded overlord tweeted a bunch of on/off stats based on the score plus/minus thinking he was doing some high level analytical work. :lmao

[ElNono]

It's the net scoring of teams when different lineups that include the player are on the floor. It's a lineup-based stat that has it's uses, especially when lineups are fairly consistent (which is generally the case in basketball, with the Spurs perhaps being an outlier).

Interestingly, it's value is actually diminished as a lineup stat when the lineups change a lot, but conversely, when that happens, it's value as a gauge of individual performance increases.

The nice thing about it as a lineup stat is that it can capture things like "good chemistry", which are often lost in stats based on pure raw individual numbers.

I think it's a useful stat, especially when analyzing player combinations. But like any other, it's just one more tool in the toolbox, and paints a partial picture.

[HarlemHeat37]

:lol there isn't a poster on SpursTalk that uses stats more incorrectly than tholdren, tbh..

This is the same poster that cited Rasual Butler's playoff PER from 7 minutes of total playing time as an argument against him, just last week..

[K...]

And here is another terrible explanation/use of plus minus. Shot alteration has nothing to do with plus minus.
Boban alters a shot and opponent scores off rebound, boban is still -2.

What is this? Fantasy and fact? Boban scored 1-2 fg and had 1 rebound. His plus minus score for the game is higher, why? Shot alteration. This is an example where plus minus plus the eye test gives useful info.


Lmao of course if boban can't get rebounds his plus minus won't be great lol.


Lmao letter on the internet. Lmao hidden player fan agenda. Lmao not understanding stats?








What's the point of this thread anyway? Did plus minus hurt you recently?

[Raven]

a mathematician would never say something so retarded.

[Uriel]

Of course raw +/- is flawed. But I've never seen anyone use raw +/- numbers to measure the worth of a player, especially when the sample size is small.

That's why we have real plus-minus. Of course, that isn't perfect either. But it isn't nearly susceptible to the same weaknesses you mentioned in your OP, since by definition, it works to extricate the impact of the other nine players on the floor and isolate the impact of the individual player in question.

You can argue that real plus-minus is "an estimate of an average of an estimate." But you can't argue that it suffers from the same vulnerabilities you pointed out in your OP.

[tholdren]

What is this? Fantasy and fact? Boban scored 1-2 fg and had 1 rebound. His plus minus score for the game is higher, why? Shot alteration. This is an example where plus minus plus the eye test gives useful info.


Lmao of course if boban can't get rebounds his plus minus won't be great lol.





Lmao letter on the internet. Lmao hidden player fan agenda. Lmao not understanding stats?








What's the point of this thread anyway? Did plus minus hurt you recently?

Im sorry you don't understand the insanely complicated formula that is plus minus.... or that shot alteration has nothing to do with it.

[tholdren]

Of course raw +/- is flawed. But I've never seen anyone use raw +/- numbers to measure the worth of a player, especially when the sample size is small.

That's why we have real plus-minus. Of course, that isn't perfect either. But it isn't nearly susceptible to the same weaknesses you mentioned in your OP, since by definition, it works to extricate the impact of the other nine players on the floor and isolate the impact of the individual player in question.

You can argue that real plus-minus is "an estimate of an average of an estimate." But you can't argue that it suffers from the same vulnerabilities you pointed out in your OP.

You see it on here after every single game

[tholdren]

Simmons had the highest +/- of the entire team at 18. He had 5 rebounds and 4 assists and 1 steal. Those are stats of someone who knows what he's doing. Sure he had a terrible shooting night, but he made up in all other things.

+/0 of West was -14 LMA was -21, Aldridge has Zero assists... that's a sign of a guy that is totally lost!!

I think Spurs might have acquired a very expensive catch and shoot player.
------------------

[All Mighty Janitor]

I don't like to use +/- because it's only useful across a whole season and is biased toward players on winning teams. It's only useful for comparing players on the same team (and sometimes not even that.) There are better stats to use like the adjusted +/-'s.

[Chinook]

Im not mixing anything up. I stated plus minus cannot be used as evidence to state individual player performance. Which you just admitted to above.

But that's not a flaw of the stat. It doesn't claim to do that. PPG doesn't claim say who's the best scorer, especially not based on one game. The problem is people who misuse the stat. The stat itself is just a raw number.


Second, no sample size can ever take "noise" completely from stats based on the definition of stats itself.

No. That's why we have confidence intervals. However, stats, especially counting stats, are NOT subject to noise save for recorder bias. It's the models that are subjective and have noise. Have you done statistical modeling before? You'd know the difference.


Plus minus is measured, but as you state, cannot be used to determine individual performance.

Yes. By itself, it does not determine individual performance. But you can totally go, "I can see why Jimmer was -12 last game. Dude sucked balls and made everyone worse." Or you can go, "It sucks for Jimmer that he was -12 last night. He got really unlucky by being on the court for SAC making seven threes in a three-minute span." The first actually happened; the second is fabricated. But neither would be exposing a flaw in plus-minus. The stat reflected the truth of the game independent of subjective interpretation. Now if Jimmer had 3000 minutes of -12 games, you have to start wondering what's going on. No one is THAT unlucky.


I am uncertain why you even felt compelled to respond that it isn't misleading when many posters on here, when reflecting on performance, try to use this stat as a way to suggest one player performed better than another. Ironically, NBA/ESPN/etc. Throw plus minus in the box score, like it is to be used as some easily transferrable/analyzed piece of data.

That's what I'm saying. The stat is misused, but it's not misleading. The fact that some people don't understand it doesn't make the stat fallable. Some people think Kobe is better than Duncan based on PPG. But that doesn't make PPG a flawed stat. It just makes those people ignorant. It's probably semantics more than anything.

I don't think your stance is bad; I just think it's a little off-target. Stats are often misunderstood and misapplied. People who don't know the philosophy and formulae just throw the numbers out and make sweeping judgments. Some use descriptive stats in a prescriptive manner. Those are people being ignorant of who stats work and the difference between the stats and the interpretation of stats. Stats are just a collection of data that are possibly churned through a formula. They are what they are. And if you keep that in mind, they're useful.

[Brazil]

+/- is only good kawhi has a net positive and Parker a net negative... It doesn't apply when kawhi is -35 and Barnes +22 for instance

[Brazil]

Tbh fwiw

[tholdren]

But that's not a flaw of the stat. It doesn't claim to do that. PPG doesn't claim say who's the best scorer, especially not based on one game. The problem is people who misuse the stat. The stat itself is just a raw number.

No, you're correct about this, but the way the stats are presented, are to make people use them incorrectly by nature. Hence, why/how players are recruited = ppg. The stat that the public infers by media representation equates the most talented basketball player as well as the best scorer.



No. That's why we have confidence intervals. However, stats, especially counting stats, are NOT subject to noise save for recorder bias. It's the models that are subjective and have noise. Have you done statistical modeling before? You'd know the difference.

Which is why Adjusted +/- is somewhat better, but in the end an average of an average of an estimate is never accurate. I read/use stats daily. And to claim that they are correct, or not misleading is just not true. The basis behind stats is only to inquire further about the number represented and how it came to be.


Yes. By itself, it does not determine individual performance. But you can totally go, "I can see why Jimmer was -12 last game. Dude sucked balls and made everyone worse." Or you can go, "It sucks for Jimmer that he was -12 last night. He got really unlucky by being on the court for SAC making seven threes in a three-minute span." The first actually happened; the second is fabricated. But neither would be exposing a flaw in plus-minus. The stat reflected the truth of the game independent of subjective interpretation. Now if Jimmer had 3000 minutes of -12 games, you have to start wondering what's going on. No one is THAT unlucky.

