"Real" Plus/Minus Stat

[Hey]

New stat from ESPN, adding up Offensive +/- with Defensive +/-. Stats are stats, so take it for what it's worth.

Explanation here

Player rankings here

Can't sort by team, but here's a quick low-down:

Lillard is 13th best PG with a score of +2.73. He's the third best offensive PG at +4.62 behind only Curry and Paul. But his defense, as you would expect, is abysmal at -1.89.

LMA ranks 9th best PF at +4.91. Surprisingly, his defense ranks better than his offense.

Batum and Matthews are both terrible on defense and their offensive rating isn't anything to boast about either. I guess we can put to rest the fantasy that we have "great wing defenders." Maybe when they want to be, but not very consistent. We already knew this, but it's interesting to see nonetheless. Remember when Batum said he was the best SF in the West outside of Durant? Well he's ranked 27th out of all SF's.

Lopez is 12th best Center. Best defender on our team at +3.22.

 

[Eastoff]

Cool idea, thanks for sharing it!

 

[Strenuus]

Good find. I'll have to dig through it some. Repped!

 

[Strenuus]

As far as I can tell, of the top 100 players on the list, Damian is the worst of them defensive wise. The next closest is Ryan Anderson, but that doesn't seem fair given he's only played 22 games.

 

[Strenuus]

To put it even moreso, Damian is 367th in DRPM... Anthony Morrow is ahead of him, Tristan Thompson as well. Yikes.

Robin Lopez is 28th in that category. LA 35th.

 

[Prue]

Crazy stats...to think Wes and Nic don't rank that well...I know we value them as pair of our top defenders so I got to imagine a lot of the 'non-quantitative' stuff is our benefit with them

 

[riverman]

I think it should be noted that Nic guards the best scorers in the NBA as LeBron, Durant, sometimes Love and even Chris Paul. Dame often doesn't guard elite pt guards but Nico usually gets the toughest assignments. Can they improve? Sure, but stats don't cover those intangibles

 

[Wizard Mentor]

A couple of comments, here:

• In general, I am a fan of advanced stats. I scoff at PPG and people (such as announcers) who try to make a point using that.
• It's unclear whether "Real" +/- takes into consideration who is on the court at the same time as you are. In particular, Lillard plays a bunch of time with bench players which is why his +/- (NetRtG) is lower than some of the others on the team.
• Matthews and Batum's NetRtG is +8 and +6, respectively. If there's a way in which those numbers are somehow screwed up in a way that makes them look bad, then I'm skeptical.
• While I like Nick Collison, he is not the 6th best player in the league or even among the better players in the league.
• I think "Real" +/- is an interesting stat, but like PER, it does a poor job of being an appropriate metric for overall player quality.

 

[jlprk]

• It's unclear whether "Real" +/- takes into consideration who is on the court at the same time as you are. In particular, Lillard plays a bunch of time with bench players which is why his +/- (NetRtG) is lower than some of the others on the team.

That's the gist of the Explanation. Read it for the answer.

 

[blue9]

That's the gist of the Explanation. Read it for the answer.

I read the explanation, and it's not clear how they make a measurement that accounts for 5 (or, actually, 10) players in aggregate and apply it to an individual. I've always felt that +/- was a pretty crap stat (along with PER) because it has more to do with the other players on the floor than it does for the individual who's being ranked. But if they actually did something to account for that, then it becomes more useful. How useful? Who knows - that depends on what they did to make it measure the individual rather than the group of players on the floor during the individual's minutes played.

 

[Blazer4ever]

I wonder what it means that Nico with his defensive talent has a negative rating. Mathewss I never thought was a great defender. Also, Lamarcus' defensive rating is as surprisingly high as Batum's is low.

 

[Blazer4ever]

Also interesting that Jeff Ayres and Nate Wolters have exactly 0 total rating as their offensive and negative defensive rating are exactly equal. They are ranked 155, 156 out of 435 players

 

[Mediocre Man]

Crazy stats...to think Wes and Nic don't rank that well...I know we value them as pair of our top defenders so I got to imagine a lot of the 'non-quantitative' stuff is our benefit with them

Some of we, not all of we. Some of we have been saying for years that Nic is a poor defender

 

[blue9]

I wonder what it means that Nico with his defensive talent has a negative rating. Mathewss I never thought was a great defender. Also, Lamarcus' defensive rating is as surprisingly high as Batum's is low.

My guess is that it means Nic plays with the bench players more than LMA does.

 

[riverman]

My guess is that it means Nic plays with the bench players more than LMA does.

Both Nic and Wes play a lot with the B squad. It's a big factor in the +/- stat line

 

[Blazer4ever]

I think maybe sometimes, other players don't play as hard on defense with Nic on the floor. Maybe without him they play more as a team defensively.

 

[TripTango]

I read the explanation, and it's not clear how they make a measurement that accounts for 5 (or, actually, 10) players in aggregate and apply it to an individual. I've always felt that +/- was a pretty crap stat (along with PER) because it has more to do with the other players on the floor than it does for the individual who's being ranked. But if they actually did something to account for that, then it becomes more useful. How useful? Who knows - that depends on what they did to make it measure the individual rather than the group of players on the floor during the individual's minutes played.

