SHU +/-

[mattyv1908]

You seem to be the resident stat geek, care to explain why it is useless?

I'll give you a perfect example.  Kevin Durant, one of the NBA's best players was one of the worst +/- players for 2 out of the 3 seasons since the NBA adopted the stat.  The other year he was one of the league's best +/- players.  His personal statistics were relative all three seasons and his team's success was relative all three seasons and yet his +/- numbers swung wildly.

The NBA plays 82 games in a season and advanced statisticians have proven that after 2-3 SEASONS +/- numbers begin to show indications of a player's effect on games he plays in and only after altering the formula for variances.

It's a pretty worthless stat.

[mattyv1908]

Also, the best NCAA player on a bad NCAA team who logs major minutes and plays all 30 games with his team will have some of the worst +/- numbers on the team.  If his team averaged 65 points/game and his team's opponents averaged 75 points/game for the season it's obvious he will be the one most affected by his mediocre team.

If a stat is inconclusive at best over multiple seasons then it is absolutely worthless over the course of a single game.

[BenCat12]

Also, the best NCAA player on a bad NCAA team who logs major minutes and plays all 30 games with his team will have some of the worst +/- numbers on the team.  If his team averaged 65 points/game and his team's opponents averaged 75 points/game for the season it's obvious he will be the one most affected by his mediocre team.

If a stat is inconclusive at best over multiple seasons then it is absolutely worthless over the course of a single game.

With all due respect, I asked Henry.  He has earned credibility with his stats and I respect his opinion.  I understand some of the shortcomings of the stat, but I question why it's "useless." 

[mattyv1908]

Bencat

With all due respect, the stats are readily available and the arguments against it being a quality metric for evaluating players from people far more qualified than anyone on this board.

[BenCat12]

Insightful

[mattyv1908]

Insightful

Hey, I tried to give you a real life scenario of one of the greatest players currently playing at the highest level and how he's fared with the plus minus statistic despite his personal numbers and his team's win percentage remaining mostly static over that same time.

I didn't make the stats up, and yet I was greeted with an 'In all due respect....'

I just was pointed out that people with the time, money and resources far exceeding us casual fans have proven how ineffective +/- stats are at telling a player's value to his team.

I thought we're living in 'the information age' and I didn't realize it was monopolized for only posters with 5000+ posts to share and digest.

[BenCat12]

I suppose I should have sent him a PM, to avoid hurting your feelings.  It's really not worth getting upset about.  You have no credibility with me, Sugar does.  I have found articles on both sides of the value of the +/- stat.  The conclusion I came to was the stat has value in certain contexts.  One of those contexts is to use +/- comparisons to determine best overall lineups.  The stat does not work for situational players like Juan, or small sample sizes like JJJ or Dawson.  But it does work for players who play major minutes, for example Todd vs. Jake.  I am currently watching the Heat Warriors game and the announcers specifically used +/- to show how much better Golden State is with Bogut (who is injured, who'd a thought) in the lineup than without.  Clearly, it has some value.  I asked Sugar because he has written extensive posts on basketball related statistics and I respect his thoughts because of it.

 

[chapman]

The conclusion I came to was the stat has value in certain contexts. 

True of all of them.  Numbers don't lie, but statistics can be spun to fit any agenda 

[Henry Sugar]

With all due respect, I asked Henry.  He has earned credibility with his stats and I respect his opinion.  I understand some of the shortcomings of the stat, but I question why it's "useless."  

Honestly, I would have just written up something similar to mattyv. His explanation was very good and credible.

The sole advantage of +/- is that it's clear. Everyone can understand the concept of +/-. However, that's not a good thing, because the stat is inconsistent across games, years, and teams. It's not reproducible and has no way of deriving why a specific result in a game is obtained. In fact, because the stat is so clear and yet also inconsistent/not reproducible, that makes it dangerous. It's like the concept of pitcher wins.

Here's a decent further explanation.
http://wagesofwins.com/2013/02/15/measure-once-predict-twice-thoughts-on/

[BenCat12]

Henry, I appreciate the feedback.
The stat should be used to evaluate single players, on a team in a single year.  Comparing +/- with different seasons or different teams in the same season, or different seasons is useless.  But using it for a single team, in a single season it can show which 5 players perform the best together and who should see the court and who shouldn't.  This is why I started measuring +/- in the first place.  This was the first season I was so baffled by the playing time and rotations that I had to come up with some reason why Buzz was doing what he was doing.  It turns out, after looking at +/-, that Buzz strongly favors consistent mediocrity (Jake and Derrick), over inconsistent upside (Mayo, the freshmen, etc).  The stat is not useless, if you know how to use it and understand its importance.  The more data there is as the season goes on the more valuable the stat is.  In other words, the stat means next to nothing after one game, but after the 25th game it holds major weight in determining what is the best combination of 5 players on the floor.

[willie warrior]

I'll give you a perfect example.  Kevin Durant, one of the NBA's best players was one of the worst +/- players for 2 out of the 3 seasons since the NBA adopted the stat.  The other year he was one of the league's best +/- players.  His personal statistics were relative all three seasons and his team's success was relative all three seasons and yet his +/- numbers swung wildly.

The NBA plays 82 games in a season and advanced statisticians have proven that after 2-3 SEASONS +/- numbers begin to show indications of a player's effect on games he plays in and only after altering the formula for variances.

It's a pretty worthless stat.

Agreed, because it does not consider who the guy is playing with. Wade is a good example. He used to rest before LeBron and Bosh and then when they went out he was playing with 2nd teamers more minutes, and many games he would have a minus stat.

[Henry Sugar]

Henry, I appreciate the feedback.
The stat should be used to evaluate single players, on a team in a single year.  Comparing +/- with different seasons or different teams in the same season, or different seasons is useless.  But using it for a single team, in a single season it can show which 5 players perform the best together and who should see the court and who shouldn't.  This is why I started measuring +/- in the first place.  This was the first season I was so baffled by the playing time and rotations that I had to come up with some reason why Buzz was doing what he was doing.  It turns out, after looking at +/-, that Buzz strongly favors below average consistency (Jake and Derrick), over inconsistent upside (Mayo, the freshmen, etc).  The stat is not useless, if you know how to use it and understand its importance.  The more data there is as the season goes on the more valuable the stat is.  In other words, the stat means next to nothing after one game, but after the 25th game it holds major weight in determining what is the best combination of 5 players on the floor.

I do not agree. However, if you want to keep using +/-, knock yourself out.

[TAMU, Knower of Ball]

It seems to me that +/- is more useful in evaluating bench/role players than it is in evaluating starters. If you are out there 30 minutes, you are going to hit runs and slides. But if you only play 10 minutes, +/- can tell you if you had a positive or negative effect on the floor.