LOL and in the first case you would be misusing the stats because it mislead you. -12 doesn't mean he made "everyone worse" it only means he was outscored by 12 when he was on the court. Just like simmons being a +18 didn't mean "he made everyone better" The +18 is misleading because it's the highest on the team, the statement is the misuse.


That's what I'm saying. The stat is misused, but it's not misleading. The fact that some people don't understand it doesn't make the stat fallable. Some people think Kobe is better than Duncan based on PPG. But that doesn't make PPG a flawed stat. It just makes those people ignorant. It's probably semantics more than anything.

It's funny you say this but you were mislead and misused plus minus in your statement about Jimmer. And yes, it is all semantics.

I don't think your stance is bad; I just think it's a little off-target. Stats are often misunderstood and misapplied. People who don't know the philosophy and formulae just throw the numbers out and make sweeping judgments. Some use descriptive stats in a prescriptive manner. Those are people being ignorant of who stats work and the difference between the stats and the interpretation of stats. Stats are just a collection of data that are possibly churned through a formula. They are what they are. And if you keep that in mind, they're useful.

Stats are always misused and misunderstood. However, based on how they are presented, the are often misleading.

[Mikeanaro]

a mathematician would never say something so retarded.

[K...]

Stats are always misused and misunderstood. However, based on how they are presented, the are often misleading.

This is waiving the white flag. Stats are not subject to abuse if you disclose the model and data. In sports we all have the data. The models are usually talked about. Abuse comes when you try to make conclusions about the stats.


Re Kawhi vs clippers. Who the fuck cares still, next season started

[K...]

"Thank You, Mathematicians".....

I'm beginning to think you aren't a mathematician at all.

[Sean Cagney]

Plus minus alone is a great tool to detect hidden performance. It can mean an average player is reaping the benefits of a better player, or that a player has non statistic superlatives.

Yep, biggest bullshit stat people use to point out or make a point for a player like say Bonner, the king of the plus minus. I don't use that +/- crap at all nor entertain it in an debate when talking players. I go by PPG, % and other things that weigh in, plus minus to me is a flat out joke.
I don't like to use +/- because it's only useful across a whole season and is biased toward players on winning teams. It's only useful for comparing players on the same team (and sometimes not even that.) There are better stats to use like the adjusted +/-'s.
I agree.

[Chinook]

Stats are always misused and misunderstood. However, based on how they are presented, the are often misleading.

See, to me, individual DRtg is misleading. It heavily factors in rebounding and is based primarily off team numbers. So it doesn't do a good job at all at isolating perimeter defense, especially by lockdown guys who don't get a ton of rebounds and who play on average defensive teams. There is bias because people choose to weigh certain factors more than others, and it has a name that makes it seem like it's more inclusive than it is. Real plus-minus has a similar flaw. It's not any "realer" than regular plus-minus. In fact, it's manufactured arbitrarily by people who are trying to fudge numbers to meet their expectations. It may or may not correlate well to the eye-test, but it's misleading.


LOL and in the first case you would be misusing the stats because it mislead you. -12 doesn't mean he made "everyone worse" it only means he was outscored by 12 when he was on the court. Just like simmons being a +18 didn't mean "he made everyone better" The +18 is misleading because it's the highest on the team, the statement is the misu

That's not what I said. I agree that it simply means the Spurs were outscored with him on the floor. I'm saying that the explanation as to why he was -12 is because he made everyone worse -- because he did. He was awful. His stats were not a trick of the light. Every time he came in, he was replacing a better player, which made the lineup tank. He messed up spacing, didn't defend well, missed his looks. That's determined by actually looking at the game, not the number. And if SAC had made seven threes in a row, that would have been the explanation of the number. The point I was making is that the -12 was the reality no matter what explanation you give. And if he has enough games with -12, it's going to become hard to create a model that explains what happened without putting the blame on him.


Which is why Adjusted +/- is somewhat better, but in the end an average of an average of an estimate is never accurate. I read/use stats daily. And to claim that they are correct, or not misleading is just not true. The basis behind stats is only to inquire further about the number represented and how it came to be.

I think you're confusing stats with the philosophy of stats. The average in this case is a real number that we can observe. It's not an estimate or approximation, because it's taken out of the entire sample. There's no ideal mean that the average is pretending to be, because there's no hidden variance. You have ALL of the data. If someone asks you how many fingers you have, you're not going to give them a confidence interval. You'll give them an integer (counting parts of fingers as whole to eliminate that loop hole). And if they asked you to give the average number of fingers each person in your house/building/whatever has, you will be able to give a correct number provided you can see everyone's hands. There'd be no variance, save perhaps your own fallability in reporting.

That's the entire point of this debate from my view. Plus-minus is not a statistical model. It's not a sampling. And it's not a suggestion of greatness. It's just a number. And no amount of misuse changes that.

[Mikeanaro]

It shows the impact of one specific player (no matter who) and if he sucked well he sucked why make a fuss over some stat when every data is showing some information, butthurted because numbers doesnt show what you want is sad.
DENIALLLLLLLLLLLLLL

[dabom]

Brazil didn't even know what rpm was one month ago. Lol

[dabom]

By the way, only the people using plus minus are trolling.

[Kidd K]

I don't like to use +/- because it's only useful across a whole season and is biased toward players on winning teams. It's only useful for comparing players on the same team (and sometimes not even that.) There are better stats to use like the adjusted +/-'s.

It's as biased towards winning teams as winning percentage is tbh. If a mediocre player like Bonner has a huge +/-, it means the coach is using him properly. It is one stat to look at; I don't think anyone is arguing is makes all other stats irrelevant.

[Johnsyounger]

This does not apply to Matt Bonner of course....

[~O~]

This is stat that's only useful unless you're witnessing the game live.

[BG_Spurs_Fan]

Stats are always misused and misunderstood. However, based on how they are presented, the are often misleading.

Yes, yes indeed, just check the OP.

[Kawhitstorm]

It shows the impact of one specific player (no matter who) and if he sucked well he sucked why make a fuss over some stat when every data is showing some information, butthurted because numbers doesnt show what you want is sad.
DENIALLLLLLLLLLLLLL

If you possess the analytical skills of an unborn fetus then I guess anything more than a raw data would be over your head.

[Kawhitstorm]

Delete

[Mikeanaro]

If you possess the analytical skills of an unborn fetus then I guess anything more than a raw data would be over your head.
Listen Kawhishitstorm, it seems you dont possess any analytical skill whatsoever and I wont even bother explaining to you stats/intangibles etc, you dont even have a personality using that stupid player nickname so you are the unborn fetus level wise turd.
Sorry but trash like you trying to ditch some stat for no reason doesnt worth a fuck.

[Drachen]

And here is another terrible explanation/use of plus minus. Shot alteration has nothing to do with plus minus.
Boban alters a shot and opponent scores off rebound, boban is still -2.

You're mistaking temperature for climate

[Kawhitstorm]

Listen Kawhishitstorm, it seems you dont possess any analytical skill whatsoever and I wont even bother explaining to you stats/intangibles etc, you dont even have a personality using that stupid player nickname so you are the unborn fetus level wise turd.
Sorry but trash like you trying to ditch some stat for no reason doesnt worth a fuck.

