It’s not often you see a 65-year old man referred to as a boy-toy. But this is what Mark Cuban called Phil Jackson this week. Cuban’s “odd” statement was in response to another “odd” statement from Jackson.
In response to the season-ending injury suffered by Caron Butler, Jackson said (according to the Los Angeles Times): “He just leaves a vacuum that’s going to be very hard for them to fill.”
Jackson attempted to clarify this comment with the following:
“I feel badly for [the Mavericks], that’s what I was saying. It’s hard to replace a player that good,” Jackson said. “They do have a good player that’s sitting behind him. Shawn [Marion] is a fine player, but it’s not Caron Butler, so it’s hard to replace a player like that.”
Hopefully it is easy to see why Cuban’s comments about Jackson are best described as “odd”. What about Jackson’s comments about the Mavericks and Caron Bulter?
First of all, I am not sure I believe that Jackson really feels badly about the Mavericks. But beyond this point, I also don’t agree with his assessment of Butler.
Before we get to what Butler has done this year, let’s note the following numbers:
- Entering this season, Butler had produced 52 wins in his career with a 0.122 WP48 [Wins Produced per 48 minutes]
- His best season was in 2007-08, when he produced 10.4 wins with a 0.216 WP48
- Last year, though, he produced 4.7 wins with a 0.080 WP48
- Butler is now 30 years of age.
Okay, now let’s look at the numbers for this year. And let’s not just look at Butler. Let’s look at the Mavericks in 2010-11.
After 34 games, the Mavericks have won 26 games and currently have the second best record in the Western Conference. This record, though, is a bit of an illusion. Currently the Mavericks are scoring 105.7 points per 100 possessions, while allowing 100.6 points. So the team’s efficiency differential is 5.1. Such a mark is consistent with a team that will win around 53 or 54 wins across an entire season. This is close to what was projected for the Lakers earlier this week, but in terms of efficiency differential, Phil Jackson’s team has the edge.
Although the Lakers have the edge, the Mavericks have been quite close. Does the injury to Butler make Jackson breathe a bit easier (contrary to his public comments)?
To answer this question, let’s move from efficiency differential to Wins Produced. The following table reports the productivity of the Mavericks’ players this season.
Again, the Mavericks are projected to win about 53 games. Of these wins, 48.6 can be linked to the play of Jason Kidd, Tyson Chandler, Dirk Nowitzki, Shawn Marion, and DeShawn Stevenson. These five players are the only Dallas players to play more than 100 minutes and post a WP48 beyond 0.100. In other words, the remainder of the roster – a remainder that includes Caron Butler – are either not playing much and/or are below average.
If we look at Butler specifically, we see a player who has posted a 0.018 WP48 and is on pace to produce 0.8 wins. Such numbers suggest that the loss of Butler does not leave “a vacuum that will be hard for the Mavericks to fill.” And moving from Butler to Shawn Marion is probably not a big problem.
What is a big problem is the loss of Dirk Nowitzki, who has missed the last six games due to a knee injury. Behind Nowitzki is Brian Cardinal, a player who has produced in the negative range in each of the past three seasons. And he is in the negative range this season as well (if only barely). So moving from Nowitzki to Cardinal isn’t very helpful.
Reportedly, though, Nowitzki only has a knee sprain. And when he returns, the Mavericks will have a fairly impressive starting line-up of Kidd, Stevenson, Marion, Nowitzki, and Chandler (or maybe, Kidd, Jason Terry, Stevenson, Nowitzki, and Chandler).
Two more observations on this team:
First, Tyson Chandler appears to be back. Here are the WP48 numbers Chandler has posted across the past seven years:
- 2009-10: 0.093
- 2008-09: 0.078
- 2007-08: 0.244
- 2006-07: 0.301
- 2005-06: 0.234
- 2004-05: 0.333
- 2003-04: 0.228
As one can see, Chandler was far above average for five consecutive seasons. But injuries (and it is injuries that seem to change player performance more often than not) caused his production to fall the past two years. Now Chandler appears to be healthy. So the Mavericks acquisition of Chandler is looking good.
