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

[drza]

The fact that Deng has the consistently super-high APM going all the way back to the '04 study that captures his rookie year, in different multi-year studies,

Please list more on that. The data I have previous gathered makes me hesitant to go too far with that.

Ilardi's 6-year from '03-'09 ranked him at 48th.
Engelmann's 5-year study he had on his side mid-season last year had Deng at 27th.

I do realize that more recent studies see climbing significantly even with many year studies, but the truth is that until recently, even the multi-year studies I've seen only saw Deng as a very good player, not what I'd call super-high.

Of course, more recent events SHOULD be able to sway rankings, but it seems really clear at the very least that we should not be looking like Deng as a guy who has been a secret superstar forever. In fact, he wasn't even a guy championed by +/- stat folks really until recently.

Your results match the ones I quoted earlier in the thread, so I guess "super high" is the term under question. I guess I could have just used "high" there. But keep in mind, a ranking of 47th would have him as either the best or 2nd best player on average on most teams. And if you look at the actual list, there are a bunch of part-time players on it which realistically puts Deng around 35th or right-near-best-on-team level in the earliest study. Which would correspond with his 27th ranking in the second multi-year study.

Bear in mind, I'm not arguing that Luol Deng is Steve Nash in terms of impact. But what I am arguing is that he's very arguably the best player on a team with lots of very good talent. I think the data would support that assertion, with the caveat I mentioned in the last post that perhaps the system hasn't completely caught up to Rose yet.

[Doctor MJ]

So a couple things:

I think the argument that Rose is disproportionately critical against tough defenses is a hypothesis with some merit, and I also think that with all of this we need to be cautious with low sample size.

Drilling down into more details. It has been mentioned that the Bulls offense is currently quite highly compared to the rest of the league right now (3rd in the league), but that comes not from improvement, but the rest of the league falling off. Also that it comes not from great scoring, so much as it comes from rebounding.

Now, that's absolute stuff which doesn't necessarily have bearing on the relative lift Rose is providing. It is worth considering though where we think Rose is having impact on the team:

Effective FG%
Free Throw generating
Turnover reduction
Rebounding

Rebounding is kind of in a category on its own. Rose obviously isn't the direct reason the team is big on rebounding.

Evan Z has recently done some great stuff with the 4 factors, here's his spreadsheet analyzing the 3-non-rebounding factors for point guards which I believe factors in the past 2.5 years:

https://docs.google.com/spreadsheet/ccc ... IM0E#gid=0

Obviously that's not capturing Rose right at this moment, but I still think it's interesting. So:

The #1 eFG guy in the study is Steve Nash at +1.21. #2 is a tie interestingly enough between Miller and Curry at +0.87. Rose ranks 17th at + 0.15.

The #1 turnover guy is Chris Paul at -1.22. #2 is Mike Conley at -0.61, literally only half of Paul. Rose ranks 9th at -0.41.

Both of those were kind of obvious #1s, I was curious who would lead in FTs...

The #1 free throw guy is Deron Williams at +1.72. #2 again is Andre Miller at +1.44. Rose ranks 16th at +0.24.

Okay so none of this necessarily means anything specific and I understand that. Part of what it makes me think about though is whether Rose is really doing something THAT unusual. Certainly he's doing something that no other Bull can do, and just as certain he's a fantastic athlete which is helpful to him. However if he's not helping any of these things specifically with huge amounts it seems to point to him having a role where much of his benefit comes simply being on a team designed to have strengths other than Rose's strengths (i.e., good balance).

And all this comes back to the idea, imho, that it's really dangerous to look at Rose's role as inherently more important to team basketball than the other roles. All roles come down to replaceability, and the more glamorous the role, and the less it requires extreme height, the more "surprisingly" replaceable we should expect it to be.

If a player like that then appears to fall into an ideal situation for him having huge impact, and he STILL isn't the clear +/- leader for his team, isn't this kind of amazing?

[Rerisen]

Quality of competition, or lack thereof, in his absence has been favorable to the Bulls.

A 10-game sample isn't exactly wonderful for evaluating everything that needs be evaluated in this context. He missed only one game last season, so that's 11 total games without him as a sample, and a weak one at that.

We actually don't have any idea how the Bulls would hold up without Rose over an extended period of time spent facing competent competition.

Right. Nor as I posted in the Bulls thread, do we have any idea how the Bulls would finish games (find offense) in the clutch against good competition without Rose. They mostly steam rolled bad teams this year without him, allowing them to avoid this issue. And even before Rose was drafted, then Ben Gordon was the Bulls closer, and his offense was disproportionately relied on late in games, relative to the first 3 quarters.

