Last week I reviewed all five teams in the Atlantic Division. This week we turn our attention to the teams of the Central Division.

**Wins and Efficiency Differential**

Following the format of last week’s review, I will start by noting each team’s won-loss record and efficiency differential (offensive efficiency – defensive efficiency).

Detroit: 15-7, 8.6

Indiana: 11-11, -0.9

Chicago: 7-13, -4.1

Milwaukee: 9-12, -4.3

Cleveland: 10-12, -4.5

Entering this season I expected Chicago, Detroit, and Cleveland to contend for the Eastern Division crown. As of now, only the Pistons are contenders. Chicago and Cleveland have clearly disappointed, although one of these teams is just as clearly about to improve. Before I get to that story, let’s move from efficiency differential to Wins Produced.

**Wins Produced and WP48 for the Central Division**

So which players are responsible for the efficiency differentials we currently see? To answer that question we turn to Table One, where the Wins Produced for each team is reported.

**Table One: A Team-by-Team Review of the Central Division**

Looking over this table we see there are many good stories to be told. In fact, looking over the posts from the past month, some of these stories have already been told.

The Bulls Forget How to ScoreThe Bulls Forget How to Score

Was I Wrong About the Milwaukee Bucks?

Judging the Value of Anderson Varejao

Beyond these columns, I wanted to offer a few more thoughts on each team.

**Detroit Pistons**

What a difference a few days can make. After 20 games Jarvis Hayes had a WP48 [Wins Produced per 48 minutes] of 0.102, which is above average. After two subpar games, his mark is now clearly below average. Obviously this could change again with a couple of good games. Still, given what Hayes has done in his career, we would expect him to be below average.

Whether or not Hayes is above average, though, is not that important for this team. The key to the Pistons appears to be Jason Maxiell. If he continues to produce – given what the team is already getting from Chauncey Billups, Tayshaun Prince, Rasheed Wallace, and Antonio McDyess – then the Pistons (the team I follow) will be hard to beat in this division.

**Indiana Pacers**

It was thought, before the season started, that the key to this team was Jermaine O’Neal. If he played like he did in the first half of 06-07, this team might contend for a playoff spot. If he played like he did in the second half of last season then this team was in trouble.

After 22 games O’Neal has a WP48 in the negative range. So clearly he hasn’t helped. Still, this team has a record of 11-11 and at this point looks like a playoff contender. I still expect a few teams to pass the Pacers as the season goes forward. But still, the play of Mike Dunleavy, Jeff Foster (when not hurt), and Jamaal Tinsley is encouraging.

**Cleveland Cavaliers**

The best player in this division is LeBron James. Unfortunately for Cleveland fans, King James doesn’t have much of a supporting cast. But with Anderson Varejao back, King James and the Cavs can expect to win more frequently. It would help even more, though, if Drew Gooden could start producing and Donyell Marshall could get healthy.

**Chicago Bulls**

Cleveland can improve by just moving people who were out of the line-up back on to the floor. Chicago only wishes its problems were this easy. The Bulls have a problem getting the ball to go into the hoop. In reading media accounts of this story, it doesn’t look like anyone in Chicago knows why Kirk Hinrich and company can’t get their shots to drop. And I don’t know either. But if Chicago can’t fix this problem, this team is not going to contend in 2007-08.

**Milwaukee Bucks**

If we look at projected Wins Produced for each player (in Table Two below) we see that the Central Division is led by LeBron James and Chauncey Billups. And sitting at number three is Michael Redd.

Unfortunately for Milwaukee, once you get past Redd, the Bucks only have two other players who can claim to be above average (Mo Williams and Andrew Bogut). Without more production from someone else, it’s hard seeing how this team is going to contend this season.

**Ranking All Players**

Before I close this post, here are all 71 players ranked in terms of Projected Wins Produced.

