ESPN adds new overall stat...REAL PLUS/MINUS..Finally!

[John Long]

Nick Collison = biggest All Star snub this year.

Seriously, if I invented a stat that told me that Matt Barnes was better than Melo, that James Harden was the 2nd most important guard his own team, that Marcin Gortat was worth more than Brow and Cousins combined, and that Channing Frye was amongst the league's elite...I'd be too embarrassed to put it up.

James Harden is atrocious defensively, but I don't see where the stat says he is the 2nd most important guard on his own team

[criteriado]

milwaukee bucks individual WAR sums up to around 1 or 2 - plausible only if you think a D league all-star team could win double digit games in the NBA

chicago bulls individual WAR sums to around 31 whereas they have won 45 games

d.j. augustin WAR rank 425 out of 435, indicating he has been one of the most damaging players in the league to his team, this despite the reality that he has been a savior for the chicago bulls

i don't question the math here. i question the application of the math in measuring a basketball player's net impact on his team. at the very least there are a number of substantial outliers

How is DJ Augustin a saviour when his team is much better when he's off the floor than with him on the floor? Seriously.

[mysticbb]

could someone sum it up real simple..??? i am a little confused how it works

The numbers have to be read as: Player x changes the result per 100 poss by y amount of points, if y is the RPM value. An average player has 0. Right now the league average rating per 100 poss is about 106.5. Meaning, an average team is scoring 106.5 points in 100 possession, while they also allow 106.5 points in 100 poss. If we take Anthony Davis here as an example, his values means that with him on the court and 4 average players, the expected result would be +1.09 over average on offense (rounding to 1.1) and +0.52 over average on defense (rounded to 0.5). Translating that to the team result would mean, that a team would be expected to score 107.6 points per 100 poss, while they would allow 106.0 points per 100 poss (a value for defense has to be subtracted), therefore creating a scoring margin per 100 poss of 1.6 points (or 1.61 points when the accurate numbers are used, like we see his value at RPM is). Hope that helps to interpret the numbers.

Obviously, someone playing more minutes than another player adds value. Also, the RPM value might be heavily effected by the role a player is assigned to on a team, while a different role might change the value in one way or another. Usually players are following a development curve over their career, but rather staying within a small range at their respective "prime".

The numbers are derived via regression analysis; that is a method used to determine the impact from various variables on a result. In that case each player on the court at a certain time (teammates as well as opponents) is seen as a variable and the overall scoring margin would be the result (adjusted for 100 poss that is). The method in case of RPM is ridge regression, which helps to get better predictive values; which comes with the price of getting a worse explaination (which is why some people adding up those WAR values aren't getting the exact number of wins for each team). The "creator" also uses a baysian distribution, adjusts for coaching differences, applies such a development curve as well as uses prior values for each player based on boxscore stats in order to achieve an even better prediction. But to understand all of that, one post is not really enough.

I hope that helped a little bit ...

[daniel3]

Nick Collison = biggest All Star snub this year.

Seriously, if I invented a stat that told me that Matt Barnes was better than Melo, that James Harden was the 2nd most important guard his own team, that Marcin Gortat was worth more than Brow and Cousins combined, and that Channing Frye was amongst the league's elite...I'd be too embarrassed to put it up.

I'm new the stat but even I saw that the stat doesn't claim what you're saying it does. Seems you just look at one aspect and went off with that without reviewing it as a whole.

If you want to hang onto one stat to make it easy on yourself, look at WAR.

[daniel3]

RK NAME TEAM GP MPG ORPM

435 Kendrick Perkins, C OKC 56 19.5 -6.19

Seems to be working.

He is only player on OKC that actually affects their winning percentage negatively.

[crazy_me_87]

could someone sum it up real simple..??? i am a little confused how it works

The numbers have to be read as: Player x changes the result per 100 poss by y amount of points, if y is the RPM value. An average player has 0. Right now the league average rating per 100 poss is about 106.5. Meaning, an average team is scoring 106.5 points in 100 possession, while they also allow 106.5 points in 100 poss. If we take Anthony Davis here as an example, his values means that with him on the court and 4 average players, the expected result would be +1.09 over average on offense (rounding to 1.1) and +0.52 over average on defense (rounded to 0.5). Translating that to the team result would mean, that a team would be expected to score 107.6 points per 100 poss, while they would allow 106.0 points per 100 poss (a value for defense has to be subtracted), therefore creating a scoring margin per 100 poss of 1.6 points (or 1.61 points when the accurate numbers are used, like we see his value at RPM is). Hope that helps to interpret the numbers.

Obviously, someone playing more minutes than another player adds value. Also, the RPM value might be heavily effected by the role a player is assigned to on a team, while a different role might change the value in one way or another. Usually players are following a development curve over their career, but rather staying within a small range at their respective "prime".