The other player who has improved is DeShawn Stevenson. Prior to this season, Stevenson had produced 0.6 wins for his entire career (a career that stretches back to 2000-o1). His best season was 2007-08, when he produced 4.4 wins – and posted a 0.082 WP48 – with Washington. So at his best, Stevenson was below average. This year, though, his WP48 is 0.139. What explains the difference? Currently Stevenson has an adjusted field goal percentage of 0.634. Obviously this mark is well above average; not just for Stevenson, but for any player in the NBA. And it seems obvious (but it may not be) that Stevenson can’t keep hitting shots at this rate. After all, for his career he has hit on less than 35% of his shots from beyond the arc and this year he is hitting on 47% of these shots.
If Stevenson can’t keep hitting his shots, the Mavericks might decline a bit. Nevertheless – even if Stevenson reverts to form — the Mavericks are a “good” team. But I don’t think they are a “great” team or a title contender in 2011 (at least, I don’t think they are as good as the Heat, Celtics, and Spurs). And I think that assessment stands, with or without Caron Butler.
- DJ
I think your projection for season wins here (and in similar posts) is incorrect, and reflects the gambler’s fallacy.
Dallas’ current wins produced suggests that they should be expected to win 65% of their games (53.3/82). Let’s take that as a given. They’ve been significantly outperforming that so far — they’re actually 26-8, 76.5%. You suggest that they should end up with a .650 record. To do that, they would have to win significantly fewer than 65% of their games for the rest of the season.
However, their current overperformance does not mean that we should expect them to underperform for the rest of the year — they should still be expected to win 65% of their future games. (Flipping a lot of heads does not reduce the chance of heads coming up in a later flip.) Thus, their expected win total for this season should be the 26 games already won + .65*(48 games remaining)=57.2, unless I’m missing something.
dberri-
Do you have any comment on David Lee’s precipitous fall this season with GSW? (I would have emailed this question directly, but I didn’t see an email address)
Nevermind about David Lee, I’d totally forgotten about his strange and horrific injury.
Fred Bush, it’s statistics. In any small sample size there is a lot of “noise” that goes on that can disrupt your results. But the larger your sample gets, the more the noise evens out and you tend to get the results you expected. The gambler’s fallacy is that because it departs from the average, that you expect it to end soon. No one’s saying they’re going to go on a 10 game losing streak this month. But things WILL average out eventually, otherwise it wouldn’t be an average. It’s the Law of Large Numbers.
Take a coin toss, for instance. The Gambler’s fallacy would be that if it came up heads 8 times in a row, I expect a tails in the next flip. And you’re right, it’s exactly 50/50. But the Law of Large Numbers states that, yes, the next flip is 50/50. But over 100 flips it’ll be about 50/50, despite the original 8 heads in a row.
Fred’s assertion points to a very large problem in statistical analysis. When people want to disprove a model they don’t take enough time to understand the model. This reminds of those other guys trying to disprove WoW by attacking the team adjustment without realizing where the team adjustment came from. Those are pretty smart guys, and yet they wasted all that time because smart people are used to their assumptions being correct. When getting into a model you can’t assume anything. Underscore anything.
So yes 95% of the time Fred would be correct and its a safe assumption to make on his part. However in this instance he is incorrect.
Some other thoughts. Wouldn’t it be great if the league figured out a way to keep teams from holding players hostage and vice versa? An example would be a system where if a guy is voted an all star starter and yet his team fails to have one of the ten best records he can sign with another team under his current contract. Conversely, if a player signed one of the ten highest contracts and his team fails to have one of the ten best records then the team can void his contract.
You could have a similar system for second teir players who make an all-star team while their teams fails to make the playoffs.
This would protect teams from the selves and protect great players from being held hostage.
Right now, OKC would be in danger of losing Kevin Durant and Joe Johnson would be in danger of losing his contract
Fred,
Is all that you are asking is that given that the Mavs have exceeded expectations so far that if they continue at “current rate” that the bonus they’ve earned will remain? I’m curious to that as well actually.
The other part that DJ mentions is some of the success has been on things that will likely regress to the mean, so it is true that the Mavs will get a little worse, but that shouldn’t discount the fact that they are overperforming their numbers right now. (This can happen in short stints, the playoffs are notorious for this)
For the benefit of non-stat geeks in this thread: let’s say that the Las Vegas Reds in the National Roulette Association are 50/50 to win any game. However, at this point in the season they’re at 18-14 due to pure luck. 50 games left.
