On December 23rd Yao Ming fractured his leg. As a result, he missed the next 32 games. One would expect that losing a center that John Hollinger has named as the most likely current big man to be listed someday as one of the top ten centers of all time (insider only) would severely harm a team. Yet in the 32 games Ming was lost to injury, the Houston Rockets managed to win 20 times. This 0.625 winning percentage exceeds the 0.586 winning percentage the Rockets have posted with Ming in the starting line-up. So does this mean that Ming is really not that good?
Not exactly. Entering this season Ming had produced 45.7 wins in four seasons. His Wins Produced per 48 minutes [WP48] stood at 0.231, which more than doubles the WP48 of an average player (0.100). This season Ming’s WP48 stands at 0.209, a mark that is again well above average. Clearly Ming is a “good” player.
How can losing such a “good” player, though, not impact the team’s ability to win?
To answer this question, let’s turn to the Wins Produced of the Houston Rockets after 61 games.
Table One: The Houston Rockets after 61 games
The Rockets players are ranked in terms of Wins Produced, and at the top we see Dikembe Mutombo. Mutombo entered the league in 1993 and is currently 40 years of age. Since leaving Philadelphia in 2002, he has played limited minutes for New Jersey, New York, and the Rockets. Prior to Ming’s injury, Mutombo was averaging less than 10 minutes per game this season. Given his age and limited minutes this season, I fully expected the Rockets to look elsewhere to fill Ming’s missing minutes.
The Rockets and Mutombo, though, had other ideas. In Ming’s abscence Mutombo played 27.5 minutes per night. And surprisingly, his per-minute Win Score of 0.382 surpassed the 0.294 per-minute mark offered by Ming this season.
When we turn to WP48, we see 0.340 for Mutombo, a record that again surpasses the production of Ming.
Although it’s surprising to see this production from an average 40 year old player, Mutombo has spent his career as an above average performer. To illustrate, compare the career marks of Mutombo and Shaquille O’Neal.
Table Two: Dikembe Mutombo vs. Shaquille O’Neal
ESPN recently listed Shaq as the fourth greatest center of all-time – behind Kareem Abdul-Jabbar, Wilt Chamberlain, and Bill Russell. Entering this season Shaq had produced 244.7 wins and posted a career WP48 of 0.338. Although I do not have career numbers of Abdul-Jabbar, Chamberlain, and Russell (because the NBA did not track all the stats prior to 1973), we can see that Shaq is quite a bit better than Yao Ming.
Surprisingly, though, Mutombo – who did not receive any consideration as one of the all-time greats at his position — is not far behind Shaq. Mount Mutombo has produced 183.6 wins in his career and offered a 0.303 WP48.
What do we learn from all these numbers?
- Yao Ming is very good.
- Shaq is much better.
- And Mount Mutombo appears very much under-rated. He is more productive than Ming, although not quite as good as Shaq.
Fortunately for the Rockets this mountain of productivity was waiting on the bench when Ming was hurt. As a result, the Rockets are still in the playoff picture with a month remaining in the season.
Unfortunately for Houston, the NBA considers it a Western city. So although the Rockets are at least as good as anyone in the East, it’s not likely that Ming, Mutombo, and McGrady – with support from Luther Head and Chuck Hayes – are good enough to knock off any of the big three out West.
But that ultimate disappointment should not detract from the fact that playoff basketball will come to Houston this year. And it looks like the team has Mount Mutombo to thank for that event.
– DJ
Kent
March 8, 2007
Dikembo Mutumbo for President.
Owen
March 8, 2007
Amazing that Dikembe is better statistically than Yao.
It would be interesting to ask a panel of sports writers the following. How do you explain how Houston went 17-12 with one of the future best centers of all time and 20-12 with him injured? This is exactly the kind of puzzle that your theory answers very easily but which I feel they might struggle with.
Excellent responses to Silverbird.
Kent
March 8, 2007
Owen for Vice President!
Ron
March 9, 2007
Just to clarify the rather unique structure of Yao Ming’s name. In Chinese culture, the surname or family name comes before the given name. So for the Rockets center, his last name is Yao, while his first name is Ming.