They are running a "ridge regression" on every player for every possession, implicitly taking into account teammates on the floor as well as opposing players. In theory, with enough data this would allow the model to separate the performances of really good players from those of their teammates who just happen to often be on the floor with them at the same time. Like with any statistic it should be taken with a grain of salt, but at the very least it's making an attempt to separate player performance from teammate performance, which is more than can be said for certain other popular advanced stats.

The thing is, none of the above is new -- it's been done already with adjusted plus-minus (APM) and regularized adjusted plus-minus (RAPM). The article mentions improvements to these stats, such as "the use of Bayesian priors, aging curves, score of the game and extensive out-of-sample testing", but all of that is just statistical hand-waving without more specific details.

 

[rasheedfan2005]

Fuck stats

Go with the pop line of thinking, play hard and know your role = wins

 

[riverman]

Fuck stats

Go with the pop line of thinking, play hard and know your role = wins

Well said

 

[TripTango]

Fuck stats

Go with the pop line of thinking, play hard and know your role = wins

I have no idea what "play hard" has to do with NBA player evaluation. My 4-year old plays hard. He also knows his role, as long as he's had his nap. Wins?

 

[rasheedfan2005]

I have no idea what "play hard" has to do with NBA player evaluation. My 4-year old plays hard. He also knows his role, as long as he's had his nap. Wins?

If youre skilled enough to make the nba the major difference in anygame will be effort. Look at the suns. Not a super star in sight but they beat our asses the other night off sheer determination. Spurs constantly turn d league scrubs into winners.

There are very few people in the nba imo that are much more gifted athletically than others. Lebron griffin and howard maybe but the rest is just pure effort.

 

[TripTango]

If youre skilled enough to make the nba the major difference in anygame will be effort. Look at the suns. Not a super star in sight but they beat our asses the other night off sheer determination. Spurs constantly turn d league scrubs into winners.

There are very few people in the nba imo that are much more gifted athletically than others. Lebron griffin and howard maybe but the rest is just pure effort.

Well, then some players just consistently give more effort than others, statistically speaking.

One of the nice things about stats like this is that it doesn't matter how the numbers arrive -- it just measures their patterns. If a team tends to win more when a certain player is on the floor, it doesn't matter if the underlying reason is due to athleticism or basketball IQ or skill or "effort" or using the damn Force. It will show up in the results.

 

[bobf]

If it works like "adjusted plus minus" which it probably does... then it does a linear regression where each individual player has two variables: offensive rating and defensive rating. For every possession there are 5 offensive variables and 5 defensive variables plus an additional variable for home court advantage. It then does a linear regression to fit all the variables to minimize the error (squared).

Given that, it does factor in which players are on the floor, and it takes into account all kinds of intangibles that PER cannot measure.

That main problem with this method is that the standard deviation is too high, so even with 1 full season of data some players will be significantly misrated due to variance. It takes about 5 full seasons of data to get the standard deviation down to reasonable levels... but then the flaw is that players abilities change over the course of 5 years.

Aside from this variance problem... it is far superior to PER.

 

[rasheedfan2005]

Well, then some players just consistently give more effort than others, statistically speaking.

One of the nice things about stats like this is that it doesn't matter how the numbers arrive -- it just measures their patterns. If a team tends to win more when a certain player is on the floor, it doesn't matter if the underlying reason is due to athleticism or basketball IQ or skill or "effort" or using the damn Force. It will show up in the results.

Every team has players better at certain things sure. But if the team as a whole plays harder 99/100 times theyll win (assuming refs arent throwing the game)

 

[rasheedfan2005]

You cant tell me these guys get to the nba and cant play defense.

 

[Sinobas]

I think it should be noted that Nic guards the best scorers in the NBA as LeBron, Durant, sometimes Love and even Chris Paul. Dame often doesn't guard elite pt guards but Nico usually gets the toughest assignments. Can they improve? Sure, but stats don't cover those intangibles

Oh God, that is such fucking bullshit. Other SFs are guarding Lebron and Durant too. If Batum and Matthews were such great defenders our defense wouldn't suck.

 

[TripTango]

Every team has players better at certain things sure. But if the team as a whole plays harder 99/100 times theyll win (assuming refs arent throwing the game)

We'll just have to agree to disagree.

You cant tell me these guys get to the nba and cant play defense.

"Can't play defense" is a funny phrase. I'd be willing to bet the scrubbiest scrub in the NBA could shut you down. They are the best in the world, playing the best in the world. At that level, even small differences in athleticism or talent can become magnified. Yeah, effort counts. But when it comes down to it, players are who they are. No amount of "effort" is going to turn Ronnie Price into Chris Paul, even though both are elite players, on a global measure.

 

[illmatic99]

This stat might now be my most favorite of all. They really took into account everything I can think of and addressed the point I hated most about +/- numbers to determine impact.

Really interested in seeing the variability or CV% in the stats though. Or at least some sort of error calculation to gauge how valid their numbers are.

 

[BLAZER PROPHET]

Crazy stats...to think Wes and Nic don't rank that well...I know we value them as pair of our top defenders so I got to imagine a lot of the 'non-quantitative' stuff is our benefit with them

Stats have to be taken with a HUGE grain of salt.