Did the Hawks change their mind after a week?

[Guest]

I just wanna throw a slide-rule in the room and watch them kill each other over it.

[Guest]

2 minutes ago, kg01 said:

I just wanna throw a slide-rule in the room and watch them kill each other over it.

[AHF]

17 minutes ago, REHawksFan said:

If THJ has a better PER and higher VA and EWA, how does he not compare to KK?  Looks to me like THJ is clearly statistically superior while Prince is about the same in limited minutes.  Everyone is better than Baze but we all knew that. 

THJr has been better this year.  It is why I advocated starting him over Korver some time ago.  We've been discussing whether Bazemore, Prince and Bembry represent an upgrade on him.  Prince's numbers are improving which is what I hope to see continue to happen.  I have to think Bazemore's will improve at some point.  Bembry is probably a year off from being a part of the rotation.

 

Player	PER	WS/48	VORP
Kyle Korver	11	0.092	0.4
Tim Hardaway	13.9	0.087	0
Kent Bazemore	9.5	0.024	-0.1
Taurean Waller-Prince	10	0.055	-0.1
Malcolm Delaney	7.8	-0.003	-0.7
DeAndre' Bembry	7.7	0.004	-0.1

[AHF]

33 minutes ago, sturt said:

I must be missing it, but I'm failing to find an indication that they are not using what they've endorsed. Forgive if I'm just blind... always possible.

Wrote to them to ask that direct question.

The 82 games article I posted was critical of simple plus/minus and said that it doesn't control for key variables as adjusted plus/minus attempts to do.  They only make simple plus/minus available and don't offer a plus/minus stat that attempts to control for variables outside of performance like opponents faced, teammates, etc.  

[sturt]

9 minutes ago, AHF said:

The 82 games article I posted was critical of simple plus/minus and said that it doesn't control for key variables as adjusted plus/minus attempts to do.  They only make simple plus/minus available and don't offer a plus/minus stat that attempts to control for variables outside of performance like opponents faced, teammates, etc.  

Just trying to understand where you're getting it that the stat they're using is simple instead of adjusted. I just find it fairly suspicious that (a) I see a difference between their plus/minus (-15) and what we find listed at NBA.com under traditional, which would naturally seem to be the simple (ie, box score) plus-minus (-4)... and (b) that the site celebrates a stat they don't even use. 

(Finding no success trying to contact anyone there at 82games--links don't actually work, and neither does their phone number. Resorted to write Dr. Ilardi himself to see if he has any insight.)

 

[phoostal]

According to Peachtree hoops they seem to think that ownership stepped in and told Coach bud and Wilcox not to trade Millsap and that they do not want to rebuild has anyone else heard that?

I'm skeptical because ever since Horford left Peachtree Hoops has been pretty much negative towards the Hawks because their favorite player left and I would find it hard to believe that the ownership group would step in and take over control of Coach Bud's final decision since he's the president of operations, but what have you guys heard?

[Lurker]

31 minutes ago, phoostal said:

According to Peachtree hoops they seem to think that ownership stepped in and told Coach bud and Wilcox not to trade Millsap and that they do not want to rebuild has anyone else heard that?

I'm skeptical because ever since Horford left Peachtree Hoops has been pretty much negative towards the Hawks because their favorite player left and I would find it hard to believe that the ownership group would step in and take over control of Coach Bud's final decision since he's the president of operations, but what have you guys heard?

From that same guy that someone said is an unreliable source, it's actually the opposite. Ownership wanted to tear down and Bud was kicking and screaming about it.

Not rumor related or anything, but I bet this is the last season we see Bud in a presidential role unless fortunes continue sliding even better in this season. I don't know that it continues sliding better, but I think we wind up stabilizing at about the point we're at now ROS. It really depends on how this month ends. If it's taken advantage of, as it very well can be, I can actually see upwards of around 52 wins (what a change from about 5 weeks ago). If not? Probably about 45-48 wins...same as last year.

