Talk Me Down: How is Deng not MVP of the Bulls?

[coldfish]

The lineup matchups from each game. Everytime a new player comes in, a new line in the matchup file is created. We have the data of 5 home players vs. 5 away players. The basic formula is:

Margin = HCA+a_1P_1+a_2P_2+a_3P_3+a_4P_4+a_5P_5-a_6P_6-a_7P_7-a_8P_8-a_9P_9-a_10P_10

where Margin is the scoring margin adjusted to 100 possessions, HCA is the intercept, a_1 to a_5 the coefficients for the home players, a_6 to a_10 the coefficients for the away players. Obviously, the index runs from 1 to ca. 450 (one index for each player). The regression finds the best fitting coefficients in order to approximate Margin. That boils down to matrix algebra. For the ridge regression a lambda is added and the coefficients are depending on the lambda now. For each lambda you get a different set of coefficients. It was shown that there is always a lambda for which RMSE (ridge) < RMSE (OLS). The best fitting lambda can be found via crossvalidation.

FWIW, I'm pretty dam impressed. I didn't think that kind of information was even available. I'm the guy who made the bad assumption.

It still doesn't match the eye test and the only hypothesis I have is regarding unit mixing and sample sizes, but I can't poke holes in the methodology as easily as I thought.

[mysticbb]

It still doesn't match the eye test and the only hypothesis I have is regarding unit mixing and sample sizes, but I can't poke holes in the methodology as easily as I thought.

I would say that the human error and bias is one of the reasons. Overall I think you are much more in agreement with those results than you might be willing to admit. Take the example of the Knicks. No idea, but somehow I had the impression that you were pretty sure that the numbers would say something different than your own instinct. It can also be the case that the adjustment to 100 possessions is not taken into account when evaluating a list. The results are not giving back the best player overall in order, but they sort the players in order of their impact per possessions they played.

[coldfish]

It still doesn't match the eye test and the only hypothesis I have is regarding unit mixing and sample sizes, but I can't poke holes in the methodology as easily as I thought.

I would say that the human error and bias is one of the reasons. Overall I think you are much more in agreement with those results than you might be willing to admit. Take the example of the Knicks. No idea, but somehow I had the impression that you were pretty sure that the numbers would say something different than your own instinct. It can also be the case that the adjustment to 100 possessions is not taken into account when evaluating a list. The results are not giving back the best player overall in order, but they sort the players in order of their impact per possessions they played.

On other boards, Deng gets regularly trashed by casual fans. I have been arguing for quite some time that his impact on winning is far more than his box score numbers say, and Thibodeau obviously agrees.

That being said, I still think Rose is more important to the Bulls than Deng. I'm not even sure its close. I'm just not sure how to tease it out of the stats at this point.

One thing that has been discussed on the Bulls board is that Rose is "cruising" a lot more this year. That would be really hard to quantify, but you can see it in his usage number, which is down from last year. Basically, Rose has another gear he can regularly turn on when the team needs him whereas Deng doesn't seem to. I know, that's really subjective but its an impression we have got.

One other random question I would have is the amount of times that Deng gets assisted by Rose. I think the number is relatively high and its part of the reason why Rose gets credit because Rose seems to get Deng going. I would like to see what percentage of Deng's baskets get assisted by Rose when he is on the court versus when Watson or someone else is.

Again though, good job with the stats. Your fundamental methodology is exactly what I would have done (I not even touching the ridge analysis or anything, I mean the 5 man unit evaluation method). I hadn't read through the thread to see that. I NEVER would have thought someone had the information and the gumption to actually do what you did.

[Rapcity_11]

One thing that has been discussed on the Bulls board is that Rose is "cruising" a lot more this year. That would be really hard to quantify, but you can see it in his usage number, which is down from last year. Basically, Rose has another gear he can regularly turn on when the team needs him whereas Deng doesn't seem to. I know, that's really subjective but its an impression we have got.

Does it matter though? If Rose is truly "cruising", we'll see his impact jump come playoff time, right?

And for now it's all about what guys are actually doing, not what they're capable of doing.

[rrravenred]

One other random question I would have is the amount of times that Deng gets assisted by Rose. I think the number is relatively high and its part of the reason why Rose gets credit because Rose seems to get Deng going. I would like to see what percentage of Deng's baskets get assisted by Rose when he is on the court versus when Watson or someone else is.

The raw numbers are easy enough.

PbP data from BR indicates that this year it's 35 of 190 FG (18%). Last year it's 90 of 250 (35%)

This is against overall Assisted rates for Deng (courtesy 82 games) of 63% (this year) and 70% last year.

