A few weeks ago the Dallas Mavericks traded for Jason Kidd. At the time, this move seemed somewhat desperate. Yes, Kidd is one of the most productive players in the game. But at 35 years of age, he seems to be closing in on the end of his career (although his production suggests otherwise).
At the time the move was made it was clear the Mavericks had to do something. Last year – led by MVP Dirk Nowitzki – the Mavericks boasted a 7.8 efficiency differential (offensive efficiency minus defensive efficiency). No, this wasn’t the best mark in the league (the Spurs had a mark of 9.1). Still, Dallas was clearly a title contender in 2007 (despite the actual playoff results).
When this trade was made, though, the Mavericks differential was only 4.3. Six teams bested this mark in 2006-07. So it didn’t look like Dallas was going to contend for a title this year (see The Missing MVP for an explanation of why this team had faltered).
To make the trade work, though, the Mavericks had to give up Devin Harris and DeSagana Diop. Harris and Diop are not quite as productive as Kidd. Still, both are above average and hence the benefit of adding Kidd was reduced.
In fact, when we look at the team’s record, it looked like the subtractions actually exceeded the addition. The Mavericks efficiency differential when this trade was made was consistent with a team that would win 52 games in a full season. Such a team could expect to win about 11 out of each 17 games played. But in the 17 games the Mavericks played with Kidd on board (and yes, that is a small sample), the team only won 10 contests. And all seven losses were to teams with winning records.
Of course, wins and losses are not the only way we evaluate teams. When we turn to efficiency differential, we see the Mavericks posted a mark of 7.4 in the first 17 games after Kidd was acquired. This mark is consistent with a team that would win 60 games in a full season, or 12.4 of each 17 games played. In sum, it looks like Kidd – who is just about as productive with the Mavericks as he was with the Nets (Win Score per 48 minutes was 12.7 in both places) — actually did help Dallas.
Unfortunately, after 17 games the Mavericks lost Nowitzki. With Kidd, Nowitzki had raised his WP48 (Wins Produced per 48 minutes) from 0.218 to 0.319 (similar to the 0.306 mark he posted last year). Without Nowitzki, though, the Mavericks now have to turn to Malik Allen (WP48 of 0.026), Juwan Howard (WP48 of -0.113), and Brandon Bass (WP48 of 0.055).
Bass is the best of the bunch, and he is still below average. If the Mavericks were to replace Nowitzki with Bass for an entire season, the team would lose about 16 more games. In other words, the 60 win title contender would morph into a 44 win also-ran. Fortunately, Nowitzki is not lost for the entire season, just a few weeks. Nevertheless, in a very competitive Western Conference, the loss of Nowitzki might be enough to drop Dallas out of the playoffs completely.
And if that happens, I suspect people might think the Kidd trade was clearly a mistake. I think, though, that the numbers tell us – both before and after the fact – that this move was a good idea. Unfortunately, a bit of bad luck here and there, certainly makes the story appear quite different.
– 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.
Sam Cohen
March 29, 2008
Was Dirk’s play improving before the Kidd trade (i.e. what was his WP48 in the 17 games before the trade?), or did his improvement truly coincide with the arrival of Kidd?
I guess I’m just surprised the efficiency differential changed that much. When I had glanced at the per possession stats last week, I thought I had seen that the Mavs defense had gotten better with Kidd, but it also looked like the Mavs offense had gotten worse.
dberri
March 29, 2008
Sam,
The defense is much better with Kidd. Offense is slightly worse (although still quite good).
Across the first 28 games Dallas played, Nowitzki had a 0.176 WP48. Over the next 25 games (right before the trade) his mark was 0.265. With Kidd, it was 0.306.
I would note that it is not necessarily because of Kidd that Nowitzki improved. He has played this well before (without Kidd).
ty w
March 29, 2008
You’re absolutely right. I was just going to write the same thing on Bucks Diary in response to all the clap trap about what a great “future” prospect Dallas gave away in Devin Harris. Harris as a Maverick was what he was and that was about all he was going to be.
As for the trend lines on the other major trades: the Suns in March appear to have gotten back to their Pre-Shaq levels on offense but they are a bit worse on defense; the Lakers (when they have Gasol) have been better on offense in March, but worse on defense; and the Cavs are better on both offense and defense, but only slightly.
tony cohen
March 30, 2008
with the point differential going up, yet quality wins literally disappearing, it seems the point spread may not tell us everything we need to know.
Maybe there is some truth to playing down to the level of the competition. Maybe the Mavs are one of those teams that doesn’t, so when they play an inferior team, they whip them…but when they run into a quality team, they show exactly how good they truly are.