The numbers are derived via regression analysis; that is a method used to determine the impact from various variables on a result. In that case each player is seen as a variable and the overall scoring margin would be the result. The method in case of RPM is ridge regression, which helps to get better predictive values; which comes with the price of getting a worse explaination (which is why some people adding up those WAR values aren't getting the exact number of wins for each team). The "creator" also uses a baysian distribution, adjusts for coaching differences, applies such a development curve as well as uses prior values for each player based on boxscore stats in order to achieve an even better prediction.

I hope that helped a little bit ...

thanks dude

yeah it did help

but i dont see the use of it to be honest

i just cant translate it to my "common basketball sense" ^^

anthony davis for example leads the NBA in Blocks and is a good rebounder.. so by everything i know and experienced in nearly 20 years watching the NBA he cant he just cant be just an average NBA defender..sure he may not be amazing because he makes some mistakes because he is young but a player who blocks alot of shots,rebounds well and stays on the floor was a defensive anchor for as long as i remember...

it just seems to be a useless stat at least to me... what do i take from knowing that Hashem Thabeet is a Top five defensive guy by that rating?? ... nothing because i still cant play him many minutes because he sucks at everything else other than beeing huge and blocking some shots..so if i have him i can feel assured i have a defensive gem for 9 minutes a game?

i might sound a bit too sarcastic here and i am sorry for that but what good is a new Stat that doesnt really help me compare players or draw new conclusions

i already knew by every other stat already that Paul George is awfully mediocre on offense and a little bit overrated on defence

i already knew that Lebron is as good as ever offensivly but lacks a bit this season defensivly

so where is the "ah thats new,i didnt knew that" hype that some here have? i just dont see why some here are so excited.. sure its nice to have one more stat but is anything special about this one?

[dice]

milwaukee bucks individual WAR sums up to around 1 or 2 - plausible only if you think a D league all-star team could win double digit games in the NBA

chicago bulls individual WAR sums to around 31 whereas they have won 45 games

d.j. augustin WAR rank 425 out of 435, indicating he has been one of the most damaging players in the league to his team, this despite the reality that he has been a savior for the chicago bulls

i don't question the math here. i question the application of the math in measuring a basketball player's net impact on his team. at the very least there are a number of substantial outliers

How is DJ Augustin a saviour when his team is much better when he's off the floor than with him on the floor? Seriously.

because the team is NOT better when he's off the floor

bulls w/o augustin:
8-12

bulls w/ augustin:
37-20

obviously he's only a part of the story, but anyone who has consistently watched the chicago bulls this year recognizes that a good many games have been salvaged by d.j. augustin. it's plain as day that his contributions on the offensive end have been invaluable. and his defense has been respectable. both the eye test and the stats confirm all of this

meanwhile, luol deng has quite a respectable WAR on the season despite playing poorly for the cavs. he was dealt there for draft picks

bulls w/ deng:
10-13

bulls w/o deng:
35-19

treat RAPM as a guide, not a religion

[mysticbb]

i just cant translate it to my "common basketball sense" ^^

Well, that is good that you can observe that as well, but the new thing here is: It is quantified, not just based on some sort of "eye test", but real data (real game action) is taken and converted into a measurement, which then can be used to predict results. For some players it might be a hint that they may be better than someone thought, others might be a bit worse than you thought. If a value is not in-line with your "common basketball sense", you may want to investigate the player further and focus on things which could be seen as positive/negative contribution.

As long as you don't act like the stats would tell you everything you need to know, you should be good and the stats will maybe help/motivate you to understand basketball better. And like a lot of thing this stat will likely not get it right for every player, but the majority of players should have a value at least close to the actual value they present (otherwise the prediction wouldn't be good).

[schnakenpopanz]

this states the obvious goran dragic was the biggest all star snub this season

[Winsome Gerbil]

Nick Collison = biggest All Star snub this year.

Seriously, if I invented a stat that told me that Matt Barnes was better than Melo, that James Harden was the 2nd most important guard his own team, that Marcin Gortat was worth more than Brow and Cousins combined, and that Channing Frye was amongst the league's elite...I'd be too embarrassed to put it up.

I'm new the stat but even I saw that the stat doesn't claim what you're saying it does. Seems you just look at one aspect and went off with that without reviewing it as a whole.

If you want to hang onto one stat to make it easy on yourself, look at WAR.

WAR is just the final punchline at the end of the tunnel. Stat people sometimes have a tendency to spiral down rabbit holes of their own creation. So you create RPM. Basing it on earlier stats. Then you create WAR based upon RPM. With each step you try to add further legitimacy upon your stat by masking its origins. Pretty soon DeAndre Jordan is the 7th most important player in the league this year.

