NET PERs after 20 Games + some obervations

[Jammer]

After 20 Games, looking at NET PERs, which factor in what you're doing (shooting %, turnovers,
fouls, rebounds) and what your man is doing (scoring, rebounds, etc.); it's clear that Rondo and
Perkins have improved significantly from last season, Pierce is playing a little better,
Kevin Garnett is getting back to where he was, and Ray is playing like typical Ray,
and not the sickly efficient Ray of last season.

However, in straight on court, off court plus/minus,
Ray, who plays the same amount of minutes as Paul Pierce (35.5 minutes per game),
Ray averages + 16.1 per game, compared to Pierce's +1.0 per game.

What this says, basically, is that even though Pierce has been the Celtics
second most productive player INDIVIDUALLY, factoring in defense,
the 5 man unit plays best when Ray is on the court. This is completely
illogical to me, because Pierce has played really well, outstanding at times,
but the HUGE difference in their on court off court plus/minus basically says
that the team has annilated the competition when Ray has been on the floor,
although Ray hasn't been putting up as good individual numbers as
Garnett, Pierce or Rondo. Ray's on court off court plus minus led the team
last year as well, so for all the comments about Ray being expendable, the
on court off court would indicate the TEAM does best when Ray is on the floor.
I'm still trying to come to grips with the true value of on court, off court plus minus
(which is very different than straight plus minus).

The other thing that jumps out at me is that
Rasheed Wallace is using up 8.8 of the Celtics 76 Field Goal Attempts per game.
I wish that he would use more of them around the basket where his length should
lead to a higher shooting percentage (Sheed is shooting 37% overall, 30.2% from three).
Sheed, however, is still an improvement over what Glen Davis did last year
during the Regular Season, or Scal.

Eddie House has slipped a bit from his steady play of the last 2 years,
and House's shots per game have dropped from 7 (past 2 years) to 6.

The bottom 3 guys (Giddens, Walker and Scal) only average about
1 shot per game in TOTAL since they don't play in every game.

I also list minutes per game (MPG) and Field Goal Attempts Per Game (FGA/Game)

NET PER ***** Player ****************** MPG ************** FGA/Game
+09.1 ********* Kevin Garnett ******* 30.4 **************** 11.6
+07.7 ********* Paul Pierce ********** 35.5 **************** 12.2
+07.3 ********* Rajon Rondo ******** 33.3 ****************** 9.4
+04.9 ********* Ray Allen ************ 35.5 **************** 12.3
+03.8 ********* Kendrick Perkins **** 27.2 ***************** 7.2

+02.6 ********* Sheldon Williams *** 13.9 ****************** 2.7
+00.2 ********* Marquis Daniels **** 20.4 ****************** 5.0
-00.5 ********** Eddie House ******** 16.4 ****************** 6.0

-03.0 ********** Rasheed Wallace ** 20.7 ****************** 8.8

-09.1 ********* JR Giddens *********** 3.8 ******************* 0.4 (last 3 guys don't play every game)
-10.3 ********* Lester Hudson ******* 3.9 ******************* 1.1
-13.6 ********* Brian Scalabrine ***** 7.1 ******************* 0.9


Obviously a backup SF that can shoot (Ryan Gomes) and a
backup PG that can dribble and shoot are the two things still on the shopping list.

Gonna be tough to get either one when you're offering some combination of
Tony Allen, Scal, Walker, Hudson and Giddens; even if you throw in a first round pick.

And I don't expect Perkins to be able to sustain that type of performance against
DEEP playoff team starting competition.

The biggest area for improvement is obviously getting Rasheed Wallace to
use his best judgement and just be efficient, even if it means
posting him up on the block more (and he HAS TO make the shot).

[Hemingway]

These stats are pretty meaningless.

[Jammer]

These stats are pretty meaningless.

Au contraire, my friend.

The list of the top NET PER is the list of the All Stars, with
2 or 3 exceptions out of the top 24.

The NET PER list is a list of the ackowledged top performers,
although it's a brutal stat.