Perhaps an interesting analysis might be in constructing a point spread based on winning percentage of an opponent. For example, have one spread against teams with up to a .350 winning %, another from .350 to .425 and so on. I wonder if this would show some interesting deviations across the league…at the least it would make for interesting post about which teams play down to the level of their competition.
Evan
March 30, 2008
Tony —
Wouldn’t that suffer from a huge sample size problem?
Brad
March 30, 2008
This is exactly what Harold Almonte said would happen.
antonio
March 30, 2008
i also think because of sample size in this situation, win differential does not tell us as much as it usually does. each game has too much of an effect on the differential, and a 30-point blowout would be given too much weight
Brad
March 30, 2008
tony, it’s even worse in football with only 16 games. What I do is calculate it game by game so 40 point blowout vs 30 point blowout counts the same but 11 point win counts a lot more than a 1 point win. You get much, much better results this way.
Brad
March 30, 2008
Under my approach you get 0.51 wins for winning by 1 point and 0.99 wins for winning by 40 points. For those 2 games combined you get 1.5 wins and an expected 75% winning percent for the season. Under the other way you get an expected winning percent of almost 100%!
Sam Cohen
March 30, 2008
I guess I’m just curious about the Kidd-Nowitzki connection because I seem to recall the major reason for the trade (aside from “intangibles”) being that Kidd would get Nowitzki more open looks.
If Dirk was doing better (especially offensively) before Kidd arrived, then it seems like a trade that ended up working, but not for the reasons the team decided to make the trade. (i.e., the Mavs made the trade for offense, but gained on the defensive end)
Animal
March 30, 2008
What happened to TG Randini?
Harold Almonte
March 31, 2008
Harris will probably be in the future what he is right now, but he has a future; Kidd, with a Harris’level in the present (and a HOF level in the past), at his 35 and aging, doesn’t have one.
Although Dallas was losing efficiently, they won the “must win” games and lost the “must lose” games they probably would have won and lost without a trade. You can say that basically, in the short valid sample size, the trade was even. They just changed PG’s points to PG’s help+other’s points at the same level.
Christopher
March 31, 2008
“Of course, wins and losses are not the only way we evaluate teams. ”
Actually, I would disagree. You make trades to better your franchise, to win more games, go deeper into the playoffs etc. Post-trade the Mavs can not really past muster against a team playing .500 or better ball. And are 10-9 after the All-Star break; Dallas was at .658 pre-trade. Dallas made that trade to get them over the hump. There is no long-term plan with a 35 y.o. PG. FWIW, REBS and AST per game are unchanged (pre vs post- All Star break). TO per game are 0.1 worse. PPS has improved from 1.26 to 1.28. The big increase one might have expected comparing WoW metrics of Harris to Kidd is just not there. The trade was a failure but will now not look as bad because so much will be heaved on losing Dirk. This panic trade and Cuban’s odd blog policy are really diminishing his star…
dberri
March 31, 2008
Christopher,
Just thought I would toss this out there…In the 18 games after the trade, rebounds for the Mavs increased by 3.1 (entirely on the defensive side, ORB virtually unchanged). Assists were also up.
That doesn’t change the fact that Cuban’s blog policy is nothing short of strange.
Bball
March 31, 2008
What’s Cuban’s blog policy?
MarkT
April 1, 2008
Ty W’s comment about Harris never being better than he was with the Mavs is bizarre. Since the trade, the guy has had career highs in every stat.
On the subject of the Kidd impact on Dallas, I think the sample size is still too small to evaluate it.
Harold Almonte
April 1, 2008
The impact? He improves the team where it was already good: mid and long distance shooting. It could mean to survive one or maybe two playoff series without depending too much on hothand nights, but it doesn’t make the team champ material. The trade didn’t make up the main Dallas weakness: inside offense.
Christopher
April 3, 2008
“On the subject of the Kidd impact on Dallas, I think the sample size is still too small to evaluate it.”
Disagree. The trades is supposed to help NOW. There is no long range frequency interpretation here. Right now, the trade has not helped. Only last night did the Mavs finally beat a .500 or better team in 11 tries since the trades). It’s not working out as planned.
“Just thought I would toss this out there…In the 18 games after the trade, rebounds for the Mavs increased by 3.1 (entirely on the defensive side, ORB virtually unchanged). Assists were also up.”
I looked at pre and post All-Star splits as ESPN as my break. This roughly matches the trade but is obviously not a perfect fit. If there’s another site that allows sortable team stats based on trades etc. I’d love a link.