RealGM

I have been studying the NET PER by position at this link:

http://www.82games.com/0708/0708BOS.HTM

First, click on the players name.

Then, after you jump to a new web page, click on By Position in the middle top of the page.

You get detailed PER, by position on offense, and of the men a player is guarding at any one instant.

The difference is Net PER data (Your PER minus the PER of the man you are guarding at any point in a game).

A player page looks like:

http://www.82games.com/0708/07BOS10C.HTM

and you will quickly see why Kevin Garnett is much better at power forward than center.

Note: Guyclinch (Pete) posted why he thought adjusted plus/minus was perhaps more relevant, I'm not convinced. You can have a guy seemingly great in adjusted plus minus on a crappy team (i.e. Jamal Crawford) who everyone knows can't defend worth sh*t so his net PER is not comparable to his adjusted plus minus. Crawford's NET PER is worse than Eddie House, Rondo, or Tony Allen; but his adjusted plus minus is top 20 in the NBA. Which is why I'm not sold on adjusted plus minus.

Moving on.

Looking at the Celtics, it becomes pretty clear, when factoring in how effective a player is on offense AND how effective he is defending a position, that:

Ray Allen should not be played at small forward
Paul Pierce is MUCH better as a small forward than anything else
Kevin Garnett is MUCH BETTER as a Power Forward than Defending Centers
Leon Powe is MUCH BETTER as a Power Forward than defending Centers
Tony Allen is clearly best used as a shooting guard only
Sam Cassell has a marginally better NET PER than Rajon Rondo and Eddie House, meaning he may take minutes from both of them.

Pollard, Scalabrine and Perkins have the worst NET PER's of anyone on the Celtics, followed by James Posey and Glen Davis.

The Celtics still frequently use a closing lineup of KG at Center and Posey at Power Forward. On days when James is not shooting the rock well from 3, PJ Brown may allow an alternative where KG plays Power Forward and PJ assumes the post defender role as Center. Not all the time, but when appropriate.

The Celtics biggest void is they get killed at center.

Looking at the minutes distribution of the team since KG's return, and right before his injury, it's something like:

PG: Rajon Rondo 31 / House 17

SG: Ray Allen 38 / Tony Allen 10

SF: Paul Pierce 38 / James Posey 10

PF: Kevin Garnett 24 / James Posey 12 / Leon Powe 12

C: Kendrick Perkins 24 / KG 11 / Glen Davis 13

Analyzing the results for these players, and who you would want to replace, would leave me to believe that PJ Brown may be more than insurance. He may actually get time at the C spot, taking time from Kendrick Perkins and Glen Davis; or shifting Garnett back to Power Forward a bit more and allowing KG to play less Center. But, when you look at his NET PER from his last season with the Chicago Bulls, it is doubtful that he will take time from Perkins, Davis, or Posey; although he could quite possibly be a better defender than Perk, and definately better than Leon, at Center.

Ironically, even though PJ is longer, with a greater standing reach, can play post defense on centers, and hit jumpers out to 18 feet, whether he still has game there will remain to be seen.

Sam, on the other hand, looks to cut into House's minutes, and maybe some of Rondo's. Depends on tempo, and how well Rondo and/or House are playing that day.

NET PER (Offensive PER minus Man You are Guarding PER)

+16.0 Kevin Garnett
+ 9.7 Paul Pierce
+ 5.7 Ray Allen
+ 4.1 Gabe Pruitt
+ 2.7 Tony Allen
+ 1.9 Sam Cassell 2007-2008 LA Clippers
+ 1.5 Eddie House
+ 1.2 Rajon Rondo
- 0.3 Leon Powe
- 0.7 Glen Davis
- 3.9 James Posey
- 4.4 Kendrick Perkins
- 5.8 PJ Brown 2007 Chicago Bulls
- 8.1 Brian Scalabrine
-12.8 Scot Pollard

February 28th, 2008
Cs wait for Cassell to clear waivers
Posted by Mark Murphy at 8:40 pm
Though the Clippers, as expected, waived Sam Cassell today, the Celtics will wait until the veteran point guard clears waivers on Monday before attempting to sign him.

