Question for stat guys about PER....

[REHawksFan]

Has anyone broken down the PER formula to figure out what has the most influence?  I was perusing some stats this morning and noticed the following:

Player A:   37.2 mpg / 44.8 FG% / 37.5 3PT% / 86.2 FT% / 6.2 rpg / 7.5 apg / 1.7 spg / 4.7 TOpg / 13 DD / 2 TD / PPS = 1.54 / PER = 31.4

Player B:   35.2 mpg / 44.2 FG% / 36.9 3PT % / 84.5 FT% / 4.5 rpg / 8.5 apg / 1.2 spg / 4.8 TOpg / 13 DD / 2 TD / PPS = 1.38 / PER = 23.5

 

Player A clearly has superior stats, but they magnitude of difference is not that great. IF Player Efficiency Ratio is supposed to measure the per game efficiency of each player, why is there such a difference in PER when their normal efficiency stats are generally similar. Does it factor in volume of shots? Is the biggest difference PPS (attributable to FT frequency and 3PA differences)?  

 

*In case anyone hadn't figured it out yet, Player A is Harden and Player B is Trae. 

[enrique]

18 minutes ago, REHawksFan said:

Has anyone broken down the PER formula to figure out what has the most influence?  I was perusing some stats this morning and noticed the following:

Player A:   37.2 mpg / 44.8 FG% / 37.5 3PT% / 86.2 FT% / 6.2 rpg / 7.5 apg / 1.7 spg / 4.7 TOpg / 13 DD / 2 TD / PPS = 1.54 / PER = 31.4

Player B:   35.2 mpg / 44.2 FG% / 36.9 3PT % / 84.5 FT% / 4.5 rpg / 8.5 apg / 1.2 spg / 4.8 TOpg / 13 DD / 2 TD / PPS = 1.38 / PER = 23.5

 

Player A clearly has superior stats, but they magnitude of difference is not that great. IF Player Efficiency Ratio is supposed to measure the per game efficiency of each player, why is there such a difference in PER when their normal efficiency stats are generally similar. Does it factor in volume of shots? Is the biggest difference PPS (attributable to FT frequency and 3PA differences)?  

 

*In case anyone hadn't figured it out yet, Player A is Harden and Player B is Trae. 

PER was never meant to provide comparison between 2 players who play different positions. If you run the average PER per position (on a side note, that expression really looks weird...PER per...but any who...), you will find that rebounding heavily influences it. Note the average C or PF PER will be higher than the other positions. All that being said, when you have a Unicorn (e.g., a Center that distributes the ball well, a PG who rebounds well, etc.) you will see some really strange numbers pop up. 

PER is simply one stat to use among several in evaluating people. To my knowledge Rebounding is overemphasized by the formula. 

Here is an article from a few years ago that is helpful... https://www.huffpost.com/entry/nba-statistics_b_5190651

 

[enrique]

Here is the formula...I knew it was out there:)

https://www.basketball-reference.com/about/per.html

[enrique]

Here is a nice comparison I just ran on point guards in the NBA

https://www.espn.com/nba/stats/player/_/position/point-guard/table/general/sort/PER/dir/desc

Note that Dame is slightly higher than Trae. They post almost identical percentages and numbers across the board. Trae is slightly higher in PPG and APG. His larger TO numbers however reduce his PER so that Dame is listed above him on PER. 

PERsonally I don't mind PER when used this way. It is basing the numbers on a clear formula that isn't completely objective, but it attempts to balance possessions and usage alongside the traditional box scores.

[lethalweapon3]

Noting, the individual player's stats are adjusted by team Pace. Atlanta's is back in the top 10, but Houston's is second only to Milwaukee. I believe, the larger the team's Pace, relative to the league average, the lower the adjusted PER becomes. (i.e., theorizing: the Hawks' lower pace makes this particular gap between the Players closer, rather than wider).

Perhaps making the gap wider: the ratio of team-wide assists per team-wide FGs made (the larger that ratio, the lower the unadjusted PER). As you might imagine while watching the Beard, Atlanta has a higher ratio than Houston there.

(speaking of Portland, in reference to Dame, they're the only team with a lower assist ratio than Houston)

They're not all that different in terms of DReb%, but that seems to be a huge multiplier in how PER gets calculated. Less impactful, but still significant, are steals and blocks, where Player A totals about twice the volume of those stats combined. It's probably why centers like Whiteside are huge PER guys while also leading the league in making fans tear their hair out.

Threes made is a direct factor in unadjusted PER, where Player A holds the edge.

 

~lw3

[JayBirdHawk]

This is some good stuff. Thanks guys.

[NBASupes]

PER is nonsense. 

[enrique]

1 minute ago, NBASupes said:

PER is nonsense. 