Deng has been a player who struggles to find his offense in 4th quarters traditionally, being such a system offensive player. Likewise, as most players numbers fade in the clutch, you need superstars to create offense then more than ever. And it's entirely possible the Bulls could stay in a lot of games without Rose up until the last 5 minutes, but ultimately end up losing more of those than RAPM thinks they should, due to the inability of anyone to produce or playmake offense when defensive intensity picks up and pressure intensifies. The reduction of efficiency across the board in clutch minutes is just too pervasive to pretend that the team's performance is just going to stay the same in these situations.

Something like this could only be measured accurately if the Bulls were able to go without Rose for an immense length of time, such that the sample minutes of these specific situations made up a proper % of the whole non-Rose minutes. Right now they account for virtually nil of our Bulls non-Rose minutes, as Deng without Rose plays almost always against second units, and then this year, against mostly bad teams that didn't produce close games.

If there was no such thing as a superstar factor late in games, then teams like Denver and Philly would be held in much higher esteem. Likewise, even the Bulls to Miami, as the Bulls underlying metrics are just as impressive as the Heat, yet everyone has Miami as the favorites this year, precisely because of the belief in superstars to take it to the next level, and the relative lack of ability of the Bulls to produce offense if a team can bottle up Rose. The problem would be 10x fold as bad without him. That it doesn't show out via a ton of 5 minute increments per game when Deng is playing against backups without Rose, isn't necessarily a safe thing to be relying on to extrapolate from I don't think.

[Doctor MJ]

Right. Nor as I posted in the Bulls thread, do we have any idea how the Bulls would finish games (find offense) in the clutch against good competition without Rose. They mostly steam rolled bad teams this year without him, allowing them to avoid this issue. And even before Rose was drafted, then Ben Gordon was the Bulls closer, and his offense was disproportionately relied on late in games, relative to the first 3 quarters.

It made me look up 82games' clutch status to get a better sense of how well Rose is playing that role.

What did I find: That the Bulls' players have basically spent NO time in clutch situations this year. It's kind of amazing. For example, they have Rose listed playing less than 20 minutes in clutch time all year. This is why 82games league-wide clutch page lists players from every team EXCEPT the Bulls. None have played enough to qualify.

So in this sense, truly, the Bulls really haven't proven they can thrive in the clutch without Rose, but only because thriving in the clutch has proven of so little importance for the team. Odd.

[Rerisen]

I noticed that too the other day Doc, however if you look at last year, the Bulls reliance on Rose in the clutch was pretty staggering. His usage was off the charts. 2nd in pts/48 of clutch time to Kobe, and also generated 9.8 assists as well. So it's a big black hole here how things would go without him. But just looking at the limitations in ball handling and shot creation among the rest of the team, it doesn't look very inspiring.

[Doctor MJ]

I noticed that too the other day Doc, however if you look at last year, the Bulls reliance on Rose in the clutch was pretty staggering. His usage was off the charts. 2nd in pts/48 of clutch time to Kobe, and also generated 9.8 assists as well. So it's a big black hole here how things would go without him. But just looking at the limitations in ball handling and shot creation among the rest of the team, it doesn't look very inspiring.

All true.

[JustCame]

Haven't read anything in this topic, but my first thought after reading the title is that advanced are great and all, but they're not everything. This is taking it way too far. lol

[SideshowBob]

Haven't read anything in this topic, but my first thought after reading the title is that advanced are great and all, but they're not everything. This is taking it way too far. lol

I think you should read it in that case

[rrravenred]

Interestingly, looking at last season's clutch numbers, the per48 numbers for +/- (yeah, I know, small sample size) has the two at +21 (Rose) and +15 (Deng). Both, however, are dwarfed by Kyle Korver, who was +29 per 48 (+89 gross) in comparable overall minutes.

Touching on what Ronny was mentioning, if we're talking about performance in role, then Korver's role in late-game situations was very effective, be that in spacing the floor or knocking down the 3 (often, but not always, on the assist from Rose) and basically not doing much else wrong (yes, it's entirely true that he's not asked to do much else).

Why? Well I suspect it's partially due to Rose leading the league in TO per 48 in the clutch due to somewhat suspect decision-making (he's a young player so that's something he can learn). To put this in context, the much-maligned Wade and James clutch duo totalled only 6.8 combined per 48.

Once again, whilst Deng's clutch statistics are extremely ordinary, his (raw, admittedly) +/- production seems to outstrip it.

So whilst it may be a black-hole without Rose, it's still not an exploding supernova WITH him. To repeat myself, however, SMALL SAMPLE SIZE!

[RichardsRival3]

I noticed that too the other day Doc, however if you look at last year, the Bulls reliance on Rose in the clutch was pretty staggering. His usage was off the charts. 2nd in pts/48 of clutch time to Kobe, and also generated 9.8 assists as well. So it's a big black hole here how things would go without him. But just looking at the limitations in ball handling and shot creation among the rest of the team, it doesn't look very inspiring.

