What kind of Flaws Are In PER stat

Moderator: Doctor MJ

90sgoat
Bench Warmer
Posts: 1,404
And1: 397
Joined: Aug 05, 2016

Re: What kind of Flaws Are In PER stat 

Post#21 » by 90sgoat » Sat Jan 21, 2017 7:11 pm

The_Hater wrote:It overrates rebounding and usage rate/volume scoring. It underates efficiency and it doesn't really tell us much about defense.

It still has its usefulness but not as a stand alone stat.


This.

It also doesn't take into account changes in playing style at all.

It generally overrates 'do it all' jack of all trades and underrates players who play within a system.

It's a stat made for the pre-2001 era.

At the very least, it should not be used in direct comparison across different eras.
Johnlac1
Assistant Coach
Posts: 4,472
And1: 1,059
Joined: Jan 21, 2012
 

Re: What kind of Flaws Are In PER stat 

Post#22 » by Johnlac1 » Wed Jan 25, 2017 6:38 pm

sanitylaker wrote:It overrates volume scoring and and it doesn't measure defense properly. That's exactly why low-usage defensive specialists like Bruce Bowen have a really low career PER.

Bill Russell, who won 11 titles, has a lower career PER than HOFer Walt Bellamy who never even played in a conf. final despite a long career. Bellamy was the better scorer, but that's about it. Not one expert in 100 would take Bellamy over Russell.
User avatar
acrossthecourt
Pro Prospect
Posts: 984
And1: 722
Joined: Feb 05, 2012
Contact:

Re: What kind of Flaws Are In PER stat 

Post#23 » by acrossthecourt » Fri Feb 3, 2017 3:35 am

Twitter: AcrossTheCourt
Website; advanced stats based with a few studies:
http://ascreamingcomesacrossthecourt.blogspot.com
76ersfan
Pro Prospect
Posts: 809
And1: 192
Joined: Feb 03, 2014
       

Re: What kind of Flaws Are In PER stat 

Post#24 » by 76ersfan » Wed Feb 8, 2017 8:22 pm

Johnlac1 wrote:
sanitylaker wrote:It overrates volume scoring and and it doesn't measure defense properly. That's exactly why low-usage defensive specialists like Bruce Bowen have a really low career PER.
Yes, Bill Russell has a lower career PER than Walt Bellamy who never even appeared in one conf. final.

Wow that's an awesome stat. One I gotta use in the future!


Sent from my iPad using RealGM mobile app
Ben Simmons
laika
Starter
Posts: 2,064
And1: 1,095
Joined: Mar 22, 2011

Re: What kind of Flaws Are In PER stat 

Post#25 » by laika » Thu Feb 9, 2017 1:45 pm

76ersfan wrote:
Johnlac1 wrote:
sanitylaker wrote:It overrates volume scoring and and it doesn't measure defense properly. That's exactly why low-usage defensive specialists like Bruce Bowen have a really low career PER.
Yes, Bill Russell has a lower career PER than Walt Bellamy who never even appeared in one conf. final.

Wow that's an awesome stat. One I gotta use in the future!




Hardly anyone knows who Bellamy is.
A better comparison- Enes Kanter, Carlos Boozer, Greg Monroe, Al Jefferson and David Lee are all better than Bill Russell according to PER.
76ersfan
Pro Prospect
Posts: 809
And1: 192
Joined: Feb 03, 2014
       

Re: What kind of Flaws Are In PER stat 

Post#26 » by 76ersfan » Thu Feb 9, 2017 5:50 pm

laika wrote:
76ersfan wrote:
Johnlac1 wrote:Yes, Bill Russell has a lower career PER than Walt Bellamy who never even appeared in one conf. final.

Wow that's an awesome stat. One I gotta use in the future!




Hardly anyone knows who Bellamy is.
A better comparison- Enes Kanter, Carlos Boozer, Greg Monroe, Al Jefferson and David Lee are all better than Bill Russell according to PER.

That's amazing. Just goes to show you that no stat is perfect.


Sent from my iPad using RealGM mobile app
Ben Simmons
crazybranman360
Starter
Posts: 2,043
And1: 34
Joined: Aug 12, 2009
     

Re: What kind of Flaws Are In PER stat 

Post#27 » by crazybranman360 » Sat Feb 11, 2017 3:49 am

It overrates per minute efficiency over players who can consistently produce over long periods of time
Image
Dr Positivity
RealGM
Posts: 36,398
And1: 4,780
Joined: Apr 29, 2009
       

Re: What kind of Flaws Are In PER stat 

Post#28 » by Dr Positivity » Sun Feb 26, 2017 7:21 am

PER does not measure defense and spacing well and is too PPG driven. Stuff like Gay having a higher PER than Draymond exemplifies this problem
asubstituteforwar.wordpress.com
Fencer reregistered
RealGM
Posts: 13,212
And1: 2,795
Joined: Oct 25, 2006

Re: What kind of Flaws Are In PER stat 

Post#29 » by Fencer reregistered » Sat Mar 18, 2017 7:36 pm

paulbball wrote:
Take plus minus from boxscore and adjust it.