[Guest]

1 hour ago, phoostal said:

According to Peachtree hoops they seem to think that ownership stepped in and told Coach bud and Wilcox not to trade Millsap and that they do not want to rebuild has anyone else heard that?

 

50 minutes ago, Lurker said:

From that same guy that someone said is an unreliable source, it's actually the opposite. Ownership wanted to tear down and Bud was kicking and screaming about it.

You can't both be right. So I'll say both 'sources' are wrong. 

[macdaddy]

I think the truth is that they were listening to offers for Sap until they realized the offers were crap.  So Bud told ownership that they weren't getting anything and might as well go all in on the season.   

We have so assets to trade so maybe we can add someone to this team for a decent playoff run.   

[AHF]

22 hours ago, sturt said:

Just trying to understand where you're getting it that the stat they're using is simple instead of adjusted. I just find it fairly suspicious that (a) I see a difference between their plus/minus (-15) and what we find listed at NBA.com under traditional, which would naturally seem to be the simple (ie, box score) plus-minus (-4)... and (b) that the site celebrates a stat they don't even use. 

(Finding no success trying to contact anyone there at 82games--links don't actually work, and neither does their phone number. Resorted to write Dr. Ilardi himself to see if he has any insight.)

 

You can see their calculation is the definition of simple plus/minus.  It is purely what happens when on the court and what happens when off the court.  That is the very definition of not controlling for other variables.

[sturt]

I've reviewed what's on the 82games site, including the Rosenbaum piece that talks about comparing the adjusted to the simple box score plus-minus to see if maybe I presumed something in error. I don't think I have.

If I'm understanding you correctly, what you might be missing is that the "adjusted" part of the calculation is already completed within the process of generating the four numbers that equate to the final net plus-minus number. So, iow, we are only being fed the end numbers generated by the adjustment formula for each: On-Court Offense (102.8 in this case with Korver), On-Court Defense (111.1), Off-Court Offense (110.2) and Off-Court Defense (102.9). By definition, then, the plus-minus part of "adjusted plus-minus" results from doing the "plussing" and "minusing" necessary between the four numbers generated.

 

Again, what I said before serves as testimony to that point. The current simple, or box score plus-minus for Korver listed on NBA.com is around -4. 

(Regrettably, I'm still waiting for an e-mail response from the Kansas professor to gain a fuller explanation. I suppose I could pursue tracking down Rosenbaum if I don't hear something soon, but as the semester gets started next week, this isn't likely to survive my priority list.)

[AHF]

1 hour ago, sturt said:

I've reviewed what's on the 82games site, including the Rosenbaum piece that talks about comparing the adjusted to the simple box score plus-minus to see if maybe I presumed something in error. I don't think I have.

If I'm understanding you correctly, what you might be missing is that the "adjusted" part of the calculation is already completed within the process of generating the four numbers that equate to the final net plus-minus number. So, iow, we are only being fed the end numbers generated by the adjustment formula for each: On-Court Offense (102.8 in this case with Korver), On-Court Defense (111.1), Off-Court Offense (110.2) and Off-Court Defense (102.9). By definition, then, the plus-minus part of "adjusted plus-minus" results from doing the "plussing" and "minusing" necessary between the four numbers generated.

 

Again, what I said before serves as testimony to that point. The current simple, or box score plus-minus for Korver listed on NBA.com is around -4. 

(Regrettably, I'm still waiting for an e-mail response from the Kansas professor to gain a fuller explanation. I suppose I could pursue tracking down Rosenbaum if I don't hear something soon, but as the semester gets started next week, this isn't likely to survive my priority list.)

The bold and underlined point is 100% wrong.  Adjusted box score adjusts for the opponents you face, the teammates you play with, etc.  There is no such adjustment going on here.

For example, if Miami's championship teams featured lineups where they ran Lebron, Wade and Bosh together all the time with a crappy PG and they ran those guys as a unit you would see a great plus/minus for that crappy PG despite that PG's very crappy play because he is playing with a bunch of studs and they will inevitably do better than Miami's reserve unit which is worse than usual because the team spent all the money on its stars.  The PG gets a nice plus/minus from two factors (1) the guys on the floor with him are studs and (2) the guys on the bench behind those studs are terrible.  Thus the differential is much larger than it would be on a team with 10 guys who are all pretty good. 