Though Cassell has reportedly reached a buyout agreement with the Clippers, the Celtics would be responsible for $6.1 million, in addition to another $2 million in luxury taxes, according to a league source.

Once Cassell emerges from waivers on Monday, Dallas, Denver and Phoenix are all expected to make pitches for the guard as well. Expect Kevin Garnett and Ray Allen - both former teammates of Cassell - to make recruiting pitches once the waiver period ends.


So we owe Cassell 6 mill, how do the Celtics have the money to sign him?

Because he will clear waivers and we will sign him for the vet min.

Matt34520 wrote:February 28th, 2008
Cs wait for Cassell to clear waivers
Posted by Mark Murphy at 8:40 pm
Though the Clippers, as expected, waived Sam Cassell today, the Celtics will wait until the veteran point guard clears waivers on Monday before attempting to sign him.

Though Cassell has reportedly reached a buyout agreement with the Clippers, the Celtics would be responsible for $6.1 million, in addition to another $2 million in luxury taxes, according to a league source.

Once Cassell emerges from waivers on Monday, Dallas, Denver and Phoenix are all expected to make pitches for the guard as well. Expect Kevin Garnett and Ray Allen - both former teammates of Cassell - to make recruiting pitches once the waiver period ends.


So we owe Cassell 6 mill, how do the Celtics have the money to sign him?

If Cassell is bought out, he will sign for the vet minimum with the Celtics like Scott Pollard ($1.27 million where the Celts pay him 770K and the league pays the rest). The Celts will only owe lux tax on the 770k they will owe him. Same for PJ Brown.

The league source is mistaken.

If the Celtics actually picked Sam up off the waiver wire, then they would inherit his current contract, and pay him $3 million remaining salary for this season, and $6.15 million (entire season's salary) in luxury tax. Obviously the Celts will not pick Cassell up off of waivers, just like other teams won't.

Plus, the Celts can't pick Cassell up off waivers because they don't have cap room, or a trade exception, that would allow them to do so.

^Oh my bad, I see. Thanks.

The impact is really about the fourth quarter. We don't have to feel tired or pressed.

Jammer,

Do I understand you correctly to be suggesting that nobody on the Cs should defend the opposing center, because it's bad for their net PER?

I think they actually tried that strategy in the Golden State game ...

Jammer, you are on a roll lately! Nice analysis.

I'm very interested in seeing how the additions will effect rotation and minutes. At this point of the season I'm not so concerned about minutes for the young guys--- we're in win-it-all mode now. Besides, I think Powe may still get minutes behind KG and we'll see Pollard and Scal waiving towels more. Baby is a rookie, he'll get his day in the sun.

But your point about this depth allowing our best players to spend more time at their natural position is important. I hope we only see KG at C for very specific matchups. He shouldn't have to play C now that we have Brown to step in.

Cassell is going to give the team much more flexibility in backcourt options, and will allow Pierce and Allen to focus on their positions more. I doubt Eddie House will bristle at losing minutes. Remember, he's a hired gunner who came here for a chance to win a championship, too, and has been playing out of position. He might end up as a guy who hardly plays during a playoff series and then comes in during the fourth quarter and hits the big shots when the team is struggling.

The only major anomaly in your per stats is that according to the stats, Pruitt should should be starting ahead of both Rondo and Cassell!

The only major anomaly in your per stats is that according to the stats, Pruitt should should be starting ahead of both Rondo and Cassell!

Heh. he is an outlier because he only plays in garbage time. It's actually hard to evaluate players from very bad teams for that reason. Ask Danny who plucked Dan Dickau for us..

I don't really disagree with that Jammer analysis much..but I have a few comments..