Dude is so furious he's projecting his insecurities by repeating everything I called him out on (nice analytical skills):lmaoSomeone call a wambulance b/c we have a man down w/ a shattered ego. Praying for your recovery!

[Mikeanaro]

Dude is so furious he's projecting his insecurities by repeating everything I called him out on (nice analytical skills):lmaoSomeone call a wambulance b/c we have a man down w/ a shattered ego. Praying for your recovery!
Lol furious how? You are the small brain here and now you are calling for attention because you dont have any fact to back up your ¨intellectual statements¨, you talked about fetus but you didnt bring any substance to the subject, all I see is a bandwagoner that joined ST to make some noise a little late.

[Kawhitstorm]

Lol furious how? You are the small brain here and now you are calling for attention because you dont have any fact to back up your ¨intellectual statements¨, you talked about fetus but you didnt bring any substance to the subject, all I see is a bandwagoner that joined ST to make some noise a little late.

If your undeveloped brain could handle even the basics then you would have been able to comprehend the facts presented in this post instead of throwing a fit. Drink your milk & go to sleep.

[Mikeanaro]

If your undeveloped brain could handle even the basics then you would have been able to comprehend the facts presented in this post instead of throwing a fit. Drink your milk & go to sleep.
Undeveloped brain? it took you 3 hours to formulate this shitty response :lmao

[Killakobe81]

It's the net scoring of teams when different lineups that include the player are on the floor. It's a lineup-based stat that has it's uses, especially when lineups are fairly consistent (which is generally the case in basketball, with the Spurs perhaps being an outlier).

Interestingly, it's value is actually diminished as a lineup stat when the lineups change a lot, but conversely, when that happens, it's value as a gauge of individual performance increases.

The nice thing about it as a lineup stat is that it can capture things like "good chemistry", which are often lost in stats based on pure raw individual numbers.

I think it's a useful stat, especially when analyzing player combinations. But like any other, it's just one more tool in the toolbox, and paints a partial picture.

I'm (mostly) with El here it is good for line-up combos ...shitty for individual metrics.

[YGWHI]

you dont even have a personality using that stupid player nickname
Like those guys who have Duncan21, Tony9, TPark, TimTonyManu...as nicknames?
Sorry, I don't care about the thread discussion but...is there something wrong with having Spurs' players names as nickname that I'm missing? Kawhi's attracting new fans, you should get used to it.

[Mikeanaro]

Like those guys who have Duncan21, Tony9, TPark, TimTonyManu...as nicknames?
Sorry, I don't care about the thread discussion but...is there something wrong with having Spurs' players names as nickname that I'm missing? Kawhi's attracting new fans, you should get used to it.
It only shows they are people with no personality, it is what it is Im sorry, even the most retarded users have personal nicknames.

[tholdren]

It's as biased towards winning teams as winning percentage is tbh. If a mediocre player like Bonner has a huge +/-, it means the coach is using him properly. It is one stat to look at; I don't think anyone is arguing is makes all other stats irrelevant.

no, it may only mean that he is using him at a correct time... i.e. when he is behind in a game and someone on the team is making 3s and the other team is getting fouled making FTs.

could be subbing in and out for defensive purposes on FTS.

ITS A 10 player stat not a 1 player stat.

[tholdren]

See, to me, individual DRtg is misleading. It heavily factors in rebounding and is based primarily off team numbers. So it doesn't do a good job at all at isolating perimeter defense, especially by lockdown guys who don't get a ton of rebounds and who play on average defensive teams. There is bias because people choose to weigh certain factors more than others, and it has a name that makes it seem like it's more inclusive than it is. Real plus-minus has a similar flaw. It's not any "realer" than regular plus-minus. In fact, it's manufactured arbitrarily by people who are trying to fudge numbers to meet their expectations. It may or may not correlate well to the eye-test, but it's misleading.



That's not what I said. I agree that it simply means the Spurs were outscored with him on the floor. I'm saying that the explanation as to why he was -12 is because he made everyone worse -- because he did. He was awful. His stats were not a trick of the light. Every time he came in, he was replacing a better player, which made the lineup tank. He messed up spacing, didn't defend well, missed his looks. That's determined by actually looking at the game, not the number. And if SAC had made seven threes in a row, that would have been the explanation of the number. The point I was making is that the -12 was the reality no matter what explanation you give. And if he has enough games with -12, it's going to become hard to create a model that explains what happened without putting the blame on him.



I think you're confusing stats with the philosophy of stats. The average in this case is a real number that we can observe. It's not an estimate or approximation, because it's taken out of the entire sample. There's no ideal mean that the average is pretending to be, because there's no hidden variance. You have ALL of the data. If someone asks you how many fingers you have, you're not going to give them a confidence interval. You'll give them an integer (counting parts of fingers as whole to eliminate that loop hole). And if they asked you to give the average number of fingers each person in your house/building/whatever has, you will be able to give a correct number provided you can see everyone's hands. There'd be no variance, save perhaps your own fallability in reporting.

That's the entire point of this debate from my view. Plus-minus is not a statistical model. It's not a sampling. And it's not a suggestion of greatness. It's just a number. And no amount of misuse changes that.

See all of the misused and misleading uses of plus minus based on this thread alone. It is a number, that's presented for an individual player in a box score. Why? It's not an individual stat. It's a stat that involves a minimum of 10 players. You literally can be subbed in an out for a FT and get a plus minus for STANDING ON THE BLOCK. Therefore the whole idea of +/- is misleading as represented by ESPN or NBA.com Easily proven by the responses on here.

Stats is the philosophy of stats - that is what stats is/are. A representation of a hypothetical based on averages. A stat really is only a mark in the past that cannot be replicated 100% of the time, so you're correct, it is just a number. But the whole concept behind stats is that they "prove" something. In reality, they only prove one thing. A mark in the past. It cannot represent anything more or anything less. Which is what you are alluding to, but keep trying to disprove at the last moment of each response.

[NameLess Scrub]

Some interesting discussion here...

I think Chinook is hitting the mark in this discussion, but still, seems like what actually bothers tholdren is the use of +/- as an official individual, game to game stat (along with the similar use in this board).

Which makes me ask, regardless of its individual or lineup wise usefulness, is this stat and adequate one to be presented as it is now?

[NameLess Scrub]

This is stat that's only useful unless you're witnessing the game live.

Off topic but.. go Magic! (too)

[HarlemHeat37]

Some interesting discussion here...

I think Chinook is hitting the mark in this discussion, but still, seems like what actually bothers tholdren is the use of +/- as an official individual, game to game stat (along with the similar use in this board).

Which makes me ask, regardless of its individual or lineup wise usefulness, is this stat and adequate one to be presented as it is now?

Nobody uses single-game +/- anymore, and any knowledgeable fan would never use unadjusted +/-, regardless, same with offensive or defensive rating for an individual(probably the most misused stats in the basketball realm, the same number that has had Carlos Boozer as an elite defender, etc)..advanced +/- metrics are unquestionably useful in both the individual and team unit setting, and just like with any good stat, it should be used in conjunction with other metrics and common sense, of course(which is something anti-stats people always ignore in their criticisms of the numbers, they always act like stat people are basing an entire argument on 1 metric)..