Basically what I was saying. The tougher the D, the more the Bulls rely on Rose.

Against poor defensive teams, Rose isn't as needed.

Deng can't do **** against good defensive teams. He can't create his own shot and can't create for others.

[Doctor MJ]

Deng can't do **** against good defensive teams. He can't create his own shot and can't create for others.

This is just asserted by people, but what's the evidence for this? Not saying it doesn't exist, but in 20 games against the 4 best non-Bull defenses last year, Deng scored about 1 point less and had just about as many assists as he did in a typical game. Obviously there's more to it than that, but if this were a guy totally crippled by good defense, shouldn't we see something obvious here?

[tclg]

I think people are seeing what they want to see. Deng is about as consistent as it comes to his scoring. Its not his fault he isnt the guy the team turns to to produce buckets but he does not choke

[Rerisen]

Interestingly, looking at last season's clutch numbers, the per48 numbers for +/- (yeah, I know, small sample size) has the two at +21 (Rose) and +15 (Deng). Both, however, are dwarfed by Kyle Korver, who was +29 per 48 (+89 gross) in comparable overall minutes.

Touching on what Ronny was mentioning, if we're talking about performance in role, then Korver's role in late-game situations was very effective, be that in spacing the floor or knocking down the 3 (often, but not always, on the assist from Rose) and basically not doing much else wrong (yes, it's entirely true that he's not asked to do much else).

Korver is not much of a mystery and you touched on reasons for his success. The Bulls started games with the inept offense of Keith Bogans in order to create a defensive mindset for the game. Bogans shot ok at threes, but did nothing else well on offense and this was exploited by defenses, making Rose's life harder, and bringing down the raw +/- of every starter. Since Keith played probably 99% of his minutes with Rose (he started both halves and always subbed out first, rarely ever returning), were the corrections of APM or RAPM able to properly mete out his deficiencies? I'm suspicious that they could.

Instead they probably globbed on in some measure to all the Bulls starters he played with and further increased the split between the first unit and the Bulls stellar bench mob. In fact, when you factor Boozer's struggles past the first month and a half, due to injuries (his TS% was horrid for about his last 65 games, after a hot start), it's easy to argue that the Bulls were holding 2 superior players on their bench (Ronnie Brewer over Bogans, and Taj Gibson over Boozer). Considering Boozer had been such a strong player before last season, prior informed numbers on him probably also overrate his impact last year, and rubbing off on Rose who played with him so much.

In any case, Brewer wasn't much better for the offense. Korver however, meshed great with Rose. Rose is a drive and kick passer, Korver is a deadly open shooter. Unfortunately, Kyle had trouble staying on the floor vs certain matchups, esp Miami.

But in the regular season it worked great. Korver led the league in 4th quarter three pointers, and the Bulls were the best 4th quarter margin team in the league. Off almost entirely going to Rose in that quarter and letting Korver help space the floor.

[Rerisen]

This is just asserted by people, but what's the evidence for this? Not saying it doesn't exist, but in 20 games against the 4 best non-Bull defenses last year, Deng scored about 1 point less and had just about as many assists as he did in a typical game. Obviously there's more to it than that, but if this were a guy totally crippled by good defense, shouldn't we see something obvious here?

Rose was in those games though, allowing Deng to play his normal, off ball 3rd option role. The idea is that without Rose, Deng would struggle greatly offensively, because he would be asked to create his own shot far more, as he and Boozer would be forced to do that. And they are very bad at it.

We can't model a true test of that just by extrapolating on a stack of instances, maybe 5 minutes a game, where Deng played with the Bulls backups against other backups without Rose in the game. Yet those minutes make up almost the entirety of our Deng without Rose sample.

Except for also the 10 game stretch this year, but as mentioned the combined % of the teams they played were quite bad, the defenses were quite bad, and the team was only tested in a couple of those games down to the wire.

To add on to my last post above a bit too, Deng, because he played 2 more minutes than Derrick last year, and is also the starter that Thibs sent back in to help the 2nd unit the most, I think truly benefited from getting any minutes away from Boozer and Bogans, who because of circumstances and stat lag, may be overrated in what they contributed, to the detriment of Derrick.

The raw On Court numbers of Bogans (+6.4 per 100 poss) vs Brewer (+11.9), and Boozer (+7.0) vs Gibson (+9.3), and even Noah (+5.8) vs Asik (+15.2 <-- note very limited mpg) support that these guys seemed to be a drag on Rose (+9.0) and Deng (+9.1) but that Deng was able to benefit more from the most time with the Bulls superior bench unit.

Interestingly the Bulls best unit with over 100 minutes last year was: Rose, Brewer, Deng, Gibson, Asik.

