How reliable is +/- stats???

[Twin Towers]

https://www.nba.com/statistics/plusminus/plusminus_sort.jsp?pcomb=1&season=22012&split=9&team=Jazz

Is this reliable??? I know it all depends on who is playing with who but why is our bench way better than the starters??
Is this not a reliable way to look at players???

When I look at player combinations I feel like certain players can get boosts from another but not really deserve that value.
The same can be said with looking at just 1 player at a time but it seems that it would be more accurate.

Is it possible Jefferson makes the starters all worse. I have no idea.

Loved to get everyone's thoughts on these +/- stats.

 

[orangello]

https://www.nba.com/statistics/plusminus/plusminus_sort.jsp?pcomb=1&season=22012&split=9&team=Jazz

Is this reliable??? I know it all depends on who is playing with who but why is our bench way better than the starters??
Is this not a reliable way to look at players???

When I look at player combinations I feel like certain players can get boosts from another but not really deserve that value.
The same can be said with looking at just 1 player at a time but it seems that it would be more accurate.

Is it possible Jefferson makes the starters all worse. I have no idea.

Loved to get everyone's thoughts on these +/- stats.

If it was a one or two game sample then I don't think it is reliable. If it is taking about a quarter of the season them yes, there is a lot of truth to it.

For me it just confirms what I am seeing out on the court.

 

[DutchJazzer]

it is reliable since certain players always are in the minus.

it is no accident jazz lead the league in 10+ point comeback win.

 

[One Brow]

Last I checked into it, the standard error for someone who played about half the available minutes in a season was in the range of 3, and it is even even higher for people who play a lot of minutes, or very few minutes, or only partway into the season. So it's useful in a broad sense, but not a narrow one.
.
For example, in the table below (culled from the front page of 82games.com), there is no meaningful difference in the numbers of Favors and Jefferson, or Mo. Williams and Watson. This early in the season, the difference between Tinsley and Mo. Williams might be significant, but there is still a small, yet significant, possiblility the difference is due to random factors.
.

Player	Min	Net
Tinsley	0.31	15.7
Carroll	0.30	12.8
Ma.Williams	0.48	5.4
Kanter	0.31	5
Hayward	0.56	3.3
Evans	0.06	1.2
Favors	0.36	-2.1
Millsap	0.64	-2.9
Jefferson	0.66	-4.4
Mo.Williams	0.56	-7.7
Foye	0.54	-10.2
Watson	0.12	-11.5
Burks	0.08	-14.7
Murphy	0.01	-64.6

 

[Gyp Rosetti]

Last I checked into it, the standard error for someone who played about half the available minutes in a season was in the range of 3, and it is even even higher for people who play a lot of minutes, or very few minutes, or only partway into the season. So it's useful in a broad sense, but not a narrow one.
.
For example, in the table below (culled from the front page of 82games.com), there is no meaningful difference in the numbers of Favors and Jefferson, or Mo. Williams and Watson. This early in the season, the difference between Tinsley and Mo. Williams might be significant, but there is still a small, yet significant, possiblility the difference is due to random factors.
.

Player	Min	Net
Tinsley	0.31	15.7
Carroll	0.30	12.8
Ma.Williams	0.48	5.4
Kanter	0.31	5
Hayward	0.56	3.3
Evans	0.06	1.2
Favors	0.36	-2.1
Millsap	0.64	-2.9
Jefferson	0.66	-4.4
Mo.Williams	0.56	-7.7
Foye	0.54	-10.2
Watson	0.12	-11.5
Burks	0.08	-14.7
Murphy	0.01	-64.6

We have to consider that our back-ups ply against back-ups much of the time. However, they do play against starters here or there as well. Having said that, and maybe I'm wrong in my way of thinking here, but Marvin Williams and Hayward are the most impressive to me. Their numbers are high despite the fact that they play a lot against starters and get a lot of minutes. Maybe I'm wrong but this is how I'm looking at it. And Kanter could/may be an anomaly as he's not playing the same type of minutes. Following this train of thought, Burks is terrible as he's so low despite getting minimal playing time. His God awful percentages may back up this notion.

 

[One Brow]

We have to consider that our back-ups ply against back-ups much of the time. However, they do play against starters here or there as well. Having said that, and maybe I'm wrong in my way of thinking here, but Marvin Williams and Hayward are the most impressive to me. Their numbers are high despite the fact that they play a lot against starters and get a lot of minutes. Maybe I'm wrong but this is how I'm looking at it. And Kanter could/may be an anomaly as he's not playing the same type of minutes. Following this train of thought, Burks is terrible as he's so low despite getting minimal playing time. His God awful percentages may back up this notion.

Hayward's number is bascially indistinguishable from 0, as is Burks and Murphy, because the sample size on-court is so small. For Millsap/Jefferson, the sample size off-court is also small. Only Tinsley, Carroll, and Williams seem to be close to significantly above zero, only Foye is close to significantly below it.

