“If they want you to cook the dinner, at least they ought to let you shop for some of the groceries.”
This quote – from Bill Parcells – essentially summarizes the problem facing all head coaches who work under a general manager. The coach is responsible for the outcome of the dinner, but the general manager buys the groceries.
The Sam Mitchell Story
The latest coach to experience this problem is Sam Mitchell. Mitchell became head coach of the Toronto Raptors in 2004. After two losing seasons, the Raptors won 47 games in 2006-07 and Mitchell was named Coach of the Year. The team’s record, though, was a bit of a mirage. Toronto’s efficiency differential (offensive efficiency minus defensive efficiency) was only 1.04, a mark consistent with only 44 victories.
Last year the Raptors improved. The team’s efficiency differential rose to 3.12, the best mark in franchise history. Although this differential is consistent with a 49 win team, Toronto only won 41 contests. Consequently, it appeared that Toronto took a step back.
This past summer the Raptors sought to take a few giant steps forward. Around the NBA draft, Toronto sent T.J. Ford, Rasho Nesterovic, Maceo Baston, and the draft rights to Roy Hibbert to the Indiana Pacers for six-time All Star Jermaine O’Neal. Since the players surrendered were not All-Stars, it was generally believed the Raptors were about to become a legitimate contender in the Eastern Conference.
But this hasn’t happened. After seventeen games the Raptors have a losing record. And the team’s efficiency differential stands at -4.10.
As a consequence, Mitchell has lost his job. Here is how this decision was explained in an article by Adrian Wojnarowski and Johnny Ludden:
After strengthening their frontline in the summer with a trade for center Jermaine O’Neal, the Raptors had hoped to challenge the Boston Celtics in the Atlantic Division. Star forward Chris Bosh has responded with the best season of his career and point guard Jose Calderon is an All-Star candidate, but the Raptors have struggled to an 8-9 record. …”This team is a lot better than a 8-9 record,” (Bryan) Colangelo said on a conference call early Wednesday evening. “…We feel we were not getting the maximization of this roster we had hoped for.”
So was it the cook or the groceries? Certainly Colangelo – the general manager buying the groceries – would like to believe the problem was Mitchell. Unfortunately, I think the data suggests there isn’t much wrong with Toronto’s cook.
Why Toronto Has Declined
Table One reports where the Raptors stand after 17 games this season. Two perspectives are offered. The first looks at the outcome the Raptors should have expected given the performance of these players last season. The second looks at what the Raptors have done this season. Both perspectives tell essentially the same story. Toronto is a below average team, and this outcome should have been expected.
Table One: The Toronto Raptors after 17 games in 2008-09
Last season the Raptors had seven players who played more than 1,000 minutes and posted WP48 [Wins Produced per 48 minutes] marks that were above average (average is 0.100). These seven were Jose Calderon, Chris Bosh, Jamario Moon, Anthony Parker, Carlos Delfino, T.J. Ford, and Rasho Nesterovic. Ford and Nesterovic were sent to Indiana while Delfino migrated to Europe. Of the four remaining players, Parker has been below average this year and Moon – although still above average – is offering less. But even if Parker and Moon maintained what they did last year, this team would still be below par.
The problem is really quite simple. Jermaine O’Neal – who used to be a “good” player – has not played well. And his performance in 2008-09 is consistent with what we have seen since the middle of the 2006-07 season. As you may recall, O’Neal suffered an injury at that point, and his production suggests he has never fully recovered.
Essentially the Raptors have lost three productive players and then added a player who only has memories of being a good NBA player. Consequently, we should not be surprised that this team has struggled. What’s surprising is that the person who assembled this team expected a different outcome. Remember, we have evidence that Colangelo was asked by his owner to read The Wages of Wins. And yet, despite this assignment, Colangelo expects the team he constructed to be better.
All of this leads me to believe that Colangelo’s owner forgot a crucial part in the process of making an assignment. It really helps if you give people a test after you assign a reading. At least I have found, students who know they are not going to be tested over an assignment tend to find other ways to spend their time.