Further adding fuel to Bogans and Boozer as drags on the team and perhaps on Rose statistically.

The Bulls best unit in the playoffs? The exact same unit!

It's no wonder Thibodeau abandoned Boozer and Noah late in the ECF to go to Gibson and Asik. And of course to cut Bogans minutes way down, who averaged 4.6 pts vs Miami.

What do Derrick's (and the teams) +/- numbers look like last year if say Ronnie Brewer starts instead of Bogans. Or Korver and Brewer alone split the time there with no Bogans? I wonder.

[mysticbb]

We can't model a true test of that just by extrapolating on a stack of instances, maybe 5 minutes a game, where Deng played with the Bulls backups against other backups without Rose in the game. Yet those minutes make up almost the entirety of our Deng without Rose sample.

Over the last 1.5 years we have a sample of Deng playing 1065 minutes without Rose (regular season and playoffs). During those minutes these are Deng's per 36 minutes numbers:

16.8 points, 6.0 rebounds, 3 assists while scoring on 53 TS%. The Bulls were +3.6 during those minutes.

In the 3943 minutes with Rose:

15.1 points, 5.6 rebounds, 2.5 assists while scoring on 54 TS%. The Bulls were +7.2 during those minutes

So, Deng has a higher volume (as expected) and a little smaller efficiency.

We can make the same thing for Rose. Per 36 minutes with Deng in 3943 minutes:

23.9 points, 3.9 rebounds, 7.4 assists while scoring on 55 TS%. The Bulls were +7.2 per 36 during those minutes.

Without Deng in 655 minutes:

22.9 points, 3.3 rebounds, 7.5 assists while scoring on 53 TS%. The Bulls were +1.0 per 36 during those minutes.


I think Deng gets a little boost by Rose giving him a chance to play within a more suited role, but it also seems like Rose is benefitting from the presence of Deng on the court. Overall the Bulls with Deng and without Rose played better than the Bulls with Rose and without Deng.

And it is no wonder that the Bulls are playing worse without Rose. It is not like Rose has no impact whatsoever, he has in fact a pretty high impact. It is just that we can expect the Bulls to play even worse without Deng instead.
For his season we have a 7 game sample without Deng and with Rose. The Bulls played like a 1.3 SRS team. With Deng and without Rose in 10 games it was 3.2 SRS. In 10 games with both players, the Bulls played like a 10.6 SRS team. As I said, we see a massive drop without Rose, just a even bigger drop without Deng. And that is consistent with all +/- based data we have.

[Rerisen]

Over the last 1.5 years we have a sample of Deng playing 1065 minutes without Rose (regular season and playoffs). During those minutes these are Deng's per 36 minutes numbers:

16.8 points, 6.0 rebounds, 3 assists while scoring on 53 TS%. The Bulls were +3.6 during those minutes.

In the 3943 minutes with Rose:

15.1 points, 5.6 rebounds, 2.5 assists while scoring on 54 TS%. The Bulls were +7.2 during those minutes

So, Deng has a higher volume (as expected) and a little smaller efficiency.

We can make the same thing for Rose. Per 36 minutes with Deng in 3943 minutes:

23.9 points, 3.9 rebounds, 7.4 assists while scoring on 55 TS%. The Bulls were +7.2 per 36 during those minutes.

Without Deng in 655 minutes:

22.9 points, 3.3 rebounds, 7.5 assists while scoring on 53 TS%. The Bulls were +1.0 per 36 during those minutes.


I think Deng gets a little boost by Rose giving him a chance to play within a more suited role, but it also seems like Rose is benefitting from the presence of Deng on the court. Overall the Bulls with Deng and without Rose played better than the Bulls with Rose and without Deng.

And it is no wonder that the Bulls are playing worse without Rose. It is not like Rose has no impact whatsoever, he has in fact a pretty high impact. It is just that we can expect the Bulls to play even worse without Deng instead.

You are making a leap at the end there. I disagree not with what happened, but how useful it is for a specific scenario we are trying to envision: the team without either Rose OR Deng.

Again, how many of those minutes without Rose are against 2nd units? I noted specifically that Deng has been Thibs guy to leave in with the 2nd unit against other 2nd units. A Bulls 2nd unit that regularly outscored their opponents at greater advantages than Bulls starters did their counterparts. Of course Deng saw a helping affect from this. Rose was the player more tied to Keith Bogans all of last year, and +/- also tells us that Keith Bogans was mostly crap, relative to the rest of the team.

Likewise, how many of those minutes have been clutch minutes, when it gets harder to find system offense, and individual creation ability is maximized.

Luol Deng is a rather unique player, in how highly assisted he is for a fairly high volume guy. This year it's a little down, but career wise usually around 70%.

