PDA

View Full Version : PER is a funny stat



NYK All the Way
02-25-2009, 02:03 AM
It's weird how people say PER is the best way to judge a players performance. Tonight in the Rockets, Blazers game Brent Barry went 0-3 from the field had 0 points, 1 board, and 1 assist and his PER was +10.

In the Bobcats and Suns game Boris Diaw went 10-12 from the field put up 27points had 10 boards 6 assists 1 steal and 2 blocks...his PER was -3

Judging from comparisons like these on a daily basis I really don't see how PER is the best way to measure how good a player is. Anyone else agree?

GoatMilk
02-25-2009, 02:05 AM
I agree with you. I think it's a good stat to look at sometimes, but you can't completely judge a player on it.
John Hollinger swears by this stuff. Dude's a goon

blackjack_119
02-25-2009, 02:13 AM
You are thinking of "+/-" not PER.

DamnGoat
02-25-2009, 02:14 AM
Yeah, that's not PER, that's +/-.

The +/- is completely useless IMO, PER does make some sense though.

NYK All the Way
02-25-2009, 02:15 AM
You are thinking of "+/-" not PER.

some one told me that +/- was PER...what's PER then

madiaz3
02-25-2009, 02:17 AM
http://sports.espn.go.com/nba/columns/story?columnist=hollinger_john&id=2850240

Doesn't really capture the intangibles or any sort of defensive play other than steals and blocks.

abe_froman
02-25-2009, 02:17 AM
some one told me that +/- was PER...what's PER then

http://sports.espn.go.com/nba/columns/story?columnist=hollinger_john&id=2850240

blackjack_119
02-25-2009, 02:19 AM
The +/- calculates how many points your team improves or loses while you are on the court. It is a pretty useless statistic because it lumps the impact of all ten players on the floor and applies it to each individual player. You can take a scrub and put him on the floor with LeBron and Dwight Howard and your +/- will likely be through the roof because of their contribution despite your impact not being a major contribution.

Likewise, if LeBron is playing a game of 1 on 5, he could have 100 points on 50-50 shooting and have a +/- of zero because they will score on him every play down on the other end.

NYK All the Way
02-25-2009, 02:21 AM
http://sports.espn.go.com/nba/columns/story?columnist=hollinger_john&id=2850240

Doesn't really capture the intangibles or any sort of defensive play other than steals and blocks.


...that explains it maybe PER is a good way to measure stats then +/- not so much

blackjack_119
02-25-2009, 02:24 AM
some one told me that +/- was PER...what's PER then

PER calculates the efficiency of a player on a per minute average. It is an effective way of comparing the contribution of starters logging big minutes to reserve players who play few minutes. In addition to calculating everything on a per minute average, PER adjusts for other factors that skew statistics such as pace. Players in a D'Antoni system may put up huge statistics, but since they play at such a fast pace, the stats are reduced to what they would be if they played in a system with an average pace.

Hollinger has admitted that PER does a poor job of incorporating defense since there are very few defensive stats that are currently measured as a result, players like Shane Battier and Bruce Bowen routinely rank poorly using PER.

DenButsu
02-25-2009, 04:40 AM
Yeah, that's not PER, that's +/-.

The +/- is completely useless IMO, PER does make some sense though.

I disagree. You do, of course, have to interpret it with an eye not only on minutes played but the quality (garbage time? or not?) of those minutes, and also who was on the floor at the time (for example Dahntay Jones often gets a decent +/- that looks better than how bad he actually sucks because he starts alongside CB, Melo, K-Mart and Nene - if he came off the bench with the 2nd unit his +/- would take a major nosedive).

But actually, people are talking about PER's insufficiency in measuring defensive contributions (and other intangibles), and that is something that +/- can do to a certain extent. For example, on the Nuggets, a guy like Kenyon Martin who is not there to pad his stat sheet but to be the defensive captain, often has a stat line that looks less than impressive standing alone. But he also often gets one of the highest +/- scores on the team because the Nuggets defense takes a hit when he leaves the floor and teams start having more success scoring on us.