Bill
March 9, 2007
Dave –
I was looking for an analysis of the 06-07 Seattle Supersonics.
You recently mentioned that you’d covered every team this year, but I didn’t see Seattle listed on the summary page.
Any help would be appreciated. Sorry to stray OT.
-Bill
anon
March 9, 2007
Bill,
Dave actually reviewed the sonics in this post (scroll down to the bottom):
https://dberri.wordpress.com/2007/01/21/nba-power-rankings/
Not sure why it’s not linked in the summary page
Travis
March 9, 2007
Lies, damn lies and statistics. Very misleading stats. This does NOT factor in the known fact that T-Mac has significantly elevated his game since returning from injury that coincidently just happended to be the same time that Yao went out. The Rockets success during Yao’s absence has a lot (and I mean A LOT) more to do with the play of T-Mac than with the play of Mutombo. Deke is a great backup, but he does not offer anything near what Yao offers the Rockets. Another example of correlation being confused with causation.
Steve
March 9, 2007
And can you really extrapolate a 40-something-year-old guy’s per-minute stats over a whole game? I don’t think Mutombo could really play any more minutes that he does.
anon
March 9, 2007
Actually, I think you’re probably more guilty of correlation being confused with causation than DBerri. Rockets resurgence correlates with T-Mac’s improved numbers and return from injury. Rockets resurgence also correlates with Deke astonishing play as a Yao replacement. How do you know which one them caused the Rockets to win more games? DBerri suggests you look at win score.
You’re assuming a causal link with less compelling evidence than DBerri.
Travis
March 9, 2007
No, I am not confused in the least. It is a well known fact that T-Mac had some fairly progressive treatment done to his ailing back that he said himslef made it feel better than it has in years. He, not me, credits that with his resurgance. That, in my mind, would be the most likely causation. Not the absence (or presence) of Yao. Also, as long as we are goping to take a quantitative look at things, show me how Deke’s stats have been better than Yao’s (and I am not talking a 48 minute per stuff). Yao has a significantly higher scoring avg., slightly higher blocks, higher assists, and comparable rebounding numbers. Have DBerri look at the Rockets win score pre and post T-Mac’s back treatment and you will find the true independent variable driving the independent variable of wins.
Jack Mott
March 9, 2007
Travis, i have broken down some of their key stats on a per 48 minute basis for you:
blocks : tied at around 3
turnovers: 1.5 for dikembe, 5.4 for yao
assists: .5 for dikembe, 3 for yao
rebounds: 18.5 for dikembe, 13.3 for yao
fg% 56.6 for dikembe, 51.5 for yao
personal fouls: 5.5 for dikembe, 5.0 for yao
So it appears that dikembes main advantages are in turnovers, and rebounds, and field goal percentage.
I think if you broke down the wins produced per 48 minutes you would find that turnovers are the main contributor
Jack Mott
March 9, 2007
Obviously it would be useful to see tracy mcgradys wins produced before and after the Yao Injury to see what role that plays
Travis
March 9, 2007
The problem with that being that Yao went out at the same time that T-Mac returned. Thus my point re: causation vs. correlation and the misleading stats. Not to beat a dead horse here, but it’s all about T-Mac’s post injury / treatment production not Yao vs. Deke.
Case in point, The Rockets had lost 5 out of 8 prior to their win v. the Celts. Yao was out during most of that time (with the exception of the Cleveland game). So where was the Deke magic then? The answer is, that isn’t the answer! If you look at T-Mac’s production during the mini-skid (as well as Rafer’s and Head’s), you will find his numbers done a good bit. T-Mac was the most significant reason for the Yao-less run, not Deke.
Jack Mott
March 9, 2007
Travis, you are telling a plausible story. Maybe McGrady was more of the reason than Dikembe. At this time, we really don’t know yet, we would need to compare the wins produce delta on McGrady before and after. Is he really producing better or just throwing up more points since he has to take up the scoring load from Yao Ming?
If it is the latter, that doesn’t mean much.
Did his FG% rise? rebounds?