As with all stats, player comparisons are like examining a diamond...if you only consider one facet you might overvalue or undervalue a particular diamond. I look at metrics such as PER, RAPM, VORP, etc. as differing ways of viewing player performance...different facets, if you will. Simple box score tallies are another facet. Taking all of them into consideration will give us a better perspective IMO of a player's value and contribution. All of that being said, the downfall of all stats in the NBA (except FT%) is that they are performed within the context of teammates of varying degrees of quality that operate amongst a league of teams of varying degrees of quality that operates within an overall league-era of varying degrees of quality. That is why there will NEVER be an open and closed statistical case for any player over another. There will always be some debate. 

Personally i don't mind that as long as people are cordial about it and realize the limits of what we have. It's part of what makes basketball fun to watch for me:)

[NBASupes]

WAR and the RAPTOR are the best ones I've seen in terms of impact in their role and situation. That's all that matters to me. Too many of these value things that I might not value as much. Winning should be a prime objective. 

[REHawksFan]

15 minutes ago, NBASupes said:

WAR and the RAPTOR are the best ones I've seen in terms of impact in their role and situation. That's all that matters to me. Too many of these value things that I might not value as much. Winning should be a prime objective. 

Winning is the ultimate objective obviously. But there has to be a way to isolate a single player's impact on winning if we are to compare player vs player in that regard. Maybe that's WAR. Maybe Raptor, maybe PER or RPM.  I agree with @enrique in that there's probably not one "catch all" stat that paints the entire picture. We should look at all of the them to formulate a well rounded opinion. 

I was really just curious why players with somewhat similar box stats had what appeared to be disparate PER.  

As to your point:

Trae plays 35 mpg for a team that has 9 wins.

Eric Bledsoe plays 26 mpg for a team with 35 wins

Would you suggest Bledsoe has a bigger impact on 35 wins for his team than Trae has for the 9 wins for his? 

[HawkItus]

What about EPI?

[enrique]

55 minutes ago, REHawksFan said:

Trae plays 35 mpg for a team that has 9 wins.

Eric Bledsoe plays 26 mpg for a team with 35 wins

Would you suggest Bledsoe has a bigger impact on 35 wins for his team than Trae has for the 9 wins for his? 

That is the real downside to WinShares for sure...there is no player in the world that could play on the Hawks and have a very high WS this season. So using it to compare players is more like saying Player A is the 3rd best player on the team with the 5th best record in the NBA, rather than saying Player A is better than Player B. Player B may be the best player on a team that is really not going to win much at all this current season.  

Trae currently has 37% of the Hawks' win shares.

Bledsoe currently has 10% of the Bucks' win shares.

What can you conclude from that about the individual players? Not a hell of a lot except that Trae is more vital to the Hawks than Bledsoe is to the Bucks...

 

Edited January 15, 2020 by enrique

[Atlantaholic]

3 hours ago, REHawksFan said:

Has anyone broken down the PER formula to figure out what has the most influence?  I was perusing some stats this morning and noticed the following:

Player A:   37.2 mpg / 44.8 FG% / 37.5 3PT% / 86.2 FT% / 6.2 rpg / 7.5 apg / 1.7 spg / 4.7 TOpg / 13 DD / 2 TD / PPS = 1.54 / PER = 31.4

Player B:   35.2 mpg / 44.2 FG% / 36.9 3PT % / 84.5 FT% / 4.5 rpg / 8.5 apg / 1.2 spg / 4.8 TOpg / 13 DD / 2 TD / PPS = 1.38 / PER = 23.5

 

Player A clearly has superior stats, but they magnitude of difference is not that great. IF Player Efficiency Ratio is supposed to measure the per game efficiency of each player, why is there such a difference in PER when their normal efficiency stats are generally similar. Does it factor in volume of shots? Is the biggest difference PPS (attributable to FT frequency and 3PA differences)?  

 

*In case anyone hadn't figured it out yet, Player A is Harden and Player B is Trae. 

The big difference is obviously that Player A gets to the line much more often than player 2 which leads to a much higher pps/TS%

[NBASupes]

1 hour ago, REHawksFan said:

Winning is the ultimate objective obviously. But there has to be a way to isolate a single player's impact on winning if we are to compare player vs player in that regard. Maybe that's WAR. Maybe Raptor, maybe PER or RPM.  I agree with @enrique in that there's probably not one "catch all" stat that paints the entire picture. We should look at all of the them to formulate a well rounded opinion. 

I was really just curious why players with somewhat similar box stats had what appeared to be disparate PER.  

As to your point:

Trae plays 35 mpg for a team that has 9 wins.

Eric Bledsoe plays 26 mpg for a team with 35 wins

Would you suggest Bledsoe has a bigger impact on 35 wins for his team than Trae has for the 9 wins for his? 

WS is garbage too. 

[NBASupes]

For me: 

It's 

1. RAPTOR

2. WAR

3. RPM

4. PIE
 

Mostly, I just use stats on NBA.com advanced sections anyway. I think @TheNorthCydeRises who also uses it does a good job with aggregating the data. 