So it can be a useful barometer at times, but you have to know what it means and take into account who players are on the court with and whether the minutes they're playing are truly competitive minutes or garbage time.

blackjack_119
02-25-2009, 06:38 AM
I disagree. You do, of course, have to interpret it with an eye not only on minutes played but the quality (garbage time? or not?) of those minutes, and also who was on the floor at the time (for example Dahntay Jones often gets a decent +/- that looks better than how bad he actually sucks because he starts alongside CB, Melo, K-Mart and Nene - if he came off the bench with the 2nd unit his +/- would take a major nosedive).

But actually, people are talking about PER's insufficiency in measuring defensive contributions (and other intangibles), and that is something that +/- can do to a certain extent. For example, on the Nuggets, a guy like Kenyon Martin who is not there to pad his stat sheet but to be the defensive captain, often has a stat line that looks less than impressive standing alone. But he also often gets one of the highest +/- scores on the team because the Nuggets defense takes a hit when he leaves the floor and teams start having more success scoring on us.

So it can be a useful barometer at times, but you have to know what it means and take into account who players are on the court with and whether the minutes they're playing are truly competitive minutes or garbage time.

This is exactly why the stat is absolutely worthless. Looking at the stat tells you absolutely nothing. You have to either watch the game in conjunction with the stat or review the play-by-play in the game log to determine whether you can use the stat. In many circumstances, the plus minus is completely inaccurate. So if you look at a stat and can't be sure if it tells you anything, it is not a very good stat in my opinion.

Kyle916
02-25-2009, 06:46 AM
+/- doesn't account for potential defensive inefficiencies of the other 4 players on the court, nor does it account for offensive capabilities of the other 4 players on the court.

It's not completely useless, but it has major shortcomings.

DenButsu
02-25-2009, 07:32 AM
This is exactly why the stat is absolutely worthless.

No, it's why it's easily misused and misinterpreted, but as I explained above it can be a valid performance indicator as long as you understand it.

Chronz
02-25-2009, 07:58 AM
+/- doesn't account for potential defensive inefficiencies of the other 4 players on the court, nor does it account for offensive capabilities of the other 4 players on the court.

It's not completely useless, but it has major shortcomings.

They have adjusted +/- for that, but it takes years of data to get an accurate assessment of, and teams usually dont have that kind of time to wait on making a decision regarding a player and by then they may not be the same player anymore. Still its good enough that Morey landed a job as a GM because hes an expert at dissecting these numbers, hes been able to find several diamonds in the ruff come draft time with late lottery picks based heavily on statistical markers, I trust his opinion on stats so much that Ill take his opinion over yours when it comes to the validity of +/-.

Yes its not perfect, no stat ever will be. +/-'s positives outweigh its negatives.

JordansBulls
02-25-2009, 09:31 AM
some one told me that +/- was PER...what's PER then

No they are not the same. +/- means what happens on the court. It means the team benefited and did well why you were on the court, not necessarily that you did well yourself.

IndyRealist
02-25-2009, 01:17 PM
Wins produced per 48 minutes is a far better estimation of a player's performance. If you're a stat junkie, this is the one to look at.

http://dberri.wordpress.com/

An average player is .100 WP/48, an exceptional player .200, and an out-of-this-world player .300+
A couple of samples taken from the blog:

Kevin Garnett: .291
Lebron James: .409
Allen Iverson: .094
Chauncey Billups: .210
Nene: .211
Trevor Ariza: .271
Dwayne Wade: .350
Chris Paul: .493
Brad Miller: .163
Chris Bosh: .230
Jermaine O'Neal: .010
Jason Kopono: -.047 (Yep, that's negative wins produced)

Some of these (especially the Toronto players) are taken from a very small sample of games, since Dave Berri takes the time to write about every team.