FWIW adjusted plus/minus is a stat that "fixes" the flaws in regular +/- and it's only posted once a year at 82games. I think Jammer is talking about regular +/- and it is something to look at.

PER is a handy way to evaluate players but it can be pretty unfair. The major reason why is that teams play TEAM defense not one one defense.

On a team with a shot blocking center teams can scheme to send guys TO that center. Teams switch on pick and rolls. They rotate and send help defenders - and they double and triple team.

For example if say Rondo's guy is often left as part of the scheme with Rondo helping out he is going to be artificially lowered because his guy will in fact often be free - by design.

By looking at plus minus you could discover a great help defender or a great ball handler and see how he impacts a particular system.. The perfect example is Tony Battie back in the day. With his help defense he transformed the C's defensively - and when he wasn't in there we were easy to score on..

As another example think of poor Bruce Bowen. he almost always has to guard a teams more productive scorer. This means by PER differential he is pretty awful. But in reality that's not entirely the case..(The guy is overrated dont' get me wrong).

Going back to the C's your "PER DIfferential" evaluation plan rates Eddie higher then Rondo. Now I really like House but Rondo's +/- is quite a bit better. As he runs the offense better AND plays much better help D. I think we can almost all agree we would rather have Rondo running the point then House. though the drop off isn't always that bad..

These kind of quirks wouldn't show up in PER differentials. PER is an imperfect stat introducing subjective likes and dislikes (as it's made up) whereas +/- for all it's flaws is 'raw data." That's why 82 games uses both in it's "Roland Rating".. It's not my genius idea.. It's just common sense.

Even that is imperfect of course - as it punishes guys on bad teams.. but its more accurate then PER differential alone.

Pete

Matt34520, just so youre clear, the info you posted wasnt saying theyd owe him that for signing him, theyre saying that they would owe him 6.1 mil if they CLAIMED him, which is why they are waiting until he clears waivers so he can sign for peanuts and still get a chunk of his money from the clips per the buyout agreement

Fencerregistered: Yes, whoever the Celtics play at center gets killed, with Garnett obviously the best, although he is twice as effective as a power forward.

Dogen: Pruitt's #'s are not significant because of:

1. Fewer Minutes, likely to change if he played say, another 300 minutes
2. Plays in garbage time, not against starters like Rondo typically does
3. He's a rookie, his performance has not yet stabilized
4. He's still learning how to play point guard

Guyclinch (Pete):

My comments about Crawford were specifically regarding his adjusted +/- versus his NET PER.

His NET PER is barely positive, his adjusted +/- is top 20 in the League.

I have seen a lot of him in his career. His individual defense is not good, and it is reflected in his NET PER.

Basically, the Knicks have 4 somewhat decent players (Crawford, Robinson, Lee and Randolph); and get killed when Crawford isn't on the floor. That doesn't mean that Crawford is a top 20 player in the league, it just means that the Knicks are either really bad (true), not well coached (also true), with bad chemistry (also true), or don't play D (also true).

Overall +/- is skewed to players who have played in every game, like Paul Pierce.

+/- per minute, or per 30/36/40/48 is easier to compare players, but I don't find it as useful as NET PER.

Same with adjusted +/-, as the Crawford example exemplifies.

Even though team defense is a crucial part of professional basketball, there's still a lot of man to man. Shooting Guards are a particularly high scoring position where not having someone who can defend the position will kill a team. Ray Allen and Tony Allen are both significantly positive in their overall NET PERs.

Rondo and House's offensive PER's look great, until you realize that they give up almost as much as they produce.

The real insight on NET PER is that:

1. Garnett is clearly near the top of the league (2nd, I think);

2. Pierce is a much greater net positive for the Celtics than most fans would realize, as is Ray Allen. Both Pierce and Allen average 38 mpg, so their significant Net positives are extremely criticial for team success.