[Atlantaholic]

6 hours ago, lethalweapon3 said:

Noting, the individual player's stats are adjusted by team Pace. Atlanta's is back in the top 10, but Houston's is second only to Milwaukee. I believe, the larger the team's Pace, relative to the league average, the lower the adjusted PER becomes. (i.e., theorizing: the Hawks' lower pace makes this particular gap between the Players closer, rather than wider).

Perhaps making the gap wider: the ratio of team-wide assists per team-wide FGs made (the larger that ratio, the lower the unadjusted PER). As you might imagine while watching the Beard, Atlanta has a higher ratio than Houston there.

(speaking of Portland, in reference to Dame, they're the only team with a lower assist ratio than Houston)

They're not all that different in terms of DReb%, but that seems to be a huge multiplier in how PER gets calculated. Less impactful, but still significant, are steals and blocks, where Player A totals about twice the volume of those stats combined. It's probably why centers like Whiteside are huge PER guys while also leading the league in making fans tear their hair out.

Threes made is a direct factor in unadjusted PER, where Player A holds the edge.

 

~lw3

Whiteside has been great for Portland this year. You can look just at their record and conclude that Whiteside has been a bust but without him they would be the Hawks right now. Outside of Lillard (and Whiteside) everyone has been awful including McCollum. Their inside players outside of Whiteside are a travesty, their wings aren't much better since Hood went down. Portland's defense has not been good this year but their defensive rating with Whiteside on the floor is excellent, and he leads the league in blocks. I would take Whiteside in a heartbeat on this team. 

[NBASupes]

6 minutes ago, Atlantaholic said:

Whitside has been great for Portland this year. You can look just at their record and conclude that Whitside has been a bust but without him they would be the Hawks right now. Outside of Lilliard (and Whitside) everyone has been awful including McCollum. Their inside players outside of Whitside are a travesty, their wings aren't much better since Hood went down. Portland's defense has not been good this year but their defensive rating with Whitside on the floor is excellent, and he leads the league in blocks. I would take Whitside in a heartbeat on this team. 

Portland's killer has been their bench and there biggest need is bringing back Nurkic. Nurk had a 10.1 WAR. Outside of Lillard, he is their best player by a mile. Losing Curry who had a good WAR for a bench player and replacing him with the worst backup PG in the NBA in Simons hasn't helped either. Their bench is just awful. Right now, so much of their producing is coming from Bazemore. The worst bench in the NBA. The Hawks bench is horrible too but that's mainly due to injuries. Len and Parker both have positive value and Bembry has a role. Outside of that, it's been extremely messy although Goodwin has helped recently and once again, injuries have played a role for Atlanta. 

 

Great maybe an overstatement but he's worth 3.3 wins right now in his current role. He has clearly helped Portland this season. 

Edited January 15, 2020 by NBASupes

[Atlantaholic]

1 minute ago, NBASupes said:

Portland's killer has been their bench and there biggest need is bringing back Nurkic. Nurk had a 10.1 WAR. Outside of Lillard, he is their best player by a mile. Losing Curry who had a good WAR for a bench player and replacing him with the worst backup PG in the NBA in Simons hasn't helped either. Their bench is just awful. Right now, so much of their producing is coming from Bazemore. The worst bench in the NBA. The Hawks bench is horrible too but that's mainly due to injuries. Len and Parker both have positive value and Bembry has a role. Outside of that, it's been extremely messy although Goodwin has helped recently and once again, injuries have played a role for Atlanta. 

I would take our bench (when healthy) over Portlands bench all day everyday. It's not even close. People forget Portland lost many pieces in the offseason including their two best defenders in Aminou and Harkless. Whiteside was supposed to be their stop-gap until Nurkic got healthy and he has done a tremendous job for them. No one in Portland is pulling their hair out about Whiteside. 

[NBASupes]

Just now, Atlantaholic said:

I would take our bench (when healthy) over Portlands bench all day everyday. It's not even close. People forget Portland lost many pieces in the offseason including their two best defenders in Aminou and Harkless. Whiteside was supposed to be their stop-gap until Nurkic got healthy and he has done a tremendous job for them. No one in Portland is pulling their hair out about Whiteside. 

Agreed. They really took several steps back. Losing key guys like Collins, Curry, Harkless, and Aminu have just killed them as well but let's be real, this is really about Nurkic. He was the star outside of Lillard. 

[Atlantaholic]

2 minutes ago, NBASupes said:

Agreed. They really took several steps back. Losing key guys like Collins, Curry, Harkless, and Aminu have just killed them as well but let's be real, this is really about Nurkic. He was the star outside of Lillard. 

A healthy Nurkic instead of Whitside would have them with roughly the exact same record this year. Doubt Nurkic would be giving them considerably more production and value than  Whiteside's 18ps 17rbs and 3.5blks per 36 mins on 63% TS%. I mean, that is all star production no doubt about it. Any Hawk fan that can sit there and say these are empty numbers and wouldn't take it is insane IMO. I'm sure you could ask Miami fans right now and they would gladly welcome Whiteside back with open arms, attitude problem notwithstanding.