The type of plus/minus that is shown on 82games just gives you that raw number.  Adjusted plus/minus does regression to attempt to adjust for the fact that the plus/minus is inflated when you play only with guys who are way, way above average on the floor with you and inflated more when the reserves are so much worse than average.  The adjustment is trying to control for the all the variables that are outside of a player's control and get to a plus/minus that more directly correlates with the actual play of the individual and less about who else is on the floor or whether he facing an opponent's reserves, etc.

If you read that article I posted earlier, it talks about all this.

If you are familiar with statistics, then look to the concept of controlling for variables.  

Simple plus/minus such as that on 82games does not do this.  That doesn't make it worthless but it needs to be taken in context and is just one of many things to look at.  

[Thomas]

On 1/12/2017 at 1:02 PM, kg01 said:

I just wanna throw a slide-rule in the room and watch them kill each other over it.

Isn't it more fun to just watch the event itself. 

[sturt]

First, @AHF,  I don't mean to go all Mike McCarthy (as-in GB coach) on anyone as-if feeling a need to legitimize my standing, but I'd thought you might remember that I'm in a profession where some familiarity (at minimum) with the language of statistics ordinarily can be assumed. It's kinda like, while I'm plausibly able to engage intelligently to some degree in discussions about constitutional law, i think I've been very cognizant in how I speak to anyone with a law degree in such discussions that, regardless of whether constitutional law is their strongest area, they're most probably more than qualified for me to assume they're at least as conversant as I am.

Just sayin.

 

2 hours ago, AHF said:

The bold and underlined point is 100% wrong... The type of plus/minus that is shown on 82games just gives you that raw number.

 

Let's just pretend for the moment that it is a correct statement that your assertion is correct.

I've read both Dr. Ilardi's 2007 article and the Rosenbaum article that it links to.

With all due respect, there simply is no such evidence as you attempt to suggest there is for your assertion.

To the contrary, Ilardi says,
 

Quote

 

The basic model – which yields a player statistic known as adjusted plus-minus - has already been described in admirable detail by Rosenbaum (see Measuring How NBA Players Help Their Teams Win). My task in this article, therefore, is merely to provide a gist-level sense of what the model does, and a means of understanding the accompanying rank-ordered listings of each player in the league who logged at least 400 minutes during the 2006-2007 season...  What’s needed, of course, is some way of adjusting the plus-minus stat to account for all such potential confounds. This is exactly what the adjusted plus-minus stat does: it reflects the impact of each player on his team’s bottom line (scoring margin), after controlling statistically for the strength of every teammate and every opponent during each minute he’s on the court. Again, the gory mathematical details of the adjusted plus-minus model have been described elsewhere (and they are beyond the scope of this article)... , the adjusted plus-minus model exhibits superb face validity; it yields results that in many respects mirror our own intuitions about how players should stack up against one another. And remember, it does so by means of a complex mathematical analysis that completely ignores boxscore stats. (These adjusted plus-minus numbers are derived without a single direct input from stats like points scored, rebounds, assists, FG %, blocks, steals, free throws, fouls, etc.)

 

 

 

 

Then, Rosenbaum's article says this:
 

Quote

 

Thus, a better measure of player value would “adjust” these plus/minus ratings to account for the quality of players that a given player plays with and against.  In addition, it would account for home court advantage and for clutch time/garbage time play.  Thus, unlike in unadjusted plus/minus ratings, these “adjusted” plus/minus ratings do not reward players simply for being fortunate to being playing with teammates better than their opponents.  Contributions for individual players are isolated statistically.  In this article, I develop adjusted plus/minus ratings... I improve on past efforts by combining estimates of player value using both (a) pure adjusted plus/minus ratings and (b) a statistical index derived from these pure adjusted plus/minus ratings.