OP's point is silly, but it's mostly agenda-driven..he constantly moves the goal posts in the way that he evaluates players..the reason he hates +/- numbers is because they are very kind to players he dislikes on the Spurs(Kawhi, specifically, and DG, as well)..he was the biggest Splitter hater on this forum, too, and pro-Splitter posters would use metrics to display his on-court impact:lol..that's the primary reason, but the actual thread itself was created after one of his favorites(Jimmer) had a poor game, and I'm assuming people were citing his individual game +/-..

[Chinook]

See all of the misused and misleading uses of plus minus based on this thread alone. It is a number, that's presented for an individual player in a box score. Why? It's not an individual stat. It's a stat that involves a minimum of 10 players.

Again, there's a ton of noise if you create a model from a small sample. But if a player has thousands of minutes in many lineups and still has that bad of a number, it's going to be pretty hard to suggest it's not his fault. Possible, but pretty hard.


You literally can be subbed in an out for a FT and get a plus minus for STANDING ON THE BLOCK.

Actually, no. The second FT doesn't count against the new player. It counts against whoever was on the court when the foul occurred. Have had that same question myself.


Stats is the philosophy of stats - that is what stats is/are.

Stats are equal parts data and theory. You're trying to apply a critique of a theory to data. Plus-minus is just data. It's not mislead, it's not flawed and it's not incorrect. It is just what it is.


A stat really is only a mark in the past that cannot be replicated 100% of the time, so you're correct, it is just a number.

In some cases, maybe. But you're trying to read variance into a place that most people would agree is pointless. There could be some score-keeping bias. But besides that, plus-minus is 100-percent accurate.


But the whole concept behind stats is that they "prove" something.

Nah. The models based off the data aim to prove something. But the data themselves ARE something. And it's like saying that the number 2 is asserting something. It's not. Plus-minus is a counting stat. It is not subject to inherent bias.


In reality, they only prove one thing. A mark in the past. It cannot represent anything more or anything less. Which is what you are alluding to, but keep trying to disprove at the last moment of each response.

I think you're trying to say that stats are descriptive and not prescriptive. I agree. Jimmer's plus-minus isn't proof he'll always suck. But Jimmer's poor on-off numbers are evidence that he's sucked his whole NBA career. The weaknesses of models to predict the future are much more severe than their weaknesses against explaining the past.

Also, as a overall critique of your reasoning, it takes more than pointing out a possible issue with a theory to discredit it. That you can create a scenario in which plus-minus could be used incorrectly doesn't no invalidate the stat.

[snickles]

so to sum up.

small sample sizes and the use of a single analytical tool / stat generally will not provide any reliable conclusions. doing this shows you either have a limited understanding of statistics or are just pushing an agenda.

always use the largest sample size possible, and as many different measurements before trying to compare players.

/thread

[NameLess Scrub]

Nobody uses single-game +/- anymore, and any knowledgeable fan would never use unadjusted +/-, regardless, same with offensive or defensive rating for an individual(probably the most misused stats in the basketball realm, the same number that has had Carlos Boozer as an elite defender, etc)..advanced +/- metrics are unquestionably useful in both the individual and team unit setting, and just like with any good stat, it should be used in conjunction with other metrics and common sense, of course(which is something anti-stats people always ignore in their criticisms of the numbers, they always act like stat people are basing an entire argument on 1 metric)..

OP's point is silly, but it's mostly agenda-driven..he constantly moves the goal posts in the way that he evaluates players..the reason he hates +/- numbers is because they are very kind to players he dislikes on the Spurs(Kawhi, specifically, and DG, as well)..he was the biggest Splitter hater on this forum, too, and pro-Splitter posters would use metrics to display his on-court impact:lol..that's the primary reason, but the actual thread itself was created after one of his favorites(Jimmer) had a poor game, and I'm assuming people were citing his individual game +/-..

Thanks.. I understand.

I still wonder though.. from a box-score perspective, every traditional stat will tell us how a particular player did in that single game without question.
Scored a lot? Shoot poorly? Grabbed rebounds? Fouled?

However, is +/- a definite indication of a player's performance in a single game, so it needs to be in the box-score?..

[Kidd K]

no, it may only mean that he is using him at a correct time... i.e. when he is behind in a game and someone on the team is making 3s and the other team is getting fouled making FTs.

could be subbing in and out for defensive purposes on FTS.

ITS A 10 player stat not a 1 player stat.

Uh, if the coach puts a player in and the team makes a run, he's using him properly. Defensive subbing also = using player properly.

If it was a 10 player stat, everyone on every team would have the same stat, and yet they don't. Same with offensive and defensive ratings. Yes they are "biased towards winning teams" because winning teams have good stats in those categories.

For example, take some player who scores 25 PPG but is a ballhog and shitty defender. Some may argue he's great and has just been on bad teams, or you can look at how his production is not at the end of the day leading to outscoring the opponent because he gives up too many points or consumes too many of his team's possessions in an inefficient manner while not helping his team enough when he doesn't have the ball as he gets his 25 PPG. Thus he has a low O and D Rating and a low +/-.

[Mikeanaro]

Uh, if the coach puts a player in and the team makes a run, he's using him properly. Defensive subbing also = using player properly.

If it was a 10 player stat, everyone on every team would have the same stat, and yet they don't. Same with offensive and defensive ratings. Yes they are "biased towards winning teams" because winning teams have good stats in those categories.

For example, take some player who scores 25 PPG but is a ballhog and shitty defender. Some may argue he's great and has just been on bad teams, or you can look at how his production is not at the end of the day leading to outscoring the opponent because he gives up too many points or consumes too many of his team's possessions in an inefficient manner while not helping his team enough when he doesn't have the ball as he gets his 25 PPG. Thus he has a low O and D Rating and a low +/-.
Also winning a game doesnt guarantee certain player will end up getting a positive +/-.

[tholdren]

Nobody uses single-game +/- anymore, and any knowledgeable fan would never use unadjusted +/-, regardless, same with offensive or defensive rating for an individual(probably the most misused stats in the basketball realm, the same number that has had Carlos Boozer as an elite defender, etc)..advanced +/- metrics are unquestionably useful in both the individual and team unit setting, and just like with any good stat, it should be used in conjunction with other metrics and common sense, of course(which is something anti-stats people always ignore in their criticisms of the numbers, they always act like stat people are basing an entire argument on 1 metric)..

OP's point is silly, but it's mostly agenda-driven..he constantly moves the goal posts in the way that he evaluates players..the reason he hates +/- numbers is because they are very kind to players he dislikes on the Spurs(Kawhi, specifically, and DG, as well)..he was the biggest Splitter hater on this forum, too, and pro-Splitter posters would use metrics to display his on-court impact:lol..that's the primary reason, but the actual thread itself was created after one of his favorites(Jimmer) had a poor game, and I'm assuming people were citing his individual game +/-..
Many use single game plus minus. It's included in 2 major sports site's box scores.
Someone is mad. Sorry you overrated splitter. Don't dislike Leonard, and yes seeing people on the board using plus minus is ridiculous

[tholdren]

Again, there's a ton of noise if you create a model from a small sample. But if a player has thousands of minutes in many lineups and still has that bad of a number, it's going to be pretty hard to suggest it's not his fault. Possible, but pretty hard.

You obviously do not understand plus minus. I t takes 10 players to create a net score. A player does not have to contribute anything to get a plus minus. 9 other players could all score,defend, assist, and that one player could stand in the corner and do nothing and have a plus minus.