And hey, its all just +/- numbers at its heart right? Except that +/- has always been a dicey and highly situation dependent measure of anything. Put Hakeem Olajuwon and David Robinson on the same team subbing for each other, and both of them look like entirely mediocre players. Make D'Antoni their coach, and their offensive numbers skyrocket, their defensive numbers fall, fire D'Antoni and hire Pat Riley and all of a sudden their offensive numbers fall, and their defensive numbers soar. Make Charles Oakley a starter next to them, and watch their defensive stats get better, while their rebounding stats get worse. Put Channing Frye next to them and watch the opposite happen. Give them John Stockton and watch their FG% shoot up. Given them Jamaal Crawford instead and watch them have to create it all themselves.

I know all that. Anybody who knows the game knows all that. No stat knows all that. You won't find me hiding behind one and parading it out as ultimate truth.

[criteriado]

milwaukee bucks individual WAR sums up to around 1 or 2 - plausible only if you think a D league all-star team could win double digit games in the NBA

chicago bulls individual WAR sums to around 31 whereas they have won 45 games

d.j. augustin WAR rank 425 out of 435, indicating he has been one of the most damaging players in the league to his team, this despite the reality that he has been a savior for the chicago bulls

i don't question the math here. i question the application of the math in measuring a basketball player's net impact on his team. at the very least there are a number of substantial outliers

How is DJ Augustin a saviour when his team is much better when he's off the floor than with him on the floor? Seriously.

because the team is NOT better when he's off the floor

bulls w/o augustin:
8-12

bulls w/ augustin:
37-20

obviously he's only a part of the story, but anyone who has consistently watched the chicago bulls this year recognizes that a good many games have been salvaged by d.j. augustin. it's plain as day that his contributions on the offensive end have been invaluable. and his defense has been respectable. both the eye test and the stats confirm all of this

meanwhile, luol deng has quite a respectable WAR on the season despite playing poorly for the cavs. he was dealt there for draft picks

bulls w/ deng:
10-13

bulls w/o deng:
35-19

treat RAPM as a guide, not a religion

I'm using 82games.com and on/off the court data. Not RAPM. He is an average replacement player but he replaced an awful player, so he looks better and the record looks better.

He looks great on PER and awful at RPM, it evens out.

It's the same argument that Jazz fans argue with Burke on his campaign for ROY. Burke replaced an awful player and the Jazz were much better even though he's a below average player.

[ceiling raiser]

Nick Collison = biggest All Star snub this year.

Seriously, if I invented a stat that told me that Matt Barnes was better than Melo, that James Harden was the 2nd most important guard his own team, that Marcin Gortat was worth more than Brow and Cousins combined, and that Channing Frye was amongst the league's elite...I'd be too embarrassed to put it up.

That's not what the stat is saying. Read MysticBBs post in this thread, that will probably make things clearer.

I'm very happy that ESPN are pushing this stat, I hope they will use it when they're doing games too so that people will be more familiar with advanced stats.

I'm hoping they don't, because a lot more would be learned if those people spent less time punching a calculator and more time watching the actual game.

In 2014, is this really necessary? Do you honestly believe that people interested in this type of data/analysis aren't watching the game? It's disturbing that you're opposed to people enriching their fan experience and understanding of players/the league.

[mysticbb]

obviously he's only a part of the story, but anyone who has consistently watched the chicago bulls this year recognizes that a good many games have been salvaged by d.j. augustin.

I watched basically all games of the Bulls, and I have to say that Augustin may add points offensively here and there, but overall he costs the Bulls more often points defensively. That the Bulls won more games with him is simply the result of being more healthy than before and being able to give more minutes to better players.

Also, the values are adjusted for coaches, and Thibodeau is seen as quite a great coach by that metric, which means Bulls players should be rated down a bit. But you might disagree with the assessment that Thibodeau is one of the better coaches in the league and Augustin is actually getting the value reduced is unfair ...

and his defense has been respectable. both the eye test and the stats confirm all of this

If you believe that Augustin's defense has been respectable, you either had a very, very low expectation or your "eye test" can't be taken seriously ...

bulls w/ deng:
10-13

bulls w/o deng:
35-19

Bulls with Deng on the court played better basketball than they played without him. If at the time Deng was still on the Bulls, the team overall had performed better without him, they would have had a better record overall (and Deng would probably have not been traded). Also, the Cavs also played better in average with Deng than without him even though it needed time until it worked.

[Ayt]

I'm hoping they don't, because a lot more would be learned if those people spent less time punching a calculator and more time watching the actual game.

I'm fairly certain that these people actually watches ALOT of basketball.

There's no need to dismiss the stat just because you don't agree with a couple of cherry picked examples.

It is more that he flat out doesn't understand it.