The specific contention is not whether Deng and the team can play alright for 5 minutes a game (over and over and over) without Rose, but how he would do over a long haul and especially vs top competition or in the clutch. That sample has not happened even this year when Rose has been out to any valuable degree. The competition was poor.

Now we can say that if the Bulls didn't have Rose they'd probably find some other volume guy (they had Ben Gordon before) which would allow Deng to not have to take on scoring burdens he would not be good at, but I'm not sure that is the same thing as saying Deng is more valuable than Rose if for instance the team had to lose one of them right now, for the rest of the season.

For his season we have a 7 game sample without Deng and with Rose. The Bulls played like a 1.3 SRS team.

That is a poor sample size though. CJ Watson was also playing out of his head to start the year and most of the time Rose was out. I think his PER was up around 18-20 until finally the burden on him seemed to start taking a toll and now he's been plummeting back to norms.

Play a full season without Rose and Watson is not going to keep that level up. Just one role player getting hot over a short span like this can highly skew things.

There have been examples of good teams losing their top guy for a handful of games in a season and doing well without him. I'm not going to say based on those that those teams would not have been far worse losing their top guy over a full season comparatively. Which I think is certainly the case this year with the Bulls without Rose. I just see them collapsing terribly late in close games, when no one can produce a good shot.

Our numbers have no answers on this question because it hasn't happened enough, and just instead pretend that we can use random minute stretches drawn from between 1st through 3 and a half quarters of play, and that those stretches will mimic what happens at the end of games without Rose. I don't buy it. Player percentages change in those spans, defenses intensify, how teams plays offensively changes (they go to their superstars, etc).

Another factor unappreciated is that Derrick Rose doesn't go all out in easy wins. He defers to teammates. Once the Bulls are up 15 or 20 on a bad team with Derrick Rose, is it not realistic to think the team might let up a little bit, that Derrick Rose lets up himself is quite obvious, and that if they didn't, they could perhaps win some of these games by 30 or 40?

Comparatively without Rose, the Bulls do not have the luxury of letting up, and go all out (at least in a short span without him) and are able to achieve similar win margins against bad teams even without him there.

I think the ability of superstars to put it into another gear is being underappreciated if we put faith so religiously in these +/- numbers.

The above sentence is almost entirely why the Miami Heat, probably with good reason, remain favorites over the Bulls this year. Not because they look stronger on paper. (Bulls have virtually the same win margin, despite far greater injuries).

Last year the Bulls outscored the Heat by 10 points in the ECF through 3 quarters of all the games. But Miami outscored the Bulls by 22 points in the 4th quarters and OT of that series. This isn't I think by chance. But by what was an obvious weakness of the Bulls reliance on one player offensively, that was predictable and predicted the whole year. A problem that would be greatly magnified without Derrick Rose there over a long period.

Predicting a repeat matchup between Miami and the Bulls while not having that 4th quarter information from the ECF, would leave us without a key piece of evidence we need to understand the dynamics in the matchup. The same type of key evidence that is missing in trying to extrapolate a Bulls team without Deng relative to without Rose.

Are you telling me that if we play a playoff series between two Bulls teams, one is starting CJ Watson in place of Rose, and the other is starting Ronnie Brewer in the place of Derrick Rose, you believe the Deng led team would win?

[mysticbb]

Again, how many of those minutes without Rose are against 2nd units? I noted specifically that Deng has been Thibs guy to leave in with the 2nd unit against other 2nd units. A Bulls 2nd unit that regularly outscored their opponents at greater advantages than Bulls starters did their counterparts. Of course Deng saw a helping affect from this.

I don't think that this is actually a big issue here. We have data for 1.5 years and RAPM can handle such effects petty well. If Deng would have benefitted most from playing with superior players together against clearly inferior players, his value would have dipped while the value of the non-2nd-unit teammates would have been increased.

Rose was the player more tied to Keith Bogans all of last year, and +/- also tells us that Keith Bogans was mostly crap, relative to the rest of the team.

Deng also played a big part of his minutes with Bogans. In fact Deng played 1719 with Bogans while Rose played 1708 minutes. Overall Deng played 44.1% of his minutes with Bogans, Rose played 46.5%. We can adjust that by taking out the additional minutes relative to the total minutes while assuming Bogans had a -15.3 per 36 minutes effect on Rose. That makes +34 for Rose or +0.3 per 36 minutes for Rose, if we compensate for the additional minutes with Bogans in comparison to Deng. That means we are getting +1.3 with Rose and without Deng, while still being at +3.6 with Deng and without Rose. No, Keith Bogans does NOT explain the difference in +/-.

The specific contention is not whether Deng and the team can play alright for 5 minutes a game (over and over and over) without Rose, but how he would do over a long haul and especially vs top competition or in the clutch.