Jack Mott
March 9, 2007
I found some splits on TMAC, his minutes, rebounds, assists blocks are all pretty constant. His points did rise but not his FG% really, maybe his 3 point %… what month did yao get hurt?
http://sports.espn.go.com/nba/players/splits?statsId=3179
anon
March 9, 2007
The basic premise of correlation v. causation is that someone makes logical leap from correlation to causation without sufficiently investigating the matter. T-Mac’s return correlates with the Rockets. Therefore, with little statistical evidence, you conclude this correlation must mean that T-Mac is the reason the one. How do you know this? You don’t. Your account is a slight, but only slight, refinement of correlation means causation. You see correlation and “observe” that T-Mac is playing better.
DBerri made no observation and instead looked at WS. You can argue that his account of causation is flawed because it is insufficient. But he’s not substituting correlation with causation. He’s making a statistically informed guess (albeit, one that may not be correct).
You, on the other hand, are unequivocally substituting correlation for causation.
Travis
March 9, 2007
Good question. Yes, his FG% went up 3pts post injury (which is actually a lot). Haven’t run the assist, rebs, 3 point FG%, and FT attempts, but my guess is those numbers are up as well.
He is the reason for the Rockets success. he has been playing out of his mind.
dberri
March 9, 2007
I am impressed by the level of discussion today. I am looking at T-Mac’s numbers. There is some evidence that he played better after Yao was hurt. But I don’t think the evidence fits the story Travis is trying to tell. If I get a chance, I will post on this today.
Travis
March 9, 2007
Excuse me, but what is the entire premise behind the win per stat? That certain players contribution contribute more to a team’s wins than others. Tell me how that is not casuation. If he is merely looking at correlations and ignoring (or at least not implicitly implying causation) then let’s look at the correlation between air temperature and the Rocket’s wins. or, maybe we could look at the number of cheeseburgers consumed by Van Gundy vs. wins. If there is no suggestion or implicit causation assumed by the author then why would we care? It would be truly academic.
Jack Mott
March 9, 2007
Travis do you really thinkg if we removed Mutmbo and put in a center with an average Wins Produced value, that the rockets would keep winning?
The delta between an average center and Mutumbo is definitely larger than TMAC before and after the yao injury.
Travis
March 9, 2007
Or, maybe Deke benefitted from resurgent T-Mac (which I think is more likely the case). Don’t get me wrong, Deke was a valuable contributor, just not the real driver in my mind. I will turn the question around to you Jack, do you really think that if Yao had been there vs. Deke that the Rockets wouldn’t have done as well or better. Do a WPV on when both Yao and T-Mac play and I think you will see where the real magic of the team lies.
Owen
March 9, 2007
great thread guys…
Steve
March 9, 2007
“The delta between an average center and Mutumbo is …”
Answer: there’s no difference. Mutombo is pretty average these days, which is great for a guy his age.
Seriously, ignore Mutombo’s per-minute stats. The guy is 150 years old. He’s still a solid contributor, a very good backup, but he can’t perform at Yao-like levels for an entire game.
Evan
March 10, 2007
The funny thing is that 3 Rice guys from Hanszen comment on this site. Small world.
Paul
March 11, 2007
Or maybe the following happened : with the offense running through him more (since mutombo nevercould score) Mc Grady is playing better, thus the improved Rockets performance
Travis
March 20, 2007
Make that 4 Rice Guys (but not all from Hanszen)!
Bhavik Shah
March 21, 2007
Did you take into account difficulty of Schedule. Yao was injured on December 23rd, 2006 and returned March 5th, 2007. Motumbo started in the 32 games in between. The Rockets ended up winning 20 of those games and losing 12. Of the 20 games won, 13 (65%) were against teams that currently have under a 0.500 winning percentage. Another way of saying this is Motumbo’s winning percentage:
1. Against Above .500 teams is 50%
2.Against Below .500 teams is 72%
If you look at the 35 games Yao started (and completed) this season, Yao won 23 games. Of the wins, 8 were against teams that are currently above .500. Of the 12 losses, 4 (33%) were to teams with currently below .500 records. So that means, Yao’s win percentage:
1. Against Above .500 teams is 50%
2. Against Below .500 teams is 79%
So currently, Yao’s win percentage this year is slightly better than Motumbo’s. These statistics are from the beginning of this season up until today (March 21st, 2007).