Let me close with a few more notes on this story (comments that are a bit more serious).
- Andrea Bargnani – the first choice in the 2006 NBA draft – has improved. Last year he was the least productive player in the NBA, finishing with -5.7 Wins Produced. This year he is on pace to produce 1.4 wins. He would actually be an above average small forward, but according to 82games.com, Bargnani is playing most of his minutes at power forward. And at the four spot, he’s still below average (but not as bad as he was last year).
- Bargnani’s improved shooting efficiency is one reason he is offering more. But while Bargnani is now frequently finding the basket, O’Neal is still struggling with his shot. Last year O’Neal’s adjusted field goal percentage was 44%. This year O’Neal’s mark stands at 42%. Both marks are well below average.
- Last year O’Neal was paid $19.7 million and produced -0.4 wins. As a result, he was the Most Overpaid Player in the league. According to HoopsHype, O’Neal is receiving $21.4 million this year. He is only on pace, though, to produce 0.4 wins. Consequently, it’s possible that O’Neal could be a repeat winner of the MOP award.
- O’Neal’s partner in Toronto’s frontcourt is Chris Bosh. Bosh has declared that he wishes to be MVP in 2009. Currently he is on pace to produce 16.3 wins, which would be a career high. Although such a mark is very good, Bosh’s production levels are nowhere near what we see from Chris Paul, Dwight Howard, or LeBron James.
- That being said, Bosh must be moving up the list of potential free agents in 2010. Bosh is not on a contender this year. And given Colangelo’s assessment of this team, he’s not likely to be on a contender in 2009-10. Hence, Bosh might want to go elsewhere. And if Bosh does depart Toronto, Colangelo might have to follow. At least, losing Bosh would make the task of building a contender in Toronto extremely difficult (even for someone who read The Wages of Wins).
– DJ
The WoW Journal Comments Policy
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.
Wages of Wins Journal 
Tommy_Grand
December 5, 2008
TJ Ford is having a nice season for Indiana. Not quite as good as Calderon, but maybe top 20 at his position.
newfcollins
December 5, 2008
Nice post. One of the problems with using stats all the time though is one can’t see the intangibles. For example: Why is Bosh having a much better season? Because O’Neal is playing defence against the other guys top players, leaving Bosh with more energy to play offence. But there’s no stat for that one.
For example, when the Raptors played the Pistons, O’Neal defended against Rasheed Wallace, who likes to crash, bang and play a physical game. That means Bosh didn’t wear down as the game went on. They didn’t win, but this strategy will work out over the course of the full season.
kevin
December 5, 2008
Colangelo- the latest dumbbell in a long line of dumbbells. I’m still grateful to his dad for handing the Celtics 2 titles on silver platter by going and rashly signing Charlie Scott even though it was the Celtics, and not the Suns, who held his draft rights.
Thanks for that, Jerry. And thanks for mentoring a son who is just as idiotic as you are.
kevin
December 5, 2008
“TJ Ford is having a nice season for Indiana. Not quite as good as Calderon, but maybe top 20 at his position.”
FWIW, Rajon Rondo completely kicked Ford’s ass Wed. night. His defense was so inept, Rondo did whatever he wanted to do with impunity.
SWaN
December 5, 2008
Thanks for the analysis.
It looks like Bargnani will play at SF which should help his value. Calderon has been injured, so hopefully his numbers pick up.
The back-up PGs cannot play any worse. Hollinger seems to believe rookies (Ukic)improve slightly over the season. Supposedly Parker starts slowly, but maybe age has caught up with him.
Sadly, the acquisition of O’Neal cost Raptors many above average players.
Would be interesting to see an analysis of Philly as they were another preseason ‘contender’.
Italian Stallion
December 5, 2008
I think it’s 100% clear and obvious that Colanglo expected O’Neil to be in better physical condition and more productive this year than his recent statistical record suggested. You can argue that that was a poor assumption, but to argue that he thought O’Neils recent form was still All Star caliber or going to improve the team if it remained flat is to argue that he is basically retarded. That’s not the case – which makes your contention just plain silly. Simply say, he was mistaken if he thought O’neil was going to recover his best form because that’s what apparently happened .
