tsherkin wrote:It should be pointed out that the Bulls clearly win on the basis of their defense, to which Deng contributes more than does Rose. That's probably a large part of why the numbers indicate Deng's efficacy the way they do.
Last year, the Bulls were 11th in team ORTG, which is unremarkable; Rose wasn't driving an elite offense, though given the teammates he had and that he isn't Steve Nash, I don't think anyone is surprised. They were at 108.3 ORTG (though that's with Boozer and Noah injured and Boozer's jumper not there at all last year, compared to this season). This year, they're at 108.2, but because league offense sucks this season as a result of the compressed schedule, they're actually 3rd in the league in team ORTG. So it's tangibly identical in terms of their offensive production, and they're still winning on the basis of their defense, but by default, their offense is much more potent because few teams are scoring as effectively as they did in the last non-lockout season.
The flip side of this is that the Bulls really suck on offense without Rose. If they didn't have him shouldering the load, they'd be worse offensively and it would impact the team a lot over a lengthy period of time.
Which is more important?
Over the balance of the season, I bet we'd find a net slight weight in favor of offense. You can't win if you can't score and when you take Rose off of this team, they cannot drive offense against strong, coordinated defenses because basically no one else can dribble or create in HCO. That needs to be considered; he's their best playmaker, their best at generating shots against set defense and their top volume scorer. He's not mega-hyper efficient, so his overall impact isn't the same as someone like James or Durant or whomever and he isn't the same caliber playmaker of a player like Paul or Nash, but Rose's importance to Chicago's offense cannot be understated because of the quality, or lack thereof, inherent to Chicago's offensive lineup.
They ALL rely on other people to set them up and of the other players, only Rip really does anything to generate shots for himself. Deng has been straight-up terrible on offense this season, as have Brewer and Richard Hamilton. It's really only Boozer and Noah who are scoring competently, and we know well enough where their offense is coming from: Rose and offensive rebounds.
So we consider the record without Rose. The Bulls are 27-8 overall, 20-5 with Rose. With Derrick, they're on pace for almost 66 wins in a regular season (.800). Without him, they're on pace for about 57 (0.700), and that's after facing the Wizards, Grizzlies, Suns, Cavs, Bobcats twice, Celtics twice, Kings and Nets. Not exactly the most stirring competition, and noticeably less successful than with Derrick in the lineup.
I'd say the numbers don't necessarily line up and we're seeing an anomaly here that may be rooted in the fact that the team primarily dominates with defense, which is an area where Deng has a lot of impact in pushing the team into elite territory, whereas Rose drives an otherwise unremarkable offense despite a fair number of underperforming slackers, so he's not getting quite as much statistical credit for moving the team towards victory.
This post is actually a perfect example for what I like about the development of the APM family of stats. Because the first half of the post (up until "Which is more important?") is an example of a knowledgable basketball mind visually scouting a situation and making informed observations. An "eye test", if you will, of someone who's basketball opinion I value. Then, the second half of the post (after "Which is more important") is the informed speculation of that person, based on what they observed. Essentially, making interpretations based off the eye test. Before the APM family of stats, interpretation of the informed observer was all that we had access to outside of the box scores. The problem is, even among good basketball minds, everyone's analysis will be different. There's no way to do it objectively. Whereas with APM, we don't have to rely on changes in subjective point of view. If the rules of statistics are followed, APM gives us an objective way to answer the "which is more important?" question without relying purely on speculation.
But here's where it gets interesting to me: outside of the 4 words underlined above, APM and our intelligent observer say THE EXACT SAME THING. The only difference, then, is in the interpretation...not the observation.
Let's return to the specifics of tsherkin's post. tsherkin says of Rose's offense "He's not mega-hyper efficient, so his overall impact isn't the same as someone like James or Durant or whomever and he isn't the same caliber playmaker of a player like Paul or Nash". Rose's 4-year APM and 2011 RAPM both have him about +2.0 on offense, a very good mark (~top 30 in the league) but not the super-elite compared to James (+ 4.1 offensive RAPM in 2011), Durant (+3.9), Nash (+6.2), and Paul (+4.1). So (and tsherkin, let me know if I overstep myself in making conclusions for you based on your writing) it'd seem that our informed observer and our informed impact stat are seeing the exact same thing.
The difference comes once tsherkin takes in his observations, and says "I bet this is what would happen" whereas the APM says "this is what actually did happen". And what did happen is that Deng has measured out as one of the better impact players in the game before Thibs or Rose even got there, he's held that impact for several years, and he held that impact all of last season with both Thibs and Rose on board as well. Yes, if someone were to try to base their analysis purely off of the 10 or 15 games that Rose missed this year you could get some crazy results because the sample size is just too small. But with years and years of consistent results in the bank, for Deng to then maintain his impact in the 15 games without Rose and maintain the exact same level of impact that the multi-year studies would tell us to expect...to me, that's a much stronger result.
It's just like Melo with Denver/NY last year...we had years of data pre-trade telling us that Melo wasn't the huge impact he was made out to be and that thus his going from Denver to NY shouldn't make a huge difference. So for it to play out exactly like that once the trade was made in-season, to me someone can't claim that the Nuggets getting better while the Knicks not post-trade is a sample-size anomoly. No, we had plenty of data with which to make our prediction, and it's okay then that the test period be a bit shorter since the results fit the prediction exactly.
Anyway, the point I'm making isn't that APM is a be-all/end-all. But I like having a quantitative approach that appreciates/values the complete package/team impact things that a trained observer would notice while remaining completely objective as to the relative importance of the different players/team roles/skill sets.
I've looked at the APM and RAPM studies and generally been pleased with what they indicate, given enough of a sample. 