**Table Two: The Players of the Central Division**

One should note how simple a projection Table Two presents. This analysis simply takes the minutes played so far and projects to 82 games. So LeBron’s injury results in a forecast that’s too low. King James, if he stays healthy, will produce more than 17.6 wins.

Beyond this point I want to note that this division is dominated by non-big men. Of the top five players, four play at the one, two, or three spots.

And with that sentence, I think is time to bring this review to an end. Looking at the schedule, I think I will review the Southeast on Sunday.

– DJ

Our research on the NBA was summarized HERE.

The Technical Notes at wagesofwins.com provides substantially more information on the published research behind Wins Produced and Win Score

Wins Produced, Win Score, and PAWSmin are also discussed in the following posts:

Simple Models of Player Performance

What Wins Produced Says and What It Does Not Say

Introducing PAWSmin — and a Defense of Box Score Statistics

Finally, A Guide to Evaluating Models contains useful hints on how to interpret and evaluate statistical models.

*Basketball Stories*

Pete23

December 14, 2007

Nice review of the Central Division. Thanks.

William

December 14, 2007

I love these division-wide reviews. Please keep them up.

A few thoughts:

a) Yi Jianlian is overrated right now. He had a couple nice games early in the season, one of which was televised, and got a favorable PR rep. But he hasn’t backed that up.

b) Mike Dunleavy, on the other hand, is criminally underrated. His stats compare very favorably to Andre Iguodala, but he gets a fraction of the press.

c) If “superstars” sport a WP48 of 0.200 or higher, there are only 5 in the whole division. The Central Division is outgunned.

Mike

December 14, 2007

Why does your model overrate strictly rebounders but underrate strictly scorers – Ben Wallace and Jeff Foster are simply rebounders with little scoring and they get very high WP48 marks whereas Carmelo and AI get low WP48 marks despite being volume scorers? I guess you will say that they are inefficient scorers. Fair enough – then efficient volume scorers such as Eddie House and Kyle Korver, or whoever qualifies as an efficient volume scorer (who just score efficiently and nothing else like Ben Wallace who just rebounds and blocks shots and nothing else) should have high WP48 – right? If not, isn’t your model overvaluing defense and overvaluing offense?

Mike

December 14, 2007

I meant overvaluing defense and undervaluing offense..

don

December 14, 2007

Thanks for the information and analysis.

Drawing upon the Pareto Maxum, three of the teams are getting over 90% of their production from their top three (Mil, 94.8%; Ind, 91.3%, and Cleveland, an unbelievable 102.5%). Detroit,on the other hand, not only is getting the highest contribution from their top three (9.5 wins v. Ind’s 9.4), but not as dependent upon them (top three contributes 57.2% of wins)–I would think a real indicator of strength.

I would expect that as the season continues that the abnormally high contributions from top three will get to last season’s norm (somewhere around 70+%, if I recall correctly). The question, of course, is will that corrective come from reduced absolute contribution of the top three or increased absolute contribution of the others ( tradeoff or additive). And I suspect player health will decide that, as much as player applied talent.

andrew

December 14, 2007

Mike:

Pure boarders and scorers are valued in accordance with their contribution to victories, so in that regars, the answer is no.

Andrew

mrparker

December 14, 2007

KOBE BRYANT IS THE BEST PLAYER IN THE CENTRAL DIVISION!!!!

Oren

December 14, 2007

I have a question.

If your model is so biased towards players that get rebounds, then how come there’s only one Center in the top five of this model and he’s ranked #5?

Anyone else note that the #2, #3, and #4 players are all Guards? And I do believe that LeBron is only the third best rebounder on his team.

joe

December 14, 2007

If lebron is so good, why are his teammates so bad. If he wants to be considered one of the greats, he needs to make his mates better, like wade does always, duncan has done, and like horry always did. Championships, people. Real talk.

dustin

December 14, 2007

Joe, I think leading you team to the nba finals at the age of 22 is a good start….Lebron has a good ten years left to get a championship. I would love to see you talkto some kobe fans, that would be something.

dustin

December 14, 2007

Also, joe, does that mean Kevin Garnett wasn’t great when he was at minnesota, but now is great because his team will probably make a championship run?

mrparker

December 14, 2007

now I’ve heard everything….sitting on the bench for 30 minutes a game…then standing around the 3 point line until a ball bounces to you and shooting it makes your teammates better.