We expect the Bulls to do worse without Rose and with Deng than they are doing with both. That should be pretty clear. But the Bulls without Rose are not going down to be a lottery team. We saw that in this season, they are still able to beat below average teams, which in the end will make them above average. And the Bulls would probably come closer to the mean in clutch situations. That is to expected.

That is a poor sample size though.

Oh, if that would be the only value used for that or it would be in complete contradiction to other values, you would have a point. But the fact is that the difference is pretty much at a level we would expect given the WHOLE dataset. RAPM has Deng with +5.8, Rose with +3, that makes about +2 for a game (36 minutes played). The raw +/- has the Bulls with Deng and without Rose better by about 2 points per 36 minutes than with Rose and without Deng. And during a 7 game sample without Deng and with Rose the Bulls play 2 points worse than with Deng and without Rose. How can you say that this is just a result of sample size, if everything else is in agreement with that?

CJ Watson was also playing out of his head to start the year and most of the time Rose was out. I think his PER was up around 18-20 until finally the burden on him seemed to start taking a toll and now he's been plummeting back to norms.

Watson as a starter has per 36 minutes:

15.9 points, 2.9 rebounds, 5.7 assists with 53 TS%

as a sub:

17.0 points, 3.4 rebounds, 7.0 assists with 51 TS%.

You honestly want to say that this was just a result of Watson getting hot as a starter?

Watson is inconsistent offensively, that makes him better suited as a backup, but he did not just got hot during the games he started instead of Rose, that is not something backed up by the facts.

I just see them collapsing terribly late in close games, when no one can produce a good shot.

The thing is that this will likely result into the Bulls being closer to the mean, which will just mean that they will not win more close games than they lose. But that still makes them a better than average team without Rose, they still are good enough to win 50 in a 82 games seasons. That does not make them contenders without Rose, but nobody is saying that.
Without Deng the Bulls would just win a couple of games less, even though they might be better at winning close games.

Another factor unappreciated is that Derrick Rose doesn't go all out in easy wins. He defers to teammates. Once the Bulls are up 15 or 20 on a bad team with Derrick Rose, is it not realistic to think the team might let up a little bit, that Derrick Rose lets up himself is quite obvious, and that if they didn't, they could perhaps win some of these games by 30 or 40?

Comparatively without Rose, the Bulls do not have the luxury of letting up, and go all out (at least in a short span without him) and are able to achieve similar win margins against bad teams even without him there.

I think the ability of superstars to put it into another gear is being underappreciated if we put faith so religiously in these +/- numbers.

There are two issues with that statement: 1st, nobody is saying Rose doesn't make a big difference for the Bulls. 2nd, you put you personal belief over facts while trying to spin it as if the facts are not the real thing, but only stuff we put faith in, while your personal belief is actually the reliable thing here. Neither of those things is part of a coherent argument.

Are you telling me that if we play a playoff series between two Bulls teams, one is starting CJ Watson in place of Rose, and the other is starting Ronnie Brewer in the place of Derrick Rose, you believe the Deng led team would win?

I guess you mean Luol Deng as the 2nd "Derrick Rose". Luol Deng instead of Brewer gives us ca. +3.5, Rose over Watson around +2. So, in the end the teams are seperated by 1.5 points. The HCA is neutral, that means we expect the Deng led team to win 3.8 games in a 7 game series. That is pretty close. It depends on how much emphasize you put on "clutch" here, but for me that makes it at best a tie. Let them play that out more often and the Deng team should win more games than the Rose team. And yes, I believe this given the fact that I have seen Deng and Rose play with and without eachother more than enough. ;)

[Rerisen]

I guess you mean Luol Deng as the 2nd "Derrick Rose". Luol Deng instead of Brewer gives us ca. +3.5, Rose over Watson around +2. So, in the end the teams are seperated by 1.5 points. The HCA is neutral, that means we expect the Deng led team to win 3.8 games in a 7 game series. That is pretty close. It depends on how much emphasize you put on "clutch" here, but for me that makes it at best a tie. Let them play that out more often and the Deng team should win more games than the Rose team. And yes, I believe this given the fact that I have seen Deng and Rose play with and without eachother more than enough.

Well I have to say I’m surprised that you seem to believe (your choice of) a +/- metric has solved player value to such an exact incontrovertible degree. Indeed to a degree that makes any kind of context, or really any analysis outside these numbers, worthless or unnecessary. But I remain far more skeptical, and rather believe that in say 10 years, we’ll have something more nuanced and more refined than even the +/- metrics we have now, and that will make RAPM look inaccurate in comparison, and to a similar degree that some believe it makes APM right now, and that APM does to raw +/-.