Afridi786
02-25-2009, 03:15 PM
Likewise, if LeBron is playing a game of 1 on 5, he could have 100 points on 50-50 shooting and have a +/- of zero because they will score on him every play down on the other end.
I know its hypothetical but still. :rolleyes::rolleyes::rolleyes::rolleyes::rolleyes: :rolleyes::rolleyes::rolleyes:

DitchDat
02-25-2009, 04:18 PM
I don't think you should judge players by it, but you can definitely use it to see what lineups work.

Lost Art
02-25-2009, 05:34 PM
PER is weighted according to John Hollingers preference of which stats are most important. If weighting those stats is John Hollingers domain, then PER is a perfect way of judging a players worth.........if not, its pretty bogus. You decide.

theuuord
02-25-2009, 05:42 PM
PER is weighted according to John Hollingers preference of which stats are most important. If weighting those stats is John Hollingers domain, then PER is a perfect way of judging a players worth.........if not, its pretty bogus. You decide.

Well the weights aren't just arbitrarily decided by Hollinger. It's a statistical model based on study. There's very little (if any) subjectivity involved.

Lost Art
02-25-2009, 05:45 PM
Well the weights aren't just arbitrarily decided by Hollinger. It's a statistical model based on study. There's very little (if any) subjectivity involved.

So you're going to tell me that there is science behind how much a point is worth in comparison to a rebound? Or a rebound to an assist? And you're also going to tell me that these weights hold true in every single game? Use your eyes, its the best way to tell a players impact on the game.

JordansBulls
02-25-2009, 05:49 PM
So you're going to tell me that there is science behind how much a point is worth in comparison to a rebound? Or a rebound to an assist? And you're also going to tell me that these weights hold true in every single game? Use your eyes, its the best way to tell a players impact on the game.

You still need aggregate measurements to know how valuable a player is.

Like how do you determine who has the best stats between a guy who is getting 25 ppg, 10 rpg , 3 apg, 2 bpg, 1 spg on 55% fg vs a guy who is getting 28 ppg, 7 rpg, 5 apg, 1 bpg, 2 spg on 49% fg

You use PER to decide that. It is a statistic tool used to compare players across eras.

theuuord
02-25-2009, 06:02 PM
So you're going to tell me that there is science behind how much a point is worth in comparison to a rebound? Or a rebound to an assist? And you're also going to tell me that these weights hold true in every single game? Use your eyes, its the best way to tell a players impact on the game.

Yes, there is. You can use correlative analysis to try to further understand how much (for example) a defensive rebound is worth over the course of a basketball game, and how much defensive rebounds directly correlate to wins and winning percentage. Of course not every defensive rebound is equal, but that's why you take an average.

PER is a solid indicator of past performance, and usually a good indicator of future success as well.

Chronz
02-25-2009, 08:56 PM
PER is weighted according to John Hollingers preference of which stats are most important. If weighting those stats is John Hollingers domain, then PER is a perfect way of judging a players worth.........if not, its pretty bogus. You decide.
The concept behind PER has been around far longer than Hollinger, he just popularized it and improved upon its basic principles.

prash
02-25-2009, 08:57 PM
I think PER and +/- are both craptastic.

Chronz
02-25-2009, 09:12 PM
good one, your so funny man

agnine
02-26-2009, 07:34 PM
Wins and loses, then chemistry. That's how you judge players and teams.

theuuord
02-26-2009, 07:52 PM
Wins and loses, then chemistry. That's how you judge players and teams.

How do you judge "chemistry"?

Hawkeye15
02-26-2009, 08:04 PM
PER is weighted according to John Hollingers preference of which stats are most important. If weighting those stats is John Hollingers domain, then PER is a perfect way of judging a players worth.........if not, its pretty bogus. You decide.

is that because Kobe ranks 6th? If he ranked #1, would you be a PER fan? I think you would honestly. Salivating over you're response, since I have said forever that watching with you're own 2 eyes means more than any stat, but if Kobe leads a certain stat, his fans love it. If not, they think its dirt.