Actually, no. The second FT doesn't count against the new player. It counts against whoever was on the court when the foul occurred. Have had that same question myself.
nice to know. So where does that net point go? Strange and seemingly erroneous and invalid.



Stats are equal parts data and theory. You're trying to apply a critique of a theory to data. Plus-minus is just data. It's not mislead, it's not flawed and it's not incorrect. It is just what it is.
stats are a measure, true, but think about a stat ppg. 23.8 not able to happen nor replicated

In some cases, maybe. But you're trying to read variance into a place that most people would agree is pointless. There could be some score-keeping bias. But besides that, plus-minus is 100-percent accurate.



Nah. The models based off the data aim to prove something. But the data themselves ARE something. And it's like saying that the number 2 is asserting something. It's not. Plus-minus is a counting stat. It is not subject to inherent bias.



I think you're trying to say that stats are descriptive and not prescriptive. I agree. Jimmer's plus-minus isn't proof he'll always suck. But Jimmer's poor on-off numbers are evidence that he's sucked his whole NBA career. The weaknesses of models to predict the future are much more severe than their weaknesses against explaining the past.

Also, as a overall critique of your reasoning, it takes more than pointing out a possible issue with a theory to discredit it. That you can create a scenario in which plus-minus could be used incorrectly doesn't no invalidate the stat.

[tholdren]

No, the point is that the statistic is not valid to make any individual assumption about a player of performance.

Think of it this way. If only 5 people played from each team, then at the end 5 players from the winning team would have 5 identical positive plus minus points. The losing team would have 5 identical negative points. By using this to determine player value, there would be 5 who played well and 5 who didnt. The stat is not idicative of how any of the ten played.

[HarlemHeat37]

No, the point is that the statistic is not valid to make any individual assumption about a player of performance.

Think of it this way. If only 5 people played from each team, then at the end 5 players from the winning team would have 5 identical positive plus minus points. The losing team would have 5 identical negative points. By using this to determine player value, there would be 5 who played well and 5 who didnt. The stat is not idicative of how any of the ten played.

:lol

https://janeaustenrunsmylife.files.wordpress.com/2014/05/bae6v.gif

[Chinook]

No, the point is that the statistic is not valid to make any individual assumption about a player of performance.

Think of it this way. If only 5 people played from each team, then at the end 5 players from the winning team would have 5 identical positive plus minus points. The losing team would have 5 identical negative points. By using this to determine player value, there would be 5 who played well and 5 who didnt. The stat is not idicative of how any of the ten played.

Yes, but the reason why it's an individual stat is because games don't work that way. There are rotations and not just line subs. There are injuries, spots subs for big moments, garbage time. As each of those scenarios is encountered and factored in, the situational arguments you make lose strength. It becomes less likely that any scenario you come up with explains the whole of the model. Jimmer for example, has been a strong net-negative player despite playing on three different teams and in four or five different systems, despite playing a lot or a few minutes, despite the constant roster flux and role and despite the dozens of opponents. When someone carries a net-negative rating through all that, it's pretty likely that it's because he doesn't help his team win. Along the same lines, Bonner carrying an elite plus-minus is likely because he helped his teams win.

In your scenario, the plus-minus would still be accurate, but because it only speaks to the lineup, that would be the focus. There would be something wrong with that lineup, and switching out players would give more data to isolate bad players. Again, the stat is just fine, but the application has to be understood to use it effectively.

[tholdren]

Yes, but the reason why it's an individual stat is because games don't work that way. There are rotations and not just line subs. There are injuries, spots subs for big moments, garbage time. As each of those scenarios is encountered and factored in, the situational arguments you make lose strength. It becomes less likely that any scenario you come up with explains the whole of the model. Jimmer for example, has been a strong net-negative player despite playing on three different teams and in four or five different systems, despite playing a lot or a few minutes, despite the constant roster flux and role and despite the dozens of opponents. When someone carries a net-negative rating through all that, it's pretty likely that it's because he doesn't help his team win. Along the same lines, Bonner carrying an elite plus-minus is likely because he helped his teams win.

In your scenario, the plus-minus would still be accurate, but because it only speaks to the lineup, that would be the focus. There would be something wrong with that lineup, and switching out players would give more data to isolate bad players. Again, the stat is just fine, but the application has to be understood to use it effectively.

You keep changing the argument. The stat can no way be used to determine the individual worth of a player. It doesn't even say anything about the whole lineups performance.

If it were a useful individual stat you would bd able to tell me the best player out of the 10 players in the scenario above. But there is no way to do that. Which has everything to do with the usefulness of the statistic. You trying to argue only supports the fact that this stat is misused and misleading. Again the original post states that individual worth cannot be derived from plus minus.

You keep admitting this in ambiguous fashion. You're in too deep

[Mikeanaro]

Yes, but the reason why it's an individual stat is because games don't work that way. There are rotations and not just line subs. There are injuries, spots subs for big moments, garbage time. As each of those scenarios is encountered and factored in, the situational arguments you make lose strength. It becomes less likely that any scenario you come up with explains the whole of the model. Jimmer for example, has been a strong net-negative player despite playing on three different teams and in four or five different systems, despite playing a lot or a few minutes, despite the constant roster flux and role and despite the dozens of opponents. When someone carries a net-negative rating through all that, it's pretty likely that it's because he doesn't help his team win. Along the same lines, Bonner carrying an elite plus-minus is likely because he helped his teams win.

In your scenario, the plus-minus would still be accurate, but because it only speaks to the lineup, that would be the focus. There would be something wrong with that lineup, and switching out players would give more data to isolate bad players. Again, the stat is just fine, but the application has to be understood to use it effectively.
:cry Are you saying Jimmer sucks!?!?!?!
I think its a conspiracy to make him look bad, those 3 teams players coaches and 58 systems are involved.




:lol

[tholdren]

http://wagesofwins.com/2011/03/05/deconstructing-the-adjusted-plus-minus-model/

Here

[tholdren]

:cry Are you saying Jimmer sucks!?!?!?!
I think its a conspiracy to make him look bad, those 3 teams players coaches and 58 systems are involved.




:lol
It's not about that. Eyeballs tell you jimmer sucked, plus minus has nothing to do with it. To say jimmer sucked because he had a bad plus minus makes you look like you have had zero education

[ElNono]

Thanks.. I understand.

I still wonder though.. from a box-score perspective, every traditional stat will tell us how a particular player did in that single game without question.
Scored a lot? Shoot poorly? Grabbed rebounds? Fouled?

However, is +/- a definite indication of a player's performance in a single game, so it needs to be in the box-score?..

If a team fields the same 5 starters and the same 5 guys off the bench (more or less the same rotation), +/- doesn't really mean much individually. How could it? It's measuring a scoring differential for 5 guys on the floor. If those 5 guys are largely the same guys, then this particular stat doesn't help you measure anything individually, no matter how large your sample size is. Now, while not all 5 players are substituted at the same time (except when Pop is drunk), the rotations in most NBA teams generally are a 5 starters / 5 primary bench players. That's what you're going to get the biggest sample size from (and thus the most significant statistical data from this stat). If the team rotation is more random (due to injuries, coaches tinkering with lineups, etc), then yeah, you'll get more significant samples of player triplets, duets, and down to the individual level.

This is not even taking into account who you're playing against, which can obviously insert it's own bias, but that's what adjusted +/- tries to address.