Frank M
December 5, 2008
In effect, you are saying that a coach makes no difference.
A team’s record can be derived from its overall and, by definition, its individual players’ statistics. This is true. But does a coach’s work not show up in the statistics?
Is it your position that a coach has no impact on, say, how young players develop? Their lack of development (for example) would be reflected in their statistics but you would attribute this 100% to the player.
Tyler
December 5, 2008
1) 82games.com is notoriously unreliable for positional data; Bargs has been playing a lot more at the 3 than the 4, especially of late.
2) These stats don’t do any justice to Mitchell’s rotations, lack of mid-game defensive adjustments and other strange coaching habits.
3) JO’s defense has actually been a lot more valuable than his offense has been negative… and the decision to keep feeding him the ball on offense is a coaching decision.
4) Mitchell did a poor job of using Kapono
5) Mitchell alienated and didn’t use his assistants, which is terrible, because he’s one of the worst Xs and Os coaches in the league. An excellent motivator, he clearly lost the team and, lacking the ability to extract effort, was no longer able to do anything because he’s never been a good strategist.
6) His long resistance to using zone because he felt that every player needed to be accountable for their man and his general reluctance and lack of knowledge on defensive strategy has been an issue during his entire tenure as Toronto’s coach, though masked when Garbajosa was healthy in ’07.
Those are the major issues with deciding that Mitchell wasn’t the problem.
Having said that, there isn’t one “problem” with the Raptors; our bench is awful and Calderon is about the worst perimeter defender I have ever seen. The Raptors lack a really good second scorer and don’t have a significant iso scorer aside from Bosh. Those are roster issues and are BC’s deal, for sure.
But I think that trying to say Mitchell wasn’t a significant problem is a good example of why stats in isolation aren’t always an effective tool.
The Raptors went 12-18 to close out the season last year, and 9-17 over the last 26. Mitchell wasn’t just struggling with the team this year, he struggled with it last year. This is a guy who had confrontations with Rafer Alston, who stopped talking to Bargnani for a month, that kind of stuff. He’s a guy whose coaching inadequacies have been clear as crystal since his first season in 04-05… and have not improved at all since then. He hasn’t shown growth as a coach in nearly a half-decade. At some point, you have to hold a coach accountable for that sort of thing, even if he’s not the only reason that the team is failing. This isn’t just an issue of the ‘groceries,’ so to speak.
Sam Mitchell’s time was up; he had 4+ years in Toronto and he had his ups and downs… but he wasn’t the right coach, right down from his disputes over style of play to his various other issues. He’s a decent coach and he’ll find work elsewhere, but the Raps needed someone else. Now, when Triano starts finding certain problems, it will be clear that they are roster-related, and thus the onus will shift from Mitchell’s coaching to Colangelo’s plan.
John W. Davis
December 5, 2008
Too bad J O’neal isnt a Wins Producing player. How was he perceived as such a good player but not produce wins?
Must be the Points per game average.
Tball
December 5, 2008
newfcollins – Rasho could have just as easily matched up with Wallace to save Bosh’s energy. O’Neal has not been an upgrade over what Rasho could/did provide.
kevin – Rondo has been going by every PG he’s played the last couple of weeks, so I don’t think Ford’s performance tosses him from top 20 PG (a grouping that would include some mediocre starting PGs)
Frank M – use the search tool on “deck chairs” to find a discussion of coaching impact (above ‘recent posts’ in the upper right margin).
keving
December 5, 2008
“A team’s record can be derived from its overall and, by definition, its individual players’ statistics. ”
Actually, a team’s record is derived by how many wins and losses it accumulates. Individual player statistics can be awfully deceiving. For one thing, they don’t measure an awful lot of things that go into deciding who wins and who loses.
Tball, did you see the game? On one play, Rondo stood out at the top of the key with the ball raised above his head. Ford stood off him by about 6-8 feet. So, under no pressure at all, Rondo just waited for Garnett to get a running start to the basket and threw him a lob pass that he could easily dunk. How can you expect to win if the opposing point guard is not even shown a faint of defensive pressure.