Each team should employ this strategy

Animal

December 14, 2007

Not that I think a 3-point specialists are great… but to just simplify something like that and make it sound stupid is a really bad argument. I could say the same thing about a rebounder (danny fortson would be an example) Sit on the bench 30 minutes a game, don’t do a thing on offense, stand under the basket doing nothing until somebody shoots and then jump up and try to get the ball… all the while usually just fouling everyone. That is just not a good argument to make.

mrparker

December 14, 2007

defensive rebounding gains posession

shooting costs you posession

getting a rebound will 100% of the time gain posession for your team

shooting will about 60% of the time(45% league average + 25% of all shots are offensive rebounded) lose you posession.

Therefore it is more productive over the course of a season to grab as many rebounds as you can than it is to shoot the basketball. Unless Horry shoots the ball either into the basket or to one of his teammates 100% of the time.

mrparker

December 14, 2007

25% of all misses…I should clarify that

25% of 55% is 13.5% which adds up to about 58%

I should have written that shooting will cost you posession about 40% of the time.

dbg

December 14, 2007

This site

http://www.winsproduced.com/playerstats.php?player=4284

has Yi Jianlian’s wp48 at 0.368 and you have it at -0.016. That’s a pretty big difference. Why is that?

Mike

December 14, 2007

MrParker – that makes sense, but the point of the game remember is NOT to get as many rebounds as possible but to score as many points as possible. By your logic, teams should employ only rebounders that way they maximize possessions – but someone’s gotta put the ball in the hoop too, and percentages for that are never a 100% – Berri’s model should take that into account. My point was that Berri’s model does not accurately reflect the importance of scoring efficiently – I would think that someone who scores 15ppg in 35 mins while shooting 50% overall, 40% from three point range and 90% from the FT line and does nothing else is more valuable (or at least equally valuable) as someone who gathers say 12 rebounds in 35 mins and does nothing else…

And David, I am glad to see that you didn’t preface your Central Division reviews with the “its too early” sentiment…

spike

December 14, 2007

Mike, didn’t you hear? At the end of NBA games, they award the victory to the team that had the most possessions.

dberri

December 14, 2007

dbj,

When you look at JChan’s site you see that Win Score and PAWS have the same number. I will e-mail JChan and have him look at that. Those numbers have to be different.

Jason

December 14, 2007

Mike, the point of the game is not to score as many points as possible. The point of the game is to score more points than your opponents. A 19-18 win and a 186-184 are scored the same way in the season standings.

Mr. Parker is correct about a rebound being very valuable, but it isn’t because it gets a possession. In a game, possessions are basically equal. At the end of a possession, the other team gets the ball. They’re more or less equal by definition of what a possession is. The defensive rebound isn’t necessary to start a possession. That possession would come regardless of the rebound at some point. The other team scoring a a basket will result in the same possession that grabbing a defensive rebound will. The difference is what happened *before* the rebound. In the case of a defensive rebound, the opposition did not score, where a scoring opportunity for the opposition did not result in their scoring. Points were prevented. Preventing your opponent from getting 2 points is as valuable as scoring two points after giving up a basket in terms of the relative score.

Whether you regard the defensive rebound as the important part of the defensive stop or not, it clearly marks a defensive stop.

dustin

December 14, 2007

Dave, for some newer players he hasn’t added the position adjustment yet. So their scores are highly inflated. However, afaik everone else has a position adjustment.

Mike, efficient shooters are valuable. Reggie Miller posted a wp48 around .2 . Also, when you have a fairly specific sceario like that, you could better prove your point with calculations. Give the rebounder a position and calculate his wp48. Give the shooter a position and shot distribution (fga, 3pfga, fta) then actually CALCULATE it to SHOW that is true. Right now you say it, but no one knows whether that is true or not.