[dice]

How is DJ Augustin a saviour when his team is much better when he's off the floor than with him on the floor? Seriously.

because the team is NOT better when he's off the floor

bulls w/o augustin:
8-12

bulls w/ augustin:
37-20

obviously he's only a part of the story, but anyone who has consistently watched the chicago bulls this year recognizes that a good many games have been salvaged by d.j. augustin. it's plain as day that his contributions on the offensive end have been invaluable. and his defense has been respectable. both the eye test and the stats confirm all of this

meanwhile, luol deng has quite a respectable WAR on the season despite playing poorly for the cavs. he was dealt there for draft picks

bulls w/ deng:
10-13

bulls w/o deng:
35-19

treat RAPM as a guide, not a religion

I'm using 82games.com and on/off the court data. Not RAPM. He is an average replacement player but he replaced an awful player, so he looks better and the record looks better.

He looks great on PER and awful at RPM, it evens out.

It's the same argument that Jazz fans argue with Burke on his campaign for ROY. Burke replaced an awful player and the Jazz were much better even though he's a below average player.

1) everyone knows PER is a garbage stat. it doesn't average anything out
2) augustin has been a well above average replacement player
3) RAPM, which we're discussing in this thread and i have been referencing, is supposed to control for other factors and boil everything down to individual impact. it fails miserably in this particular case

[xfactor99]

Has anyone done a meta-study of how well RAPM, PER, Win Shares, O/DRating, etc predict future performance? (Maybe compared to regular stats as well) Where the R squared numbers at. Just curious.

Was never taught ridge regression in my stats classes, looking at it now. Looks useful.

[mysticbb]

Has anyone done a meta-study of how well RAPM, PER, Win Shares, O/DRating, etc predict future performance? (Maybe compared to regular stats as well) Where the R squared numbers at. Just curious.

Neil Paine done that:

Code: Select all

Metric   Y-1            Metric   Y-2            Metric   Y-3
-------------           -------------           -------------
RAPM    0.751           RAPM    0.646           RAPM    0.568
ASPM    0.723           ASPM    0.610           ASPM    0.532
WS/48   0.694           WS/48   0.547           PER     0.502
WP/48   0.654           PER     0.546           WS/48   0.494
PER     0.638           WP/48   0.492           WP/48   0.440

http://www.apbr.org/metrics/viewtopic.p ... 96&p=15343

Was never taught ridge regression in my stats classes, looking at it now. Looks useful.

Here are some references to original paper, which provides the underlying math:

Tikhonov, 1943: http://a-server.math.nsc.ru/IPP/BASE_WORK/tihon_en.html
Hoerl and Kennard, 1970: http://www.jstor.org/discover/10.2307/1 ... 2215866301
Chawla, 1988: http://www.sciencedirect.com/science/ar ... 5288900399

Maybe that is helpful.

[PaulieWal]

1) everyone knows PER is a garbage stat.

Lol PER is a garbage stat now?

[dice]

obviously he's only a part of the story, but anyone who has consistently watched the chicago bulls this year recognizes that a good many games have been salvaged by d.j. augustin.

I watched basically all games of the Bulls, and I have to say that Augustin may add points offensively here and there

but overall he costs the Bulls more often points defensively

perhaps marginally

That the Bulls won more games with him is simply the result of being more healthy than before and being able to give more minutes to better players.

health is part of it for sure. but there's no getting around the fact that the bulls have been far more successful with augustin on the roster than they were with deng. jimmy butler getting healthier accounts for only a fraction of that

Also, the values are adjusted for coaches, and Thibodeau is seen as quite a great coach by that metric, which means Bulls players should be rated down a bit

he's a great coach, but he must be the best coach in the history of the league if the bulls are outperforming their individual WARs by that much. the spurs aren't that far out of whack, and popovich is pretty universally considered an excellent coach

and i don't think WAR is supposed to work like that anyway. it relies solely on what has happened on the court and distributes that value to players

and his defense has been respectable. both the eye test and the stats confirm all of this

If you believe that Augustin's defense has been respectable, you either had a very, very low expectation or your "eye test" can't be taken seriously ...

i guess synergy's "eye test" can't be taken seriously either

bulls w/ deng:
10-13

bulls w/o deng:
35-19

Bulls with Deng on the court played better basketball than they played without him. If at the time Deng was still on the Bulls, the team overall had performed better without him, they would have had a better record overall (and Deng would probably have not been traded). Also, the Cavs also played better in average with Deng than without him even though it needed time until it worked.

deng was salary dumped for future considerations. he was indeed a positive for the team, which is just more evidence that losing him and bringing in augustine should not have resulted in the team taking off given the supposed huge negative WAR impact that the move had

and deng started hot for the cavs and fell off hard

[dice]

1) everyone knows PER is a garbage stat.

Lol PER is a garbage stat now?

always has been. which is widely recognized amongst those who do not use the acronym 'lol'