 

 

 

(Related but somewhat a sidebar: Look at any given player's 82games.com profile. You'll find that there are (a) plus/minus ratings supplied, as well as other stats that are relevant to (b) the statistical index part of this hybrid approach he's advocated.)

So, how does Rosenbaum say he arrives at the adjusted plus/minus?

Again, I grant some latitude here that perhaps I'm misinterpreting something in what you said, but as I take it, you're contending that essentially we know that 82games does not provide us with the adjusted plus-minus because they are providing 4 numbers (on-court offense, on-court defense, off-court offense, off-court defense) from which the net result is calcuated, and that's not how the formula is conveyed by what is found in Rosenbaum's explanation.

And while that's true--i.e., Rosenbaum conveys it in these terms...

... what you, me and everyone else learned in elementary school is that distributive law allows us to understand that the same result can be had from multiple ways of conveying a formula, as long as you're rationally plugging in the same numbers as the basis of your calculation.

In other words, it is completely plausible that the four numbers provided (on-court offense and def, off-court off and def) are produce-able and rendered according to Rosenbaum's calculation... and especially so since we know that 82games even provides us 5-man unit information that is characteristic of the non box-score data necessary to calculate an adjusted plus-minus. It simply confounds reason that they would not be using that, as opposed to using box score data.

So, look, I hardly ever tell someone that they're 100% wrong about anything. I've been wrong enough to know that's not generally a good idea for me. And I'm not going to do that here, even now, even with as much pointing to the conclusion as I've laid out. I can't know because I can't see the actual numbers plugged in. Even my assertion that there's clearly a difference in the simple plus-minus found at NBA.com for Korver (-4) and what we find at 82games (-15) can maybe someway somehow be explained by the point that the 82games number is stated to be based on a projection of 100 possessions. (But golly, that seems such a huge delta at face value.) On this, for multiple reasons, I have strong confidence in my conclusion, and I'll leave it at that.

You can have the last word, but now, right or wrong, I've got some syllabus writing to do. (That is, "last word" unless I get a reply from Ilardi, of course, in which case, I'll pass that along.)

 

 

[AHF]

Sorry if I came off as insulting, @sturt.  Truly did not mean to do so.  

82games however gives no indication that it factors in anything like the following from what you quoted:

Quote

controlling statistically for the strength of every teammate and every opponent during each minute he’s on the court

Quote

“adjust” these plus/minus ratings to account for the quality of players that a given player plays with and against.  In addition, it would account for home court advantage and for clutch time/garbage time play.

In contrast, for example, real plus/minus says the following in addition to noting it adjusts for 100 possession pacing:

Quote

RPM takes into account teammates, opponents and additional factors

Korver's RPM to this point in the season is -2.89.  Not good (at all) but far, far from -15.

[AHF]

Finally, I will note the following:

82games Net Plus/Minus Ryan Kelly = +19.7  

(1.1 ppg, 0.460 TS%, 1.1 rpg, 0.2 bpg, 0.4 spg, 12.2 PER, 0.1 WS)

82games Net Plus/Minus Zaza Pachulia = +9.8  

(5.6 ppg, .598 TS%, 5.7 rpg, 0.5 bpg, 0.8 spg, 15.4 PER, 2.4 WS)

82games Net Plus/Minus Anthony Davis = +7.0 

(29.1 ppg, .574 TS%, 12.2 rpg, 2.5 bpg, 1.4 spg, 28.2 PER, 5.8 WS)

82games Net Plus/Minus Dwight Howard = +0.3

(14.0 ppg, .628 TS%, 13.4 rpg, 1.3 bpg, 0.9 spg, 22.5 PER, 4.2 WS)

82games Net Plus/Minus Karl Anthony Towns = -0.8

(21.7 ppg, .562 TS%, 11.8 rpg, 1.5 bpg, 0.6 spg, 22.5 PER, 4.4 WS)

I'll let people make their own judgments which of these numbers tell you the most accurate story about the value these PF/Cs have brought with their play.