JChan

December 14, 2007

Hey, I hadn’t noticed that the rookies WP48 are way off on http://www.winsproduced.com because they don’t have a position adjustment in the database. I’m doing my best to find time to work on this, but unfortunately it’s taking longer than I thought it would. Keep hope, though. Some really cool stuff is coming. Eventually.

Patrick

December 14, 2007

Mike,

Efficient scorers are not undervalued. Why do you think Chauncey Billups is #2 on this list?

Michael Jordan also always had a high WP48.

Rebounds are important because possessions are important. This comment:

“Mike, didn’t you hear? At the end of NBA games, they award the victory to the team that had the most possessions.”

…is pretty funny, because if both teams score with equal efficiency, then, indeed, the team that had the most possessions will, in fact, always win.

If the point of the game is to score more points than your opponent than getting more possessions is a fantastic way to start. If you shoot 50% and I shoot 40%, and you get 10 possessions and I get 13, I win. And a 10% difference in shooting efficiency is a much bigger gap than is usually the case.

Note that steals and (negatively) turnovers are also worth as much as scoring efficiency.

In any case, this isn’t rocket science. Even coaches who have no use for stats at all will emphasize rebounding, not turning the ball over, and defending well over scoring. The saying “defense wins championships” was not coined by economists.

Mark

December 14, 2007

Mr. Parker is correct about a rebound being very valuable, but it isn’t because it gets a possession. In a game, possessions are basically equal. At the end of a possession, the other team gets the ball.Jason, while this is technically correct you are confusing the act of defining a term with analyzing basketball. It would have been perfectly reasonable to define a possession differently so an offensive rebound gave you another possession, it just makes the math easier to define it this way.

now I’ve heard everything….sitting on the bench for 30 minutes a game…then standing around the 3 point line until a ball bounces to you and shooting it makes your teammates better.I believe it does, at least the shooting part. No one keeps track of how many shots a player makes off of someone else’s assist, but clearly both halves make up a team play.

Personally, I think missing a shot has no impact on the basketball game at all, but missing the offensive rebound is where the problem kick in. We should penalize each player who doesn’t get a rebound based on the opportunity loss. Let’s put that in the regression analysis and see how the factors adjust. :-)

This part of the comment should go on the FAQ page, but I’d like to see a complete breakdown of which part of the WP formula comes from regression analysis and what parts come from debatable theory.

dberri

December 14, 2007

Mark,

All of Wins Produced comes from a regression. You can debate the formulation and results of that regression if you like. But all of it comes from a regression.

Jason

December 14, 2007

Mark, I’m not sure where you get the notion that I’ve confused definitions with analysis. Perhaps you can clarify where you think I’ve provided analysis that was actually a definition.

I’m very aware of what analysis is and what definitions are. Definitions are, however a very good place to start in analysis. They tend to make it easier to communicate when people use the same meaning of words. You’re free to try to define a possession differently, but be aware that a critique of an analysis that employed a different definition, your critique based on your definition will not have much meaning.

William

December 14, 2007

DB –

Any chance you could create a forum on your website entitled “Debate the Efficacy of WP48”, then steer these nagging commenters in that direction each day?

I’m not criticizing their opinions, or angling to have them silenced. I just believe that the Comments should generally be focused on your post…. and not a rehash of the “generating shots v. rebounding” argument, or the “rebounds aren’t created equal” argument, etc. etc.

It’s frustrating having to wade through these same arguments, and their corresponding explanations, day after day.

I like hearing from the TrueHoop crowd, and I like hearing from the UNC-G professor’s students, and I encourage peer review (such as it is), and I know it’s mostly out of your control, but when you post about the Central Division, I just wish the discussion that followed was about the Central Division too……

Owen

December 14, 2007

I hear that…

dberri

December 14, 2007

William,

This is kind of the idea behind the FAQ page. When we get that finished we can send people there.

