thoughts on PER for the Rockets & the rest of the NBA

Currently there is only one rocket that is in the top 20 for his position in PER, and that is Carl Landry with a 23.48 PER (#3 among power forwards). The rockets are the only team in the NBA with only 1 player in the top 20 at his position in PER. Every other team has 2 or more, with Boston having 6 to lead all teams.

----------

There is one team with 1 player in the top 20 by position: Houston. Average win% for these teams is 56%.

There are seven teams with 2 players in the top 20 by position: Charlotte, Cleveland, Detroit, Miami, Minnesota, New Jersey, Oklahoma City. Average win% for these teams is 41%.

There are seven teams with 3 players in the top 20 by position: Chicago, Indiana, Los Angeles Lakers, Milwaukee, New Orleans, Toronto, and Utah. Average win% for these teams is 51%.

There are twelve teams with 4 players in the top 20 by position: Atlanta, Dallas, Denver, Golden State, Los Angeles Clippers, Memphis, New York, Orlando, Portland, Sacramento, San Antonio, and Washington. Average win% of these teams is 53%.

There are two teams with 5 players in the top 20 by position: Philadelphia and Phoenix. Average win% of these teams is 47%.

There is one team with 6 players in the top 20 by position: Boston. Average win% of these teams is 76%.

-----------

so its a little unclear as to whether having more high PER players makes a team win more as we would expect. there are a lot of outliers present, like our rockets who appear to only have one "very good" player yet are winning, and the sixers/clippers/wizards who seem to have multiple very good players but do not win. Clearly missing in all of this is that I don't have PERs for all the players on all the rosters, so as to be able to calculate an average PER by team and see how that correlates with win%. if someone wants to get me that (ESPN insiders) then I'd love to check it out.

anyway, I thought it was interesting and wanted to see what you guys thought.

 

it is true that PER does not factor in true team defense. but keep in mind that this year's rockets are not a good defensive team, ranking somewhere around 20th in the NBA in adjusted fg% against.

however, it is also true that PER does a great job of measuring a player's offensive efficiency and measurable production. for the rockets to be above .500 with only one player in the top 100 in PER in the NBA is really surprising. especially so when you consider that they are not very good defensively, so there's no argument that it's okay for them to be inefficient on offense, because they make up for it on the other end.

think about that again for a second, the rockets have only 1 player in the top 100 in the NBA in PER, yet they are above .500 despite playing the hardest schedule in the league. its amazing to me.

 

its not just about scoring. PER is an attempt to capture as much about a player's production and efficiency as possible in a single number.

here's a link to the calculation of PER if you're interested: basketball reference

the final number you get includes all of a player's productive contributions (like making field goals and free throws, collecting rebounds, blocking shots, steals, getting assists) except they are adjusted by negative contibutions as well (like missing field goals and free throws, turnovers), and its all weighted to league averages, pace factors, etc, per minute.

so in the end, it's supposed to represent the quality of the play of each individual player, per minute, relative to the rest of the league (league average PER is always 15.0).

 

i looked around and didn't find a complete 09-10 PER list anywhere. If you have a link and could post it, please do.

and yes, the sum/average PER for all players has more significance, but it also reveals the biggest flaw in PER, that it produces huge swings when minutes played for an individual shrinks. you end up with league leaders in PER as end-of-bench guys, so you'd have to introduce some intelligence into the calculation in order to effectively PER-ify a team.

this wasn't supposed to be a discussion of the merits of PER as much as simply pointing out how odd this year's rockets team is from the perspective of a metric that many people use to measure the quality of individual players.

 

yeah, see, I can't use the PERs if its all individualized like that. there are roughly 400 players in the NBA. i need it in a list or csv format.

and yes, i understand the team aspect. i mentioned that in the first post. but considering that PER is also an efficiency of production calculation, having a team full of high PER players (as long as they play some defense) should outperform a team of low PER players, just because the high PER players produce more measurable positive production per minute.

I think that if you summed the PER's of each team, you'd find that the total PER by team would correlate with some decent accuracy to win %. Boston would probably have the highest team PER. You can see this by comparing teams with equivalent numbers of top20 PER performers.

for example: among teams with 3 players in the top20 in PER - if you average the PER of those 3 players, and use that to rank the teams, you get a ranking that goes like this (actual NBA standings rank among this group is in parenthesis):

1) LAL (1)
2) NOR (3)
3) TOR (4)
4) UTH (2)
5) MIL (5)
6) IND (7)
7) CHI (6)

clearly you can see that my PER-based rankings for this group agree with the NBA standings with some level of accuracy, with only Utah being somewhat miscast. jsut to check it with a bigger sample, here is the same for teams with 4 players in the top 20 ranked using PER, with NBA standing among these teams in parenthesis:

1) DEN (5)
2) ATL (3)
3) SAS (4)
4) GSW (12)
5) POR (6)
6) MEM (7)
7) SAC (10)
8) ORL (1)
9) LAC (8)
10) DAL (1)
11) NYK (9)
12) WAS (11)

the agreement is less strong, with both Orlando, GSW and Dallas being way off, but otherwise you see a decent alignment between where you'd rank a team based on the average PER of its 4-man top20 combination, and where the team ranks among this group.

Point of this all being that PER does highlight quality in some respects, and you can use it to gain an appreciation for the quality of a team based on its best players. to the degree that what you see agrees with what the numbers say, i think this supports the use of PER as a metric of player quality, and further underlines why it's so odd that the rockets are a decent to good team despite the lack of PER-friendly players.

 

what i am looking for is the PER for every player in the NBA this season, in a .csv or excel importable format. i didn't see it there, maybe i missed it though, i always have trouble navigating BBR.

what you said supports what i've been saying also. if the rockets weighted average PER as a team puts them 22nd in the NBA offensively, and their adjusted FG% against puts them 20th in the NBA defensively, it is remarkable that the team is not only above .500, but doing so against the hardest strength of schedule in the NBA.