All that doesn't mean the stat is broken or it doesn't work or it's unusable. It measures certain impact on the floor, and for example, it can show certain player duets or triplets being really dominant due to factors that individual stats could never measure (good chemistry, good communication, etc). It's no silver bullet, it's just one more stat you need to combine with others to get a good picture about the team as a whole.

[Chinook]

You keep changing the argument.

No, what I'm doing is responding to all the shit you're throwing against the wall hoping for it to stick. The argument is that plus-minus isn't misleading, and you're twisting yourself in a knot trying to find out exactly how you can misinterpret it. But again, it's just like points, rebounds, assists. It doesn't have a bias, because it's not based on what people feel is more important.


The stat can no way be used to determine the individual worth of a player.

You've done nothing to establish this. You've simply asserted that it can't in certain scenarios. Certain completely contrived and unrealistic scenarios. Saying something can fail is not the same thing as saying it doesn't work. It's like saying planes are a bad mode of transportation because they sometimes crash.


It doesn't even say anything about the whole lineups performance.

Of course it does. If a lineup constantly gets outscored by its opponents, it's a poor lineup. As Nono said, it's even stronger than individual plus-minus.


If it were a useful individual stat you would bd able to tell me the best player out of the 10 players in the scenario above.

No, again, you made an unrealistic scenario, and no real model has to be viable under unrealistic conditions. Let's take TS% for example. Player A shoots 60 percent, while player B doesn't make a shot (hell, none of his shots are particularly close) but gets 75 percent of his shots goaltended because of a rookie big man who doesn't end up seeing the end of his 10-day contract. In that situation, player B has a higher TS%, but no reasonable person would assume he was more efficient than player A. Does this mean TS% is somehow a poor stat?

Or assists: Player A runs 10 PnRs to perfection, but his running mate is Ayres who fumbles the passes out of bounds. Player B does 10 isos for 23 seconds before kicking the ball out to Mills, who manages to make a couple of those shots. Player A has no assists and potentially a couple of TOs, while player B has a couple of assists and no TOs. Is player B a better playmaker? No. Does that mean assists aren't a good stat? No.

That's why your strategy is silly. You can't break system by showing it can be beaten. No one claims models based on plus-minus always work. But there's a large gap between workability and perfection.


You keep admitting this in ambiguous fashion. You're in too deep

Man, you're in too deep trying to defend Jimmer. If you are a statistician like you suggest you are, then you know all about models and confidence intervals. And if you know about those, you know it's silly to attack stats by pointing out they don't work 100 percent of the time. And you should know about sample sizes and how having a large sample normalizes almost all variation. These are first-week concepts in any stats class. So why are all of your arguments ignoring them?

[Mikeanaro]

It's not about that. Eyeballs tell you jimmer sucked, plus minus has nothing to do with it. To say jimmer sucked because he had a bad plus minus makes you look like you have had zero education
Eyeballs have very short memory span and perception, thats why we as humans write things down and measure stuff to have references and as chinook said, if you have a bad plus minus on every one of those 3 teams and 58 systems it means you are not a great player, hence why he is struggling to find a team, he was a college sensation a ¨shimmer¨ but now he is just a ¨jiver¨.

[tholdren]

No, what I'm doing is responding to all the shit you're throwing against the wall hoping for it to stick. The argument is that plus-minus isn't misleading, and you're twisting yourself in a knot trying to find out exactly how you can misinterpret it. But again, it's just like points, rebounds, assists. It doesn't have a bias, because it's not based on what people feel is more important.



You've done nothing to establish this. You've simply asserted that it can't in certain scenarios. Certain completely contrived and unrealistic scenarios. Saying something can fail is not the same thing as saying it doesn't work. It's like saying planes are a bad mode of transportation because they sometimes crash.



Of course it does. If a lineup constantly gets outscored by its opponents, it's a poor lineup. As Nono said, it's even stronger than individual plus-minus.



No, again, you made an unrealistic scenario, and no real model has to be viable under unrealistic conditions. Let's take TS% for example. Player A shoots 60 percent, while player B doesn't make a shot (hell, none of his shots are particularly close) but gets 75 percent of his shots goaltended because of a rookie big man who doesn't end up seeing the end of his 10-day contract. In that situation, player B has a higher TS%, but no reasonable person would assume he was more efficient than player A. Does this mean TS% is somehow a poor stat?

Or assists: Player A runs 10 PnRs to perfection, but his running mate is Ayres who fumbles the passes out of bounds. Player B does 10 isos for 23 seconds before kicking the ball out to Mills, who manages to make a couple of those shots. Player A has no assists and potentially a couple of TOs, while player B has a couple of assists and no TOs. Is player B a better playmaker? No. Does that mean assists aren't a good stat? No.

That's why your strategy is silly. You can't break system by showing it can be beaten. No one claims models based on plus-minus always work. But there's a large gap between workability and perfection.



Man, you're in too deep trying to defend Jimmer. If you are a statistician like you suggest you are, then you know all about models and confidence intervals. And if you know about those, you know it's silly to attack stats by pointing out they don't work 100 percent of the time. And you should know about sample sizes and how having a large sample normalizes almost all variation. These are first-week concepts in any stats class. So why are all of your arguments ignoring them?

This has no bearing on any player, or all bearing on every player. In every scenario there are too many variables for plus minus to be used. Read the article.

If I play in a lineup with Tim Duncan and three other spurs players, I would always play the worst. Yet our plus minus would be equal. You cannot attribute value individually from that stat.

And again, the original post was about that point. Many on here look at the box and say "plus minus shows" then they list an individual. It's retarded. And no, plus minus does not give specifics about lineups either. How many times do players like harden, Bryant, kd, curry, go on runs? 9 players get plus minus adjustments from a player scoring 8 in a row, but that has no correlation with 8 players on the court other than the scorer and the defender.

If plus minus were really that great why did it get adapted? Why are there still many who don't buy in? It's evident you understand the basis of statistical measurement, but it's interesting you cannot put it into context other than just stating the typical "sample size" bs

[tholdren]

You make the perfect pointsc about TS percent and assists. Many feel hockey assists should count, or if a good pass is fumbled by a player then the to should not be attributed to the passer. However none of those instances you give relate to 10 players being statistically affected by one play, then that stat be used to determine individual worth. A better example would be opp points scored per game means everyone on the team is a good of bad defender.

[Spur Bank]

Chinook, we appreciate your attempts to educate tholdren -- I'll admit I was tempted to go line-by-line with his "arguments" as well. But I humbly suggest you stop wasting your time on him.

tholdren... I don't know what to say to you. The sheer volume of factually incorrect statements you've made in this thread boggles my mind.

[tholdren]

Chinook, we appreciate your attempts to educate tholdren -- I'll admit I was tempted to go line-by-line with his "arguments" as well. But I humbly suggest you stop wasting your time on him.

tholdren... I don't know what to say to you. The sheer volume of factually incorrect statements you've made in this thread boggles my mind.

List away.

[tholdren]

Although I respect some of Chinook ' commentary, he is not an expert on stats, educational or professional, and virtually every published article on relevance of box plus minus in sports would not side with him.

But whereas standard plus/minus stats are unreliable insofar as they are highly influenced by who a player is sharing the floor with – both teammates and opponents – RPM attempts to isolate each individual’s performance to give a measure of their true value....

[tholdren]

Since the invention of basic plus minus, the plus-minus technique has been developed in order to get more accurate results. At the moment, there are 4 advanced plus minus metrics that derived from basic one.