This doesn't exactly answer your question because it's not limited to lineup numbers (the figures for this year is especially rubbery because of Rose and Deng's extended absences. Were they contiguous?). But from the raw numbers, it doesn't look like Deng is offensively joined to the hip of Rose. Having said that, assist numbers aren't necessarily the be-all end-all of joint offensive impact.

Of interest is Engelmann's player pairs (although he doesn't appear to have compiled anything for '12 yet).

Quick summary, Deng and Rose have the highest combined +/- for a pair for the Bulls at 1.4. The next highest (intriguingly) is Korver and Noah (in far more limited minutes).

Rose's next highest pairing is with Taj Gibson at +1, wheras Deng's is with Boozer at +1.1

So once again it doesn't look like Deng's production is necessarily slaved to Rose's.

Not conclusive of course, but interesting.

[coldfish]

One thing that has been discussed on the Bulls board is that Rose is "cruising" a lot more this year. That would be really hard to quantify, but you can see it in his usage number, which is down from last year. Basically, Rose has another gear he can regularly turn on when the team needs him whereas Deng doesn't seem to. I know, that's really subjective but its an impression we have got.

Does it matter though? If Rose is truly "cruising", we'll see his impact jump come playoff time, right?

And for now it's all about what guys are actually doing, not what they're capable of doing.

I was talking more in games. Rose will cruise in the first half a lot and just let other players go to work. He only gets aggressive when he has a reason, like the Bulls being down. That leaves fans with the somewhat subjective feeling after many games that "Rose won the game". That might be part of the reason why the subjective analysis doesn't match the statistical analysis.

[Rapcity_11]

One thing that has been discussed on the Bulls board is that Rose is "cruising" a lot more this year. That would be really hard to quantify, but you can see it in his usage number, which is down from last year. Basically, Rose has another gear he can regularly turn on when the team needs him whereas Deng doesn't seem to. I know, that's really subjective but its an impression we have got.

Does it matter though? If Rose is truly "cruising", we'll see his impact jump come playoff time, right?

And for now it's all about what guys are actually doing, not what they're capable of doing.

I was talking more in games. Rose will cruise in the first half a lot and just let other players go to work. He only gets aggressive when he has a reason, like the Bulls being down. That leaves fans with the somewhat subjective feeling after many games that "Rose won the game". That might be part of the reason why the subjective analysis doesn't match the statistical analysis.

Yeah, I get what you mean. However, why would it be a surprise that Deng is having a higher per possession impact than Rose if the above were true?

[coldfish]

Yeah, I get what you mean. However, why would it be a surprise that Deng is having a higher per possession impact than Rose if the above were true?

Its not a surprise, its an explanation for why that is the case.

[drza]

Yeah, I get what you mean. However, why would it be a surprise that Deng is having a higher per possession impact than Rose if the above were true?

Its not a surprise, its an explanation for why that is the case.

I think this gets to a crucial point, in the difference between "MVP" and "best player". Rose could in theory be the best player on the team in abstract, but Deng the most valuable. All stats really, but APM especially, can't compensate for what someone is capable of doing, only what they actually did.

[Rerisen]

All stats really, but APM especially, can't compensate for what someone is capable of doing, only what they actually did.

We have to be careful here though, because all the points the Bulls scored and all the points the opponent scored are not really what 'Deng did'.

The numbers are saying Deng is more impactful essentially on the basis of that the team does better when he's on the floor without Rose than vice versa. But probably 75% or so, the vast bulk, of their minutes on the floor are together and both are getting the same credit during that time. The time that is the most crucial periods of a game, and against the top competition of the opponent. If Rose assists Carlos Boozer, then Rose and Deng both get +2. But whether they are both equally responsible for the team's production in that time is a harder nut to crack, and to assume from their much lesser time apart.

Since Rose produces more hard production than Deng (PER, Win Shares, etc), this is where you have to get into how much more value you might want to assign him during their on floor time together.

[Rerisen]

One thing I think will be interesting to follow is if Deng beats Rose this year again, but on non-prior informed. Because of course with last year weighing in heavily he is likely to have a big advantage on the prior informed for the whole season.

But purely on this year's numbers alone, if Rose comes out ahead, then it may suggest last year's numbers could be in question more - at least for someone whose more subjective, or we might say, basketball analysis, is that Rose should be more valuable. Right now Rose is creeping up on Deng (2.4 to 2.8) without any prior weighting. Taj still leading them both... does RAPM know that Carlos Boozer isn't nearly as good as he was in 2010, and is it still giving him weighted credit for 2010? I would think last year it was overvaluing his actual play, and those who played more with him were likely to suffer.