3. Rondo, House and Tony Allen are net positives.

4. Center and backup power forward are the areas where the Celts need to look to improve.

Gotta go, be back tonight.

Jammer - it's not so much I don't agree with you. It's that Net PER is nearly as unreliable on it's own as +/-.

I think my House/Rondo example proves that. House has a better net PER but is he really the better PG for our team? I don't think so.

Crawford plays a similiar role with the Knicks - he is necessary for their team offense to run (okay) as I guess Nate can't do it. So with him on the court probably the team works alot better. That doesn't mean his individual D is that great (as you point out).

None of these statistics has the game "figured out" yet. In baseball OTOH you can really look at the numbers and find the better players.

Of course this is a good thing. If you read the baseball boards all they do is throw around numbers. Even guys you like say Pokey Reese you discover sucks. It's kind of depressing.

PER is a rough measure of production - guys with awesome PER differentials are always studs - but that's about as far as it goes. You could use any "production" statistic in there like efficency or TENDEX. Tendex might be better for all we know..

Pete

Jammer wrote:Dogen: Pruitt's #'s are not significant because of:

1. Fewer Minutes, likely to change if he played say, another 300 minutes
2. Plays in garbage time, not against starters like Rondo typically does
3. He's a rookie, his performance has not yet stabilized
4. He's still learning how to play point guard

I guess I should have made it clear that I was joking about that. The garbage time is probably the biggest factor. Many times Pruitt was put in when we were already running up the score.

Slightly OT: The Cassell signing may be the best thing for Gabe. Not so much this year since he likely will not see the floor or be on playoff roster, but helping him develop as eventual backup to Rondo.

...I have to chime into this discussion and support what Clinch is saying about favoring the NET PER over other stats...

While these statistics do show a rough level of effectiveness both team and league-wide, the figures themselves are not an accurate reflection of a player's impact on the game, just an estimate...

While a player with a ridiculously high or low figure certainly stands out, the actual spread in points amongst players may vary significantly due to a number of uncontrolled factors, namely:

-Who the player is on the court with
-What the player's actual defensive responsibilities were on a given play

In basketball, often a defender spends over 1/3 of his defensive plays on the court in a help defensive position that allows "his man" to score when he isn't even the one responsible for that coverage. For players who get more spot minutes this can rise up to 50% of his plays on occasion.

In the example of Rondo and House, these players are seperated by a marginal differential in NET PER, and I believe there is really no way of telling whether one is "better" than the other in this statistic because the numbers that it relies on to calculate it could be substantially altered if one were to go back and code for the actual defensive responsibility in a given play.

The fact that the other 4 players on the court have so much to do with:

-containing dribble penetration
-blitzing the pick/recovering
-making proper weak/strong side rotations

Makes these figures very tough to digest as is...even the simple length and size of the other four players factors into this, as small lineups applied specifically for increased pace and offensive spacing also have an effect on the defense at times.

Until someone actually comes up with a formula that can properly account for team defensive responsibility and determine the number of posessions a player is NOT responsible for his positional counterpart, I can't look at one guy's +5 and another's -1 and say with confidence that they are representative figures for that players ability.

None of the stats experts i've spoken with (quite a few) rely heavily on defensive statistics...they'll look at cumulative +/- and effectiveness ratings by rotational groups, but judging individuals defensively is still primarily left up to traditional coaching or painstakingly tedious game logging...

Offense is far more reliable a measure at this time...

Regardless, this thread has a great deal of merit as a topic for consideration and many great points are brought up....I just don't want anyone to over-rely on the stats inherent in these as quotable, absolutes because there is a lot that's not factored into them...

The particular stat you are discussing has proven to be pretty inconsistent on its own. When debating things in the past (probably with Pete) we would find situations with players who would have say a PER of 16 but only have a by position PER of 12 and 13 or an opponent PER 15 and by position it is 18 and 19. To start one or two point differences in PER really isn't that significant. It is trying to give you an idea of the type of player a guy is. There are plenty of definition breakdowns on the sites that use these stats.