Net Plus-Minus*(NPM)Adjusted Plus-Minus*(APM)Statistical Plus-Minus*(SPM)Regularized Adjusted Plus-Minus*(RAPM)

[Spur Bank]

Although I respect some of Chinook ' commentary, he is not an expert on stats, educational or professional, and virtually every published article on relevance of box plus minus in sports would not side with him.
Are you?

I wouldn't say I'm an "expert," but I am an actuary and my day job is statistical modeling. Among my nerdy hobbies are playing with R on my home computer and reading and contributing to online math forums. I'll soon be publishing a paper on the underlying statistical functions for generating stochastic equity and interest rates. I majored in math, competed in math competitions in college, and was president of my university's math club, although this is barely relevant because I'm in my mid-30's now.

But, so what? What does that qualify me to say? Well -- as someone who deals with math and statistics day in and day out, both professionally and recreationally, in my opinion it means I'm vastly overqualified to say that you have very little idea what you are talking about.

[tholdren]

Are you?

I wouldn't say I'm an "expert," but I am an actuary and my day job is statistical modeling. Among my nerdy hobbies are playing with R on my home computer and reading and contributing to online math forums. I'll soon be publishing a paper on the underlying statistical functions for generating stochastic equity and interest rates. I majored in math, competed in math competitions in college, and was president of my university's math club, although this is barely relevant because I'm in my mid-30's now.

But, so what? What does that qualify me to say? Well -- as someone who deals with math and statistics day in and day out, both professionally and recreationally, in my opinion it means I'm vastly overqualified to say that you have very little idea what you are talking about.
Then you would agree kevin Durant is the 62 best nba player... thanks plus minus

[Spur Bank]

List away.
I'm not going to list away, but as an example, I'll use something from your first post:

Plus minus has absolutely no correlation to a SINGLE PLAYER

First of all, that's a nonsensical statement. Correlation is a numerical measure of the dependence, or relationship, of two measurable items. But let's assume you meant a single player's *performance* and forgot to say that word. If that's the case, to say it has no correlation is so obviously wrong that if you can't see that, I really have no hope for you.

[Spur Bank]

Then you would agree kevin Durant is the 62 best nba player... thanks plus minus
Wow, I take it all back, you have shown me the error of my ways. Many wisdom. Such brain. Wow.

[tholdren]

Wow, I take it all back, you have shown me the error of my ways. Many wisdom. Such brain. Wow.
Then you also agree Andrew Wiggins was the worst player in the league? Over 400 players ahead of him. He's dead last. But plus minus box is an individual stat.

[tholdren]

I'm not going to list away, but as an example, I'll use something from your first post:

Plus minus has absolutely no correlation to a SINGLE PLAYER

First of all, that's a nonsensical statement. Correlation is a numerical measure of the dependence, or relationship, of two measurable items. But let's assume you meant a single player's *performance* and forgot to say that word. If that's the case, to say it has no correlation is so obviously wrong that if you can't see that, I really have no hope for you.
For you to pretend you didn't know what I meant and avoid the argument with name calling is telling. Explain why plus minus had to be adjusted multiple times to estimate an individual's worth or performance and renamed.

Who is better Andrew Wiggins or Jeremy lin? Is Wiggins really not better than anyone who played a minute of nba games? How exactly does plus minus box determine individual performance?

[Cry Havoc]

Are you?

I wouldn't say I'm an "expert," but I am an actuary and my day job is statistical modeling. Among my nerdy hobbies are playing with R on my home computer and reading and contributing to online math forums. I'll soon be publishing a paper on the underlying statistical functions for generating stochastic equity and interest rates. I majored in math, competed in math competitions in college, and was president of my university's math club, although this is barely relevant because I'm in my mid-30's now.

But, so what? What does that qualify me to say? Well -- as someone who deals with math and statistics day in and day out, both professionally and recreationally, in my opinion it means I'm vastly overqualified to say that you have very little idea what you are talking about.

Yeah, just FYI, most of us gave up on tholdren long ago. Like, around the time he made his 3rd or 4th post, when it became readily apparent that he didn't know what was talking about, which isn't that bad of a problem, but the fact that he would never, ever, admit that he's wrong about anything, which is actually quite hilarious.

Inb4 he makes some off-color quip about me, since, yanno, he's already lost this debate pretty soundly and is going to try to be an ass to save face.

[Cry Havoc]

Thanks.. I understand.

I still wonder though.. from a box-score perspective, every traditional stat will tell us how a particular player did in that single game without question.
Scored a lot? Shoot poorly? Grabbed rebounds? Fouled?

However, is +/- a definite indication of a player's performance in a single game, so it needs to be in the box-score?..

That's not accurate at all, because "traditional" stats don't account for a lot of plays that are made on the court. For instance, Bruce Bowen was a terrible player with traditional stats. His best season using those numbers were 9.3 ppg, 3.9 rebounds, .8 steals, and .5 blocks per 36 minutes. His actual numbers are even lower than that. Those are pedestrian, at best, given his 32 mpg. Yet no one would ever say that Bowen wasn't key to the success of the Spurs through the 2000s. His game in, game out contributions were not at all measurable by points, rebounds, blocks, steals, and assists, but by the defensive presence he had on the court that could so thoroughly shut down a weapon on the other team.

Similarly, Duncan's defense is even better than indicated for much of his career because of what an incredible help-side defender he has been that often alters a shot before it even becomes a shot.

[freemeat]

And here is another terrible explanation/use of plus minus. Shot alteration has nothing to do with plus minus.
Boban alters a shot and opponent scores off rebound, boban is still -2.

Boban did't get the rebound and deserves to be -2.

[Cry Havoc]

Along the same lines, Bonner carrying an elite plus-minus is likely because he helped his teams win.

If Bonner played in the playoffs like he did in the regular season, he would have been a massive weapon to have. Unfortunately he doesn't play well when it counts.

[Spur Bank]

For you to pretend you didn't know what I meant and avoid the argument with name calling is telling. Explain why plus minus had to be adjusted multiple times to estimate an individual's worth or performance and renamed.
The same reason ERA became adjusted to FIP in baseball, which later led to xFIP and then SIERA. ERA has been "adjusted"/improved upon to a great degree... but it doesn't mean ERA is useless, it's just not as predictive as the others.


Who is better Andrew Wiggins or Jeremy lin? Is Wiggins really not better than anyone who played a minute of nba games? How exactly does plus minus box determine individual performance?
It "determines" individual perform in that it is an indicator of an individual's contribution. Going back to the ERA example, which like plus/minus is a stat used on individuals but is to some degree team-dependent, Clayton Kershaw had a not-very-good ERA the first half of this season. Did that tell us that Shelby Miller was better than Clayton Kershaw? No, of course not, just like your NBA example.

You seem to be missing the point. It is an indicator of individual performance, yet when used exclusively and on its own, it suffers from enormous limitations. You are speaking about it as if it is entirely useless.

[TheDoctor]

Dear Fans with Learning Disabilities in Math,

Please note that Plus Minus, has ZERO correlation to individual performance or player.

Example:

Player A is on the court for 20 minutes. In that time player A has 6 turnovers, 0/7, 0 assists, 0 rebounds, 0 blocks. During that 20 minutes the man PLAYER A is defending is the only player to score. Player A gives up 20 points during the time he was in the game. During the same time his other teammates score 30.