 

[NUMBERICA]

It's AN indicator. It relies heavily on context. It needs the guidance of other stats and the eye test especially. But it is AN indicator. It's one of the more valuable stats, IMO. There's merit to the idea.

 

[One Brow]

It's AN indicator. It relies heavily on context. It needs the guidance of other stats and the eye test especially. But it is AN indicator. It's one of the more valuable stats, IMO. There's merit to the idea.

I agree. For better or worse, it's the only indicator that takes into account the complete effects a player has on the score of the game.

 

[Twin Towers]

I think this shows that the starters are not getting it done.

Could it be the starters all together they just suck??

Could it be the combination of players (Jefferson and Millsap playing together as an example...)

Whatever the reason something needs to change. Corbin's decision making has been horrible in terms of who plays with who.

Foye should only be playing 15-20 minutes a game. Jefferson and Millsap need to be separated. Would prefer Jefferson to be just traded.

I love Burks energy on the court and how he attacks the basket. He should be playing 15-20 minutes a game also.

I just want a pass first point guard who can hit the open 3. Jose Calderon is who I personally like. Would be happy to trade Jefferson for him.

 

[Weezur]

Randy Foye is 2x worse than the second worse +/- (which is either Mo Williams or Marvin)

 

[Jazz4ever]

What's strange is Millsap is having a bad year with this metric but has always been a +/- champion. He is at -3.8 this year after being +11 last year. Jefferson too, was very good last year at +8.4 and is currently at an appalling -6.5 this year. Someone pointed out how well they worked together last year, a bromance if you will, that is completely absent this year. One would think surrounding them with three point shooters like Mo and Marvin would open up their games but it has not seemed to work at all. Its either a chemistry issue or free agency is weighing heavily on their minds, which I'm not sure.

 

[Twin Towers]

What's strange is Millsap is having a bad year with this metric but has always been a +/- champion. He is at -3.8 this year after being +11 last year. Jefferson too, was very good last year at +8.4 and is currently at an appalling -6.5 this year. Someone pointed out how well they worked together last year, a bromance if you will, that is completely absent this year. One would think surrounding them with three point shooters like Mo and Marvin would open up their games but it has not seemed to work at all. Its either a chemistry issue or free agency is weighing heavily on their minds, which I'm not sure.

Scouting reports

 

[LunaticWolf]

I know it all depends on who is playing with who but why is our bench way better than the starters??
Is this not a reliable way to look at players???

Or against who.

Last night Favors played 14:27 against Lamar and Turiaf, and 4:45 against the starters. He was +12.

Not to give the starters a free pass though, They can't seem to run the offense to save their lives.

 

[Lion's Roar]

#1 It shows where teams are making runs. Jazz are making comebacks with their subs and being beat with the starters.

#2 In a team like Utah where 1st unit and 2nd unit are almost mutually exlcusive you can't use +/- to compare players cause starters and subs face such different opposition.

#3 If the numbers hold for the season you could say that the Jazz 1st unit is bellow league average while the bench unit is above league average.

#4 In a game-by-game setting it's pretty much pointless. It only becomes an issue when trends emerge. Basically in a game setting it helps you establish a sense of a recap trough the boxscore and little else.

 

[Twin Towers]

Or against who.

Last night Favors played 14:27 against Lamar and Turiaf, and 4:45 against the starters. He was +12.

Not to give the starters a free pass though, They can't seem to run the offense to save their lives.

So true. I feel bad for Millsap though. He just stands there watching Jefferson play 1 on 1. No plays are ran for him anymore. Doesn't excuse his bad defense or lack of boxing out or pointless fouls. lol

But if we keep Jefferson past this year I will no longer watch the Jazz. I couldn't watch Jefferson ruin Kanter and Favors.

 

[Twin Towers]

#1 It shows where teams are making runs. Jazz are making comebacks with their subs and being beat with the starters.

#2 In a team like Utah where 1st unit and 2nd unit are almost mutually exlcusive you can't use +/- to compare players cause starters and subs face such different opposition.

#3 If the numbers hold for the season you could say that the Jazz 1st unit is bellow league average while the bench unit is above league average.

#4 In a game-by-game setting it's pretty much pointless. It only becomes an issue when trends emerge. Basically in a game setting it helps you establish a sense of a recap trough the boxscore and little else.

Our Bench plays long periods of time. Other teams they mix it up much more. Our bench does play starters also.

To me I see it as Jefferson being the main problem Along with Foye. But the current combination of starters is not working is what these stats show me. The rest can be debated both ways. But no way anyone can argue that the current combination of starters is working well.

 

[bordelais7]

+/- is completely reliable. Plus or minus infinity reliabilities.