We need to do adjustment because teams are not equal. Most GSW players have amazing plus minus. The pros of a plus minus based stat is that they are largely an unbiased measure of a player's effectiveness. Who cares what kind of numbers you put up or don't put up as long as your team is +20 pts per 100 possession with you playing.

APM: you take the basic plus/minus that you see on boxscores, solve a system of linear equations that takes into account the teammates and opponents on the court. Why? See above sentence. It doesn't do a good job at doing this because basketball is not a 1v1 game.


The obvious idea is:
-- Assume everybody has a "true" +/- score.
-- Take every unique collection of 10 guys on the court, look at the actual +/-, and use that as an estimate of the overall sum/difference of the 10 guys' true stats.
-- Fit this data with the best estimates for "true +/-" you mathematically can.

Weighting all that for minutes played would be straightforward. Weighting it for various late-game/end-of-quarter tactics is less so, but such flaws are shared by almost all statistical approaches, so let's ignore it.

Why doesn't that work?
paulbball
Senior
Posts: 636
And1: 340
Joined: Jun 30, 2016
   

Re: What kind of Flaws Are In PER stat 

Post#30 » by paulbball » Sat Mar 18, 2017 8:23 pm

Fencer reregistered wrote:
paulbball wrote:
Take plus minus from boxscore and adjust it.

We need to do adjustment because teams are not equal. Most GSW players have amazing plus minus. The pros of a plus minus based stat is that they are largely an unbiased measure of a player's effectiveness. Who cares what kind of numbers you put up or don't put up as long as your team is +20 pts per 100 possession with you playing.

APM: you take the basic plus/minus that you see on boxscores, solve a system of linear equations that takes into account the teammates and opponents on the court. Why? See above sentence. It doesn't do a good job at doing this because basketball is not a 1v1 game.


The obvious idea is:
-- Assume everybody has a "true" +/- score.
-- Take every unique collection of 10 guys on the court, look at the actual +/-, and use that as an estimate of the overall sum/difference of the 10 guys' true stats.
-- Fit this data with the best estimates for "true +/-" you mathematically can.

Weighting all that for minutes played would be straightforward. Weighting it for various late-game/end-of-quarter tactics is less so, but such flaws are shared by almost all statistical approaches, so let's ignore it.

Why doesn't that work?


You just described the basic ideas behind APM: http://www.82games.com/comm30.htm. Though, I don' understand your last bulletin point.

Here is an extension of APM, RAPM or regularized APM at Sloan 2010: http://www.sloansportsconference.com/wp-content/uploads/2015/09/joeSillSloanSportsPaperWithLogo.pdf

APM shouldn't really be used anymore. RAPM or ESPN's RPM is superior.
Fencer reregistered
RealGM
Posts: 13,212
And1: 2,795
Joined: Oct 25, 2006

Re: What kind of Flaws Are In PER stat 

Post#31 » by Fencer reregistered » Sun Mar 19, 2017 7:25 am

paulbball wrote:
Fencer reregistered wrote:
paulbball wrote:
Take plus minus from boxscore and adjust it.

We need to do adjustment because teams are not equal. Most GSW players have amazing plus minus. The pros of a plus minus based stat is that they are largely an unbiased measure of a player's effectiveness. Who cares what kind of numbers you put up or don't put up as long as your team is +20 pts per 100 possession with you playing.

APM: you take the basic plus/minus that you see on boxscores, solve a system of linear equations that takes into account the teammates and opponents on the court. Why? See above sentence. It doesn't do a good job at doing this because basketball is not a 1v1 game.


The obvious idea is:
-- Assume everybody has a "true" +/- score.
-- Take every unique collection of 10 guys on the court, look at the actual +/-, and use that as an estimate of the overall sum/difference of the 10 guys' true stats.
-- Fit this data with the best estimates for "true +/-" you mathematically can.

Weighting all that for minutes played would be straightforward. Weighting it for various late-game/end-of-quarter tactics is less so, but such flaws are shared by almost all statistical approaches, so let's ignore it.

Why doesn't that work?


You just described the basic ideas behind APM: http://www.82games.com/comm30.htm. Though, I don' understand your last bulletin point.

Here is an extension of APM, RAPM or regularized APM at Sloan 2010: http://www.sloansportsconference.com/wp-content/uploads/2015/09/joeSillSloanSportsPaperWithLogo.pdf

APM shouldn't really be used anymore. RAPM or ESPN's RPM is superior.


Thanks!

The bullet point you were asking about arose because it was suggested in this thread that one solve a series of linear equations in which there were many more equations than variables. That can't actually be done. Rather, one looks for some kind of best fit.

The paper you linked me to suggests that the preferred way to do that is linear regression. I'm guessing that the independent variables are 0s and 1s, with any "data reading" having exactly 10 1s (for the 10 players on the court), and the regression coefficients become each player's estimated APM.

(I just skimmed the first couple of pages of the paper for now.)

Return to Statistical Analysis