The relativeness is that it gives a relative measure between players.
The brutalness is you get deductions
for missed shots, turnovers, fouls, or if your man scores or if your man grabs a rebound.

However, you get pluses when your man commits a foul, turnover, or misses a shot.

The cumulative 2010 list is not yet available,
but the 2009 season list, top 30, was (great correlation to the league's top performers):

+25.1 LeBron James
+21.1 Dwayne Wade
+18.3 Chris Paul
+14.4 Dwight Howard
+13.3 Brandon Roy
+13.0 Kobe Bryant
+10.8 Tim Duncan
+10.7 Dirk Nowitski
+10.4 Kevin Garnett
+09.2 Danny Granger
+09.1 Tony Parker
+09.1 Yao Ming
+08.6 Kevin Durant
+08.6 Andrei Kirilenko
+08.1 Carmello Anthony
+08.0 Pau Gasol
+07.7 Andre Iguadala
+07.6 Leandro Barbosa
+07.5 Amare Stoudamire
+07.1 Deron Williams
+06.6 Kevin Martin
+06.6 Ray Allen
+06.5 Joe Johnson
+06.5 Andrew Bynum
+06.3 Chris Bosh
+06.3 Al Jefferson (way down this year)
+06.2 Paul Pierce
+06.2 Shaquille O'Neal
+06.0 Gerald Wallace
+05.7 Rajon Rondo

[Slartibartfast]

I wonder how accurate the PER against stat can be, but thanks for doing all the homework Jammer.

[Banks2Pierce]

Exactly...like if Rondo is covering Devin Harris as an example...Devin Harris gets a high pick and roll and shoots behind it, does that count against Rondo? How can this be compiled in an accurate way?

[exculpatory]

Jammer aka e e cummings with stats,

If I had the time, I would give these numbers more thought, but alas I do not. I am snowed with deadlines.

BTW, how often do GMs use these kinds of stats?

Ciao

Ex

[GuyClinch]

Exactly...like if Rondo is covering Devin Harris as an example...Devin Harris gets a high pick and roll and shoots behind it, does that count against Rondo? How can this be compiled in an accurate way?

Yes it would count against Rondo. Its not that accurate - its worse then that you can quibble with the PER metric itself. How accurate is that?

As for GMs using this statistic - I doubt it. Reason being that they likely have access to compilied statistics for each invidual moment. Better then that - they can call up every single match via video. They can do things like - show me every single time someone took a shot against Rondo - and see what was the outcome..

That being said I think its broadly accurate - in that guys with very high NET PER tend to be the best players..

Pete

[GreenDreamer]

Net PER is a decent barometer of how a player is doing. It certainly isn't an end all be all stat. Bruce Bowen used to be a perfect example of a very good role player whose net PER sucked. These stats are being pulled off of the 82 games site, yet for some reason the OP doesn't mention the oncourt/offcourt numbers, or their Roland Ratings.

Personally I like to look at a variety of systems, and get a good composite view of how a player looks. One system might favor a certain kind of player and another might make the very same player look bad. Bargnani is a the kind of guy who "looks good" when you use PER, but is slammed by the truly comprehensive systems. Rondo, on the other hand, comes across as a stud in ALL of the statistical systems (PER, 82games, Basketball-reference.com's win shares, Berri's win score, etc.)

82 games has Ray as our highest rated player, with Rondo second, KG third, Perk fourth and Paul fifth.

Basketball-reference has Paul as our top player, with Rondo second, KG third, Ray fourth and Perk fifth.

PER would have it as KG, Paul, Rondo, Perk, Ray

Kind of a big difference there, don't you think? The systems aren't the same, and so you get different results. In Rondo's case, though, his ranking is consistent (in PER only a fraction separates his PER from Paul's), and so that would lead me to believe that his production is at a very high level. Changing the point of view doesn't degrade his value. Ray, on the other hand, could be looked at as the best or worst of the starting five, depending on who you believe, and that means that none, or only one, of the systems completely grasp his impact (though in this case, I'm leaning more in the direction of 82games).

Trusting any one system is a bad move, IMO. If a guy looks good in all of them, though, he is proabably as good as they say he is.