Most likely it won’t be a catch all number at all, but more detailed breakdowns based on specific circumstances (which I’m sure FO’s have already). What we are dealing with here is just a messy collage of total play time, doing gymnastics to try and even out all the scenarios, assuming to find static player value over a span of years or more when no such thing exists, and then rather blindly exhorting that with these baked numbers, we can now solve all, and indeed totally different, situations.

One look at LeBron James precipitous drop as a player (according to RAPM) going from Cleveland (08-09 and 09-10) to Miami last year tell me no such static value exists. And that changing circumstances change player value. Because I certainly don't believe LeBron is any significant way, less of a player now.

Which is not to say these numbers should be dismissed out of hand, I value them fairly highly, nor that they don’t give valuable insights. But when one has high understanding of the context of a particular team or situation, you can also look at these numbers and see where it might be dangerous or misleading to extrapolate them to vastly different circumstances, especially of which don’t make up any significant portion of the existing data at all.

I’d wager the vast majority of Luol Deng’s minutes without Rose are against 2nd or highly mixed units the last year and a half, excepting the 10 games Rose missed this year, which is mostly against poor competition. I do not believe we can take that information and get a greatly accurate reading on how the team does just flat out without Rose in the lineup over 82 games.

Then we have to understand that Deng’s minutes with Rose are also far greater than his minutes without Rose, and that while they are getting the same credit for those minutes together in raw +/-, their level of contribution to them is not actually the same. And that taking the much smaller time apart and leveraging that unduly against the time together is highly speculative.

I.e. You cannot give me Matt Bonner’s cumulative minute totals over the last several years, built with very specific matchup management by Greg Popovich one game at a time, and use that to try and say Matt Bonner should be starting for the Spurs. Or that it proves he can handle 35 minutes a game because look how the team did in his 1432 minutes last year. Nor that taking him off this years Spurs hurts them more than taking off Tim Duncan. (Bonner has a higher +/- and RAPM, than Duncan this year).

You have to understand the context in which numbers were created. And whether those contexts are similar enough to the new scenario you are trying to project them onto, to make them reliable enough to be a good predictor. In the case of imagining the Bulls without Rose vs without Deng, the scenarios are just not similar enough to the data we have, to do that with reliability.

What the numbers can suggest better is that in the existing environment, *with* Rose healthy and on the team, it’s quite possible the team will continue to do better in Deng without Rose minutes, than Rose without Deng minutes. While still understanding, that the coach is balancing lineup strength differently around the two players, which is undoubtedly playing into the results.

That however is quite a bit different thing that saying who is more valuable not within the current team framework, but in one in which either player would not be there at all.

[mysticbb]

Rerisen, I think you are trying to rationalize your personal belief here much more than "putting numbers into context". Yes, numbers have to been seen in context, but I doubt that someone would claim that the +/- numbers would give a "greatly accurate measure" of what will happen. RAPM is still the best tool in terms of predicting out of sample data. It gives us a pretty good impression, if handled correctly.

Your example of Bonner does not take into account that we are comparing players with similar minutes played against similar strong opponents. Deng is hardly playing garbage minutes and no team is completely without any starting players when Deng is on the court and Rose not. Bonner is getting minutes in garbage time, he is used more when the matchups are in favorite of him. Given the amount of possessions played and the circumstances, it is easy to argue that it is not a good idea to start Bonner and give him more minutes. The only way to apply that example in a useful way to our discussion would be Watson/Brewer. Do you want to say that Brewer can rather maintain his performance level against any kind of opponents than Watson, and thus the difference between Rose and Watson is actually bigger overall, big enough to overcome the rest and not only tie it, but also gives the Rose team a lead?
The issue I have with that is the fact that the results aren't showing it. If the difference is indeed bigger and Rose is more important we should see that either in the results directly or we should see some sort of inconsistency in the data. But neither of those things is there. The results are pretty much in agreement. It just adds up.

Also, Vegas seems to agree with the view that Deng is more important. They have the Bulls in average at +8.5 with both, while at +5 with Deng and without Rose and +4 with Rose and without Deng. They count Deng with +3.5 and Rose with +2.5 (usually a starter is worth between +1 and +2). And I think Vegas knows what they are doing.

For James we have to take into account, that his value dropped, because of the overlapping in terms of skills between him and Wade. We see a similar drop for Wade, btw; Chris Bosh on the other hand stayed pretty consistent at the same level.

So, yeah, circumstances have to be taken into account, we can't ignore context, but in the end that context should show up in the results at some point or should at least create some sort of inconsistency.

[Rerisen]

Rerisen, I think you are trying to rationalize your personal belief here much more than "putting numbers into context"

Everyone's is a 'personal' belief, your personal belief is just in the infallibility of RAPM as a predictor, at least relative to anything else, while I prefer a more wholesome approach using all metrics we have to evaluate players, including firsthand observations, and understanding that players do not have static values.