Rondo is having a nice season but he beat up on Ford like he was playing against a high school player.
Joe
December 6, 2008
It is nice to see that the new coach is maximizing the minutes from Bargnani, Graham, and Kapono at the cost of minutes for Parker, Moon, and Humphries. That is a good recipe I think…
Vince Gagliano
December 6, 2008
For the Pacers, O’Neal’s loss has been Troy Murphy’s gain.
In 2006-07, his first season in Indiana, Murphy posted a 9.6 WS/48 with O’Neal as a running mate.
The next year, it was 13.4, albeit with more playing time than ever before. Note that O’Neal played only 42 games that year, compared to 69 the year before. Also, O’Neal played seven more minutes per contest in 2007 than in 2008.
Now, without O’Neal around, Murphy has a 15.6 WS/48 through 18 games. This is roughly equivalent to a .282 WP48 at power forward.
In fact, Murphy is probably playing the best ball of his career, including his Warrior years.
Vince Gagliano
December 6, 2008
And while it’s tough to gauge, it looks as if Bosh has benefited from having O’Neal around.
Last year, he posted a 14.3 WS/48 mark, which is up to 15.6 through seven games. Even O’Neal has benefited, if only by a little.
Sadly, what might have made the offseason trade so bad for Toronto wasn’t what they got, but rather what they gave away to get it.
Kevin
December 6, 2008
Still don’t think he should have got fired. I wonder how the Raptors will do the rest of the season.
Oh yea, please help my friends and I resolve an arguement at:
http://thetossup.wordpress.com/
I’d really appreciate it.
NOEngineer
December 7, 2008
I would LOVE to see a table showing the projected WP48 for the top 10 players at each position who are able to play but NOT currently employed. Box score data may be hard to come by, but maybe it could be done…
Jason E
December 7, 2008
“Actually, a team’s record is derived by how many wins and losses it accumulates. Individual player statistics can be awfully deceiving. For one thing, they don’t measure an awful lot of things that go into deciding who wins and who loses.”
A team’s record is the number of wins and losses. It isn’t derived from this. A team’s wins can be derived or approximated without knowledge of the actual wins and losses from the cumulative statistics. It isn’t perfect, but it’s very, very close.
*Everything* that determines wins and losses is recorded in the individual statistics. Wins and losses are determined by which team scores more than the other. That’s it. There are no points for intangibles. By simply adding up the points scored by all the players on one team and comparing it to the points scored on the other team, you can determine exactly who wins each game.
Of course, in terms of the contribution to the win, apportioning it according to raw point total misses a considerable deal and does not do well for predicting *future* wins since other tracked categories factor heavily into how the other team was prevented from scoring, how a player added to or took away from the output of teammates.
In fact, while there are limits on what is recorded that results in some error in the apportionment of contribution to a win (or loss), simply going by what is recorded in the box score does a rather good job of predicting how a team will do. There are undoubtedly some things not recorded that affect how we should apportion responsibility for the outcome, but *most* seems to be rather accurately reflected in the statistical categories covered. It is not “awfully deceiving.” If it were deceiving, the predictive ability would be low. The predictive ability from statistical output is not low. It is actually quite good. If the “awfully deceiving” part is the error (and what else could it be?) there is not much that is “awfully deceiving.”
keving
December 7, 2008
“*Everything* that determines wins and losses is recorded in the individual statistics.”
You’re kidding, right?
Are deflected passes that prevent easy baskets but don’t result in a loss of possession recorded?
Are opponents FG% (FG% of the point guards while Rondo covers them, for example) recorded?
Is “induced fatigue” recorded?
Is “switches that result in a missed shot or turnover” recorded?
Are “blocked shots that result in a loss of possession” recorded separately, rather than lumping them in with ones that are slapped right back into the hands of the opponent or batted out of bounds?
Are “entry passes prevented by denial defense” recorded?