The bizarre thing to me is that RAPM likes Boozer more in 2011 than 2010. When his level of production and certainly watching him would scream the opposite. That Boozer was a gimpy broken down mess for most of 2011. My impression would be that going to a much better supporting cast in Chicago, that ended up winning 62, gave him something of a rising tide lifts all boats effect. No matter that these things are supposed to be parsed out if you have ultimate faith in it.

[mysticbb]

If Rose assists Carlos Boozer, then Rose and Deng both get +2.

That is not true. Rose is seen as the more impactful offensive player, thus, he got so far more credit for the points scored by the Bulls than Deng.
Deng makes his impact on the defensive end, the Bulls defense is better with Deng to a greater degree. And that's what we also see when we compare games with and without Deng.

And I see that you are trying to interpret the numbers in a way that it fits your preconception. Last season the non-informed RAPM for Rose is 3.4 (with +2.4 on offense), while Deng had 3.8 (+3.2 on defense). So, the difference in this season is basically the same as last season. Boozer was +1.4 on offense and -0.8 on defense. The prior for those two were +1.8 on offense and -2.0 on defense. So, no, I would not say that Boozer's prior had a big impact on the result from last season.

[Rerisen]

Gotcha. But tell me if this is correct. RAPM is suggesting if the Bulls had the chance they should not have traded in their injured for 75% of the season Carlos Boozer, for the much healthier and more productive Carlos Boozer of 2010?

Now I can understand that Boozer may have for whatever reason, say the Bulls better defense, been more valuable to Chicago than he was to Utah in 2010. But he is still the same player style wise, and therefore even on top of being a better fit, I would take the year younger, healthier Boozer from 2010 in an instant, who I think would have put up an even better number for the Bulls in 2011.

[ahonui06]

Obviously, the MVP of the Bulls is Derrick Rose.

How good would Chicago be in the playoffs without Rose?
How good would Chicago be in the playoffs without Deng?

I rest my case.

Chicago needs Rose to be successful because he is pretty much the foundation of their offense.

[mysticbb]

Gotcha. But tell me if this is correct. RAPM is suggesting if the Bulls had the chance they should not have traded in their injured for 75% of the season Carlos Boozer, for the much healthier and more productive Carlos Boozer of 2010?

Offensively Boozer was better in 2010, that's what the numbers are telling us. But Boozer was never someone who was able to anchor a defense, he needed someone next to him to help. That is seen in the defensive numbers. Boozer on the Bulls plays in a defensive system which can cover his poor defense better (I think that is something many Bulls fans expected before 2010/11, if I'm remembering that correctly, at least I assumed that the defensive system by Thibodeau will help Boozer). So, in the end Boozer might have eaten into the defensive value of Deng much more here than into the offensive value of Rose.

Now I can understand that Boozer may have for whatever reason, say the Bulls better defense, been more valuable to Chicago than he was to Utah in 2010. But he is still the same player style wise, and therefore even on top of being a better fit, I would take the year younger, healthier Boozer from 2010 in an instant, who I think would have put up an even better number for the Bulls in 2011.

I agree with that, but I think that would have had mainly an effect on offense; the Bulls would have been a better offensive team.

@ahonui06

The Bulls so far did not show that they are playing better with Rose and without Deng than vice versa. ;)

[Rerisen]

Obviously, the MVP of the Bulls is Derrick Rose.

How good would Chicago be in the playoffs without Rose?
How good would Chicago be in the playoffs without Deng?

I rest my case.

Chicago needs Rose to be successful because he is pretty much the foundation of their offense.

Well the Bulls did blow out the Pacers by 20 points with Rose going 5-16 for 13 points. Deng was +33, Rose +14. So score one for Luol.

Now if you watched the game, however, Rose's 3rd quarter barrage of threes certainly jumped out as something that fired the Bulls momentum and seemed to break Indy's back. Is there such a thing as momentum or is that all perception (rhetorical!).

[mopper8]

Obviously, the MVP of the Bulls is Derrick Rose.

How good would Chicago be in the playoffs without Rose?
How good would Chicago be in the playoffs without Deng?

I rest my case.

http://en.wikipedia.org/wiki/Begging_the_question

[RichardsRival3]

MVP Deng is leading the Bulls offense currently.

Bulls can't do jack against the 6ers and lost last night to the blazers.

[ahonui06]

Only advanced stats clowns would think Bulls would be more successful with Deng than Rose. It's the only reason I can think of that someone would even suggest Deng is more important than Rose to Chicago.

[Krodis]

The Bulls would hardly be the first team to struggle offensively against the Sixers.

Besides, I don't think anyone ever claimed Deng was some sort of OFFENSIVE guru.