The issue I have with that is the fact that the results aren't showing it. If the difference is indeed bigger and Rose is more important we should see that either in the results directly or we should see some sort of inconsistency in the data. But neither of those things is there. The results are pretty much in agreement. It just adds up.

The problem is nearly all our ‘results’ are *with* Rose on the team as a starter, other than a weak 10 game sample this year. Which you are giving high esteem to since it just happens to jive with past results, but it’s still not a strong sample of the type and length we truly need to figure out how this team would fare without either player long-term, in order to determine true value.

You say the Bonner example isn’t congruous because of his minutes. And I’m saying that it is similar because you are telling me snippets of Deng playing a few minutes without Rose per game, are going to tell us how the team does at large for 48 minutes every game without him, including closing every tight game the team gets into. It’s not remotely the same circumstances to draw from. That the team choose to purposely rest Rose probably longer than they needed to because they knew they could run over about 5-7 poor teams isn’t really proving much here.

You mentioned Vegas, that’s interesting, be curious to see the source. Last I read about this, none of the stat based models are beating Vegas yet. Though I think last year, Hollinger was the only one to do it. Of course he predicts with a high reliance on PER, and PER doesn’t like Luol Deng too much (it obviously undervalues him), and so does he after his crack about Deng not being a worthy All-Star. But historically, simply using the last season records of most teams, and repeating them as a prediction, ends up beating most stat based predictions for team wins.

What we find as reasoning against this failure to predict often, is that ‘if only’ players played the same as they had during the time the data was built on these stats, that the predictions would have held true. Yet something is missing there. Do players not play the same only because they get better or worse, which certainly happens, but maybe also because their circumstances change? And that most stats have very little way to understand these circumstance changes.

Ronnie Brewer and Joakim Noah are tied for 3rd biggest impact on the Bulls by RAPM. So are both equally valuable to the Bulls assuming even minutes? RAPM would apparently think so, but losing Joakim would hurt far more, because the Bulls have Rip Hamilton now.

If we play a Bulls season without Rose and without Deng, and the Bulls won noticeably less with Deng, your reasoning would be that Deng just played much worse, impacted less. As indeed he might (at least offensively) if the team was robbed of a legit first option for an entire season.

But just perhaps his decline would be because the burden on him suddenly became much greater. Much greater than it is in a giant pool of minutes, but where most of that pool is made up of little drops from individual games where Deng only plays without Rose for a few minutes at a time, against mostly subpar opponents, and where he is never asked to produce as much in crunch time. When expectations were highest for Deng to become a superstar in 08, and he was nearly traded, he pretty much fell apart mentally and had a depressed season under the strain.

I could be wrong and that the Bulls would do worse with Rose and no Deng. And that at the end of a season Luol Deng would have the exact same RAPM with or without Rose on the team. However, I don’t think so, because I do not believe in an absolute static value of players regardless of circumstance. And a team losing its best player (or I guess 2nd best, if you believe Deng is secretly the Bulls best player…) is a pretty massive circumstance change. Not as big as Lebron going from Cleveland to Miami, but pretty big.

However, a team losing its *perceived* best player, and unquestionable leader, which is certainly Rose, would without question be more devastating. I.e, losing Rose over Deng. The Bulls are built on the symbiotic drive of Thibodeau and Rose, coach and best player working together. That Rose misses 10 games means little if everyone knows he is coming back (I think Vegas knew). But If he is not coming back… then your team morale and cohesion sees something much different likely occur. Not admitting this is living in a fantasy world where human beings and psychology do not exist.

Specifically about Deng’s RAPM, it has not stayed static as it is the last few years. Common belief is that a glue guy gains value the better a team becomes, and I think this makes sense. It also matches up with Deng’s value being highest when the Bulls have been better. Take Derrick Rose off the Bulls and they are a lot worse. Take Deng off and they are too of course, but in either situation where the team is noticeably worse, good chance Rose’s ability to impact wins increases relative to Deng, and even RAPM apostles might see how history shows this to be true.

Deng RAPM by year:

2007 4.3 (Team won 49)
2008 1.8 (Team won 33)
2009 1.8 (Team won 41)
2010 2.6 (Team won 41)
2011 4.8 (team won 62)
2012 5.8 (team on a 64 win pace)

So has Deng suddenly become much better? He is pretty much the same player he’s always been, other than Thibodeau imposing on him that he shoot threes instead of so many long two’s. Yet his efficiency has went down this year despite it, and his overall production has yet to again match what it was in 2007. Deng is really not that different of a player, but he is a player that is being maximized in the current team environment. Change that environment around drastically, especially making it worse, and I think Deng is far more at risk than Derrick Rose in losing impact. So... I could almost buy Deng as MVP on the current Bulls... but they could afford to lose Rose far less. ---- Sorry for length.