When a 24 second clock turnover is recorded, is the induced turnover assigned to the player most responsible for inducing it?
Are fouls sustained by players separated into good and bad ones, so one can get an understanding if the player is moving his feet and anticipating on defense or not?
Are “fouls induced” recorded?
And when, God willing, someday they will be, are they separated into non-shooting and shooting categories?
Are individuals given statistical credit for getting the opposing team in foul trouble early in the quarter, thus limiting their strategic options?
Are “breaking up 2-on-1 fastbreaks” recorded?
Really, if I wanted to, I could think of another half dozen other things to add to the list.
Jason E
December 7, 2008
““*Everything* that determines wins and losses is recorded in the individual statistics.”
You’re kidding, right?”
No. The only thing that determines a win or a loss is the points both teams have. Nothing else determines who won the game. There are no ‘intangibles’ where one team scores more than the other, but loses.
“Are deflected passes that prevent easy baskets but don’t result in a loss of possession recorded?”
They are not, but deflected passes don’t determine who won or lost. The *result* of those deflected passes might, but it’s the result (a scored basket or a change of possession without scoring) that is what the outcome is based on.
Deflected passes may (or may not) be useful in assigning credit to a particular player (e.g. what percentage of the total points he is responsible for) but knowing the number of deflected passes a *team* has does not tell us more than knowing how many points both teams scored. Please do try to differentiate between assigning individual credit and looking at the team. The two are not the same thing.
“Really, if I wanted to, I could think of another half dozen other things to add to the list.”
You could, but since it didn’t seem like the list you provided indicated you knew the difference between apportioning credit (something some of those might be useful for) and what determines a victory, you should save the keystrokes.
Though any of those items might affect how much credit to give to a particular player for the points scored or surrendered, they are not the same thing as deriving a *team’s* victory total from the box score. While perhaps recording these things would better refine the distribution of credit, distributing it simply by what *is* recorded has a rather good predictive value. Much of the game is recorded, certainly enough to know quite a bit about what happened and how this is likely going to influence future results.
kevin
December 7, 2008
I see I’m dealing with a rather opaque person here.
In order to deconstruct the points, you have to understand how they are being created and prevented. Since you yourself included the qualifier “individual” here, just counting up the points a player scores doesn’t tell you very much about relative contributions towards wins and losses.
This is so obvious it really shouldn’t bear mentioning.
kevin
December 7, 2008
“Much of the game is recorded, certainly enough to know quite a bit about what happened and how this is likely going to influence future results.”
This simply isn’t true. If it were, then, perusing the statistics of players like Kiki Vandeweghe and Dennis Johnson, one would think the former’s a superstar and the latter’s a no-talent . It’s this kind of thinking that gets guys like Hollinger in so much trouble. He doesn’t seriously critique his own BS and is seemingly unaware of the weaknesses in the raw data.
dustin
December 7, 2008
kevin,
“Though any of those items might affect how much credit to give to a particular player for the points scored or surrendered, they are not the same thing as deriving a *team’s* victory total from the box score.”
It appears you imply jason is dense, then proceeded to paraphrase him…
Jason E
December 7, 2008
Kevin, I suggest you learn a bit more about actual analysis of statistical measures. It is not the same as “perusing”.
kevin
December 8, 2008
Jason, it appears you’re not the person who should be making that judgment, since you seem so confused about how questions can be addressed by using them.
Jeremy
December 11, 2008
Kevin,
good pt about Rondo destroying Ford last week. Rondo destroys everyone now. Id love to see some up to date league wide WP48 stats.
Dominik
January 2, 2009
Nice post, Professor Berri. Your book has made my marketing research class over here way more interesting.
The Raptors are like a park in Chernobyl – they have nobody sittin’ on the bench.
Kind regards,
Dominik from Germany
dberri
January 2, 2009
Dominik,
Just curious… who would assign The Wages of Wins in a marketing research class?
Dominik
January 2, 2009
Sadly, The Wages of Wins wasn’t an complementary literature assignment in our marketing research class. It just made me enjoy going through the regression analysis chapters and working with SPSS.