Their expectation is to win 25 of their last 50 games and finish at 43-39, not to “even out” to .500 by the end of the season.
Xavier, you can see that they’re at .5625 right now and their season expectation is to finish at .524, which is much closer to their per-game theoretical expectation of .500, but they won’t get all the way down to .500 by the end of the season; 50 more games is not a Large Number.
Wins produced and actual wins are not a perfect correlation mostly because of the role of chance in how a team performs in close games, as I understand it. I think Prof. Berri wants to say “they’re playing like a 53-win team”, but he is saying instead “I expect them to win 53 games this season”. At the beginning of the season, those two statements are equivalent, but at this point we can account for some of the luck factor, so that should be considered when projecting season win totals.
To be clear, this statement
“Such a mark is consistent with a team that will win around 53 or 54 wins across an entire season.”
I accept,
but this statement:
“Again, the Mavericks are projected to win about 53 games.”
I reject.
Fred, why don’t you come back at the end of the season and say “I told you so.” But, in the event that you are wrong, you should come back and say “WoW. I gotta give you some credit for that one, Berri.”
I just go off of my gut, personally. I don’t think the Mavs will win more than 50 games this. So, yes there is a 10 game skid in the horizon and some additonal hicups (Dirk or no Dirk).
At this point I get the feeling Jason Kidd will still be throwing off WP48s north of .200 three years after he’s dead.
Fred is correct.
If a legitimate 53 win team gets off to a very good start because of either luck or temporary above average performance by several players, it is expected to win more than 53 games at the end of the season.
1. You cannot assume a streak of bad luck to revert things back to the mean.
2. Even if out performing players revert to the mean for the remainder of the season (which could very well happen) it would tend keep the team on a 53 win pace for the “remaining games” which will tend to lead to more than 53 wins because of the good start.
I don’t think this is an issue because I’m sure D Berri understands this.
This is more of a lesson on communication than a lesson on statistics. Since I’m a poor writer and do this kind of thing all the time, I don’t see it as worthy of much conversation.
I didn’t respond to FB for much the same reason IS. I interpreted “Such a mark is consistent with a team that will win around 53 or 54 wins across an entire season.” as the same as “Again, the Mavericks are projected to win about 53 games.” Both statements can be interpreted several different ways which are all correct in context. Even taking the harshest interpretations, it’s kinda nitpicky since 53 wins is similar to 57.
Good post, although I’d discuss one other player beside those that Prof. Berri discussed: Jason Terry. Terry has been a good player (WP48 > .100 ) for most of his career, until last year. And this year he is also below average. If Terry can revert to having at least a WP48 of .100, then the Mavs should be expected to be even better than Prof Berri noted, even if Stevenson declines. On the other hand, Terry is getting fairly old, so perhaps we shouldn’t expect him to revert back to his prior form.
Isn’t it possible that the lower scoring margin is a result of one of the toughest schedules in the league? The Mavericks have a 0.53 SOS which is 2nd in the league.
Now we can assume that they will play weaker opponents until the end of the season which can bring their expected wins closer to the real amount of wins. Seems as reasonable as an assumption to me as saying they will not win as many games as their record indicates.
It is also a fact that the Mavericks are typically outscoring their opponents in close games. That isn’t something which is new in this season. In fact over the last 7 years the Mavericks outscored with Nowitzki on the court their opponents by +17.3 points per 36 minutes in the final minutes of close games. That is +12 per 36 minutes above their average scoring margin in all other minutes. Their ability to execute the high pick&roll/pop with Terry and Nowitzki in the 4th quarter is the reason for that. Assuming that will go away all of the sudden seems a bit weird to me (well, probably not that weird at all, if I factor in the current injury to Nowitzki).
Btw:
Kidd: 0257 WP/48
Barea: -0.018 WP/48
Nowitzki: 0.229 WP/48
Cardinal: -0.001 WP/48
Well, according to those numbers the difference between Nowitzki and his replacement is smaller than between Kidd and his replacement. That means Kidd is more important to the Mavericks than Nowitzki by that metric. If we check the reality it shows us that the Mavericks are outscoring their opponents without Kidd on the court while they get completely destroyed without Nowitzki. Either that metric isn’t really working or the reality is just wrong. I guess it must be the reality… :D