PLAYER A WOULD HAVE A +/- OF = +10
...........................................
Example:

Player B is on the court for 20 minutes. In that time player B has 0 Turnovers 10/10, 6 steals, 8 rebounds, and 3 blocks. During that 20 minutes the man that Player B is defending does not score, get an assist, or a rebound. Unfortunately for Player B during that 20 minutes his team was outscored 20-30.

PLAYER B WOULD HAVE A +/- OF = -10
-------------------------------------

Some of you would look at the plus minus and claim that Player A had the better game, but as you can see, this would not be the case. Plus minus has absolutely no correlation to a SINGLE PLAYER. I have no idea why/how anyone with minimal knowledge of statistics could misconstrue this statistic. The only thing that this stat would possibly do is generate questions.

STOP USING PLUS MINUS AS SOME TYPE OF EVIDENCE.

Thank You,

Mathematicians

Thank you for taking the time to let us know.

[NameLess Scrub]

That's not accurate at all, because "traditional" stats don't account for a lot of plays that are made on the court. For instance, Bruce Bowen was a terrible player with traditional stats. His best season using those numbers were 9.3 ppg, 3.9 rebounds, .8 steals, and .5 blocks per 36 minutes. His actual numbers are even lower than that. Those are pedestrian, at best, given his 32 mpg. Yet no one would ever say that Bowen wasn't key to the success of the Spurs through the 2000s. His game in, game out contributions were not at all measurable by points, rebounds, blocks, steals, and assists, but by the defensive presence he had on the court that could so thoroughly shut down a weapon on the other team.

Similarly, Duncan's defense is even better than indicated for much of his career because of what an incredible help-side defender he has been that often alters a shot before it even becomes a shot.

I completely agree. But what I meant was that traditional stats tell us how the player did in those particular areas.
Can +/- do that for a single game?

At this point in time, I guess anybody could use eye test and advanced stats to show how valuable Bruce Bowen was,
by going away from the regular box scores..

It seems like most people here agree that +/- is useful as it is observed over time, and that one single game will tell almost nothing.

So I'm wondering then, how useful is it to have it listed for every single game?
Can we put it in the same level of reliability of the other stats in the box score?

[HarlemHeat37]

:lmao what a meltdown from.this moron that doesn't understand a single thing about using stats, tbh..

:(I'm a statistician, yet I'm the same person that cited Rasual Butler's Playoff PER from 7 minutes of playing time as an argument against him :(

[Chinook]

Advanced stats are an attempt to take objective data and finagle them to fit a subjective evaluation. Sometimes, the new numbers correlate better. RAPM and PER are examples of this. Sometimes, they don't correlate very well. DRtg is an example of this. While they are improvements over counting stats in the sense that they often make better models, they do NOT replace those stats.

Plus-minus is not an advanced stat. It wasn't invented by someone looking to make a point or conform to a standard view. It doesn't have to correlate to performance any better than points do. In that way, the articles you're linking make perfect sense. No one would build a model based on points alone. So anyone who bases a model off plus-minus will probably be inaccurate.

But just as we care about how many points a player scores in a given game and as an average across a season or career, we should care about plus-minus. Because it matters. People who are good at it over a long career are good for a reason, and people who are bad at it for a career are bad for a reason.

Bonner may not be a great player, but you can argue that he's one of the most pivotal guys is modern NBA history. His ability to space the floor and guard opposing bigs has led to a movement toward stretch-bigs that eventually became the small-ball movement we have today. The entire league changed because guys like Bonner started to break teams like the 2010 Lakers. A team like Memphis is a paper tiger because they're allergic to spacing. Golden State got away with playing a 6-7 center because they spaced the floor. It's insane how much the NBA has changed because Bonner and guys like him absolutely own in plus-minus.

[Brazil]

nvm

[Brazil]

:lmao what a meltdown from.this moron that doesn't understand a single thing about using stats, tbh..

:(I'm a statistician, yet I'm the same person that cited Rasual Butler's Playoff PER from 7 minutes of playing time as an argument against him :(

:lol tbh...

[Macca76]

Dear Fans with Learning Disabilities in Math,

Please note that Plus Minus, has ZERO correlation to individual performance or player.

Example:

Player A is on the court for 20 minutes. In that time player A has 6 turnovers, 0/7, 0 assists, 0 rebounds, 0 blocks. During that 20 minutes the man PLAYER A is defending is the only player to score. Player A gives up 20 points during the time he was in the game. During the same time his other teammates score 30.

PLAYER A WOULD HAVE A +/- OF = +10
...........................................
Example:

Player B is on the court for 20 minutes. In that time player B has 0 Turnovers 10/10, 6 steals, 8 rebounds, and 3 blocks. During that 20 minutes the man that Player B is defending does not score, get an assist, or a rebound. Unfortunately for Player B during that 20 minutes his team was outscored 20-30.

PLAYER B WOULD HAVE A +/- OF = -10
-------------------------------------

Some of you would look at the plus minus and claim that Player A had the better game, but as you can see, this would not be the case. Plus minus has absolutely no correlation to a SINGLE PLAYER. I have no idea why/how anyone with minimal knowledge of statistics could misconstrue this statistic. The only thing that this stat would possibly do is generate questions.

STOP USING PLUS MINUS AS SOME TYPE OF EVIDENCE.

Thank You,

Mathematicians

"MathematicianS" plural ? So you admit that you're not alone in your head ? :lol

[JeffDuncan]

Yeah, well, whatever you're talking about, I'm just glad Jimmer came into his own in the Heat game. His +/- was a respectable 0, while Patty Mills fell off a cliff at -10. I always knew Jimmer was the better player. Bye bye Patty, here comes Jimmer!

[tholdren]

The same reason ERA became adjusted to FIP in baseball, which later led to xFIP and then SIERA. ERA has been "adjusted"/improved upon to a great degree... but it doesn't mean ERA is useless, it's just not as predictive as the others.


It "determines" individual perform in that it is an indicator of an individual's contribution. Going back to the ERA example, which like plus/minus is a stat used on individuals but is to some degree team-dependent, Clayton Kershaw had a not-very-good ERA the first half of this season. Did that tell us that Shelby Miller was better than Clayton Kershaw? No, of course not, just like your NBA example.

You seem to be missing the point. It is an indicator of individual performance, yet when used exclusively and on its own, it suffers from enormous limitations. You are speaking about it as if it is entirely useless.

Unfortunately, you are missing the point.
First of all, ERA cannot be linked to plus minus. Terrible analogy. Pitcher has to throw the ball to get ERA. Basketball player doesn't have to touch the ball the whole time he is on the court and he will end up with a plus minus - whether he does good or not - his plus minus will more than likely be based on the actions of 9 other players more-so than the player himself in many instances.

Take the Knicks. Carmelo Anthony is ball dominant. He goes down and shoots 6 times in a row. Based upon his 6 shots and the defense of the man defending him. 8 other players get added or subtracted values based upon the outcome of that shot. How can these numbers give me an indication of the 8 other players. It can't. Now take that in combination with the 48 minutes of play, where some plays 2 players are involved and other plays 10 players are involved. Plus minus is adjusted after every point. Ignorant to think you can decipher individual performance.

[tholdren]

Jason Washburn played against the clippers stellar game - he played 3 minutes, recorded one turnover and no other in game statistics, and finished with a plus minus of +6
in the same game CJ wilcox (4 min) had the same stats, one turnover, with a -6.

[Mikeanaro]

http://www.aolcdn.com/photogalleryassets/moviefoneuk/902420/identity-410a-062510.jpg