Far more important than this is that all of these ratings have flaws, the people who created them tell you that. That have put a ton of work in to developing multiple instruments to provide information about player effectiveness. If you look at multiple ratings and they all tend to say the same thing then you can start to make some conslusions. But harping on one stat vs another doesn't make any sense.

There is also reality. I have been harping on the idea that Powe shouldn't play center and most of that is what my eyes tell me as much as stats and ratings. But to be fair the reality is when Leon plays center it generally means Posey (or worse Scal) is at the PF spot compounding the mismatch. So in comparison to Davis or Perkins who get a healthy amount of time next to KG when they play center, Powe rarely plays with Garnett and that has to factor into the evaluation.

good post. Having PJ will allow posey to defend small forwards more thereby giving pierce more of a blow. having cassell will similarly give ray allen more rest and take away some of the turnovers caused by having eddie and tony run the offense- although I have to say that eddie ran the O well against cleveland and that he and Powe have developed nice chemistry on the pick and roll. I envision a nice defensive unit of PJ ,Garnett, Posey, Rondo and Allen and I can also see PJ playing at the end of game along with garnett and posey. that would be a tough front line to score against. You could put pierce and allen or cassell in with that group and have a nice veteran group at the end of a game. Looking ahead to next year we are going to need a big man as well as a big guard. I expect we will lose pollard, sam and pj after this year and possibly scal so there will be a lot of roster spots to fill next year.

I think the biggest thing PJ has brought is that while the Posey/KG frontcourt works the Posey and either Powe or Baby is shaky. While I really have come to like the Posey/Powe/Baby line up I can also see that being a disaster in the postseason with Powe/Baby getting killed by officials and just the whole idea of essentially two first year guys in the playoffs. PJ gives Doc an option there.

I have kind of soured on House. I really like what he can do but we are seeing first hand what every team that he has played this much of role on for a season has seen. He is a good scrub time player but he is actually getting killed out on the court he is just trading 3's for 2's so if they are dropping at 40% it is awesome, because he also baits the opposition. But if he struggles even just a little bit his solution is to shoot more and the opponent can just go right at him defensively.

In the postseason the need for bench mins dries up, plenty of rest between games and in games, so starters will increase their mins. You need your first 3 guys on the bench to be good enough to match up with starters not with backups.

I believe Riley, who Doc played for, says play 8 and trust 6. With Sam, Doc now has KG, PP, Ray, Tony, PJ, Sam, to trust and Rondo and Perk to play.

I believe Riley, who Doc played for, says play 8 and trust 6. With Sam, Doc now has KG, PP, Ray, Tony, PJ, Sam, to trust and Rondo and Perk to play.

I think Doc can trust 6 and play 9. I think you forgot Posey in the trust side, and I think TA belongs on the 'play' side with Perk and Rondo. TA has been playing pretty damn solid, but he's not quite a guy I trust yet, he still has his moments where he plays like he just smoked crack.

Trust
Pierce
Ray
KG
Sam
Pose
PJ (eventually)

Play
Rondo
Perk
TA

I think Eddie House becomes more of a situational player, I'll be interested what his role becomes, I hear some people saying he's gonna slide to the 2, but I see only short minutes for him at that position.

The real question for me is who plays at the end of games come playoff time, Cassell or Rondo? I think Sam will have no problem coming off the bench, but I think when it gets down to it, crunch time of playoff games, he's gonna want to play down the stretch. Yet I still think Rondo brings some things to the table that Cassell clearly doesn't, better man to man D, and the ability to get in the lane at will.

By the playoffs, I hope Doc will have settled on his closer, whether its Cassell or Rondo. I don't want those two getting yanked around, the guy in there at the end of the game has to know he's gonna be in there no matter what. The ball is really in Rondo's court, to prove that he should be the closer for this team.