That being said, this is an open forum. I am reluctant (expect in very specific circumstances) to censor anyone.

By the way, if you want to see some silly comments, check out this post at the Yglesias site.

http://matthewyglesias.theatlantic.com/archives/2007/12/score_one_for_adjusted.php

The post is on plus-minus. Yglesias does not mention me at all. But the comments basically attack some guy named Berri. And as PJ notes, they get the story wrong in the process.

Again, these comments are an open forum. You can’t stop silly stuff. But hopefully the FAQ page will help.

Mark

December 15, 2007

Jason,

It just seemed to me that mrparker had the correct analysis but used the wrong definition, but your response made it sound like his analysis was incorrect.

DB,

Based on what I read elsewhere, I was sure that the opportunity cost associated with a missed shot was an assumption in the model. It took me forever to find on this site what should be obvious, the actual formula for WP, but sure enough, all factors are determined via regression:

PROD = 3FGM*0.065 + 2FGM*0.032 + FTM*0.018 + FGMS*-0.034 + FTMS*-0.015 + REBO*0.034 + REBD*0.033 + TO*-0.034 + STL*0.034 + FTM(opp.)*-0.018 + BLK*0.018 + AST*0.022

(I know, PROD isn’t WP, but this is the part that matters unless I’m terribly confused.)

So really, the only assumption going in is that these statistics are all that is needed and a reasonable model is linear. Did you ever try throwing in other factors like AS/TO or FG% to see if the regression produced a more accurate result?

Anyway, you should put the actual formula in the FAQ.

One last question, how to you assign FTM(opp.) to an individual? It seems you would need to lump that in with the later team corrections. (I know, I should buy the book, assuming it answers these questions.)

Kent

December 15, 2007

So the weightings for WP come from a regression where:

(1) the dependent variable is probabiling of winning (i.e. winning % projected from point differentials).

(2) the independent variables are FGM, FTM, FGMS, FTMS, REBO, REBD, TO, STL, FTM(opp.), BLK, AST

And it’s just a plain vanilla multivariate regression?

Is my understanding corrrect?

Tkx.

Kent

December 15, 2007

(sorry for typo. should have read “probability” in point 1)

Kent

December 15, 2007

Mark asks “Did you ever try throwing in other factors like AS/TO or FG% to see if the regression produced a more accurate result?”

As a related and more general question, as part of my prior question I want to also ask whether any other statistics or interactions of the statistics were looked at in the regression.

Thanks!

Jason

December 15, 2007

Mark, the way Mr. Parker wrote it, he said that the value of a rebound was in getting the possession. I don’t think that this is correct. The value is in getting the ball without surrendering points to do so. It is certainly not correct using the same definition of a possession that Dave has used in his analysis and requires a definition where possessions are unequal if it is to have structured meaning to efficiency differences. Possessions are only valuable if you do something with them. Getting the ball in itself does not bring the value. A defensive rebound is valuable because it means that your opponent did not gain value from their possession.

mrparker

December 15, 2007

In my world gaining posession = taking the ball from the other team.

You do not take the ball from the other team when they put the ball in the basket. Instead the rules stipulate that you get it back.

So counting a possession after a made basket as a “given posession”….given possessions + gained possessions = total possessions.

Each team will have the same amount of total possessions but not the same amount of gained/given possession.

I hope tht clears up any holes that are punched in my argument by semantics.

All possession are not created equally.

dberri

December 16, 2007

Kent,

What you report as the model is not correct. In the technical notes at wagesofwins.com I report the actual model that is estimated. I should also note how this model was derived (its in the paper detailing the model) but I haven’t gotten around to this.

firebird

December 18, 2007

So counting a possession after a made basket as a “given posession”….given possessions + gained possessions = total possessions.

Each team will have the same amount of total possessions but not the same amount of gained/given possession.