APBRmetrics

schtevie wrote:MW00, you make several interesting points. First, you offer an alternative perspective of Point (1) of my argument on behalf of the "Russell Dynasty" (Coincident with Russell's arrival and departure, the Celtics took a discrete and large step up and down in defensive performance.) As for his arrival...

MW00 wrote:On Russell, the “before after” story posited with him is missing a number of salient points. The first being Boston’s increase in performance came with him missing a significant portion of the season and winning a better percentage of games with him away.

This is true (as I discovered, owing to your intervention) but I am quite sure that my interpretation stands. To begin, in this general exercise wins are the tail, and per possession efficiency at scoring and preventing points is the dog. This is the essence of a +/- perspective. And if you go to the box scores at basketball reference, you see that though the Celtics' winning percentage was a bit higher in its first third of the season compared to the remainder, you also see that the points scored per game and points allowed per game were marginally higher and lower, respectively, in the latter two thirds of the season. So, on those terms the offense and defense were both a bit better with Russell playing. But again, we care about efficiency, so we need estimates of possessions per game. And sadly we don't have them (or at least I don't). But we do know a couple of very relevant facts that suggests what the truth of the matter is.

First, we know that as a general rule, defenses rule early on, with offenses catching up as the season progresses. And this is not a trivial factor (witness the decrease in average scoring efficiency in two strike-shortened seasons). And second, we know that during the years in question the NBA possessions per game were increasing significantly on an annual basis. And if any of this increase was happening within-season, this too would serve to decrease offensive efficiency and raise defensive efficiency.

Then, as for the interpretation of the year after Russell retired...

MW00 wrote:Then when he left it coincided with the leaving of Sam Jones and the rapid falling off of Bailey Howell, one of the “couple, future hall of famers” who they retained (more on this shortly). The loss of Russell and Jones meant Boston went from a “win now” team to a “build for the future” team. Thus Don Chaney, then shooting 36% started taking a portion of Larry Siegfried (competent playmaker and another of those Russell teammates with a strong defensive reputation)’s minutes. Howell’s falloff is a major issue in the drop off too. Whether it was age, the less competitive situation, inferior playmaking or the absence of Russell we can’t be absolutely certain, but I’d credit them in roughly that order with Russell’s absence having minimal direct impact (indirectly influencing through worse playmaking a little, and competitiveness significantly).

My response to this is basically indirect. The kind of "event analysis" we are engaged in here is inherently imprecise and subject to varying interpretations. My perspective is to take a "macro" approach in accounting for the estimated deterioration: 1.4 and 4.4 points per 100 possessions on offense and defense respectively, compared to the NBA average. Looking at the season summaries, I know who left, their ages, and the minutes they played. Similarly, I can see who came in, their ages, and make a pretty good guess of how the minutes of the departed were allocated by position. And what this exercise tells me is that my interpretation is robust. This is to say that, all else equal, replacing the minutes of 34 and 35 year old players with those of younger players on average should net you a positive gain on aging curve grounds. But as it turns out, things went way south. Furthermore, we tend to know that centers are really and disproportionately important for defense (and Russell's minutes were by far the greater).

So, I think it is most plausible to say that my Point (1) stands: Russell's arrival suggests a defensive identity being stamped on the Celtics, just as his departure indicates that he remained a defensive force to the end of his career.

But what about the years in between, Point (2) notes that the estimated defensive arc corresponds to that of a great defensive player, and I offered Ben Wallace as a thought experiment from a modern context.

I am not sure, but in saying this, I get the impression that the analogy might have made you a bit unhappy...

MW00 wrote:Wait, Ben Wallace fully explained Detroit’s above-averageness in ‘04? Any reason why Detroit with the four other starters in place (admittedly Sheed for a full season), in 07-08 were (+)6.67 SRS without Ben Wallace. That is to say with the same core they remained about the same (substantial) distance above average and previously that was all provided by Big Ben, and later it was from the players who played most of the minutes that brought no net value above .500?

To this, I can only reply that if you have any such questions, go to Jeremias' site for the (coherent) +/- answers. I related the accounting his work provides for 2003-04. And 2007-08's can be found at http://stats-for-the-nba.appspot.com/ratings/2008.html. Punchline: there is no mystery.

MW00 wrote:I’m not unsympathetic to the idea that some of the members of the Boston Dynasty have been overrated, or that defense is underrated in compiling these sort of rankings. But when we’re hypothesizing that Ben Wallace’s negative impact on an offense, whilst offering a .441 ts% on 15.3% usage, was marginal, I’m not convinced. And then taking that approach back to an era where a lot of data is missing to conclude that Russell won all Boston’s titles single-handed.

Again, it is not my point that Russell won all those titles single-handed, only that the contribution of the Other is approximately average. This is a key distinction.

This aside, given the above comment, I am supposing that you won't be convinced. Too bad, the data as rendered in Jeremias' xRAPM consistently shows that Ben Wallace wasn't a major drag on offense (his average ORtg from 2001 to 2012 was only -1.5). What this means is that his really terrible scoring ability (ranked 88 out of 91 centers and center type forwards on TS% between 1997 and 2009 for players playing more than 20 mpg - this range being chosen to correspond to the exact same years of Russell's career) as well as his mediocre TO% and AST% (both 57 out of 91) doesn't prevent you from being not too bad on offense, as long as you can get offensive rebounds (ranked 12 out of 91) and presumably to a bunch of other non-box score things well as well.

But we're really talking about Bill Russell here. And doing a similar (but necessarily limited) comparative ranking (and taking into account the consensus narrative about blocks and rebounding) we get the impression that he and Ben Wallace were very, very similar players. Of particular note is the cumulative career comparison for PF/36 minutes. Russell ranked 2nd best out of 71 players, and Wallace, best out of 91. The only seemingly significant difference between the two players that I identify is that Russell was relatively speaking a distinctly better scorer. His TS% was average, instead of terrible, 39 out of 71.

So, there you go. Again, it was the Russell Dynasty.

The bullet point version of my response:

-Boston probably were better with Russell. But the difference is relatively marginal. The important point to take is that they were good without him. I don't think anything you have said or can say will refute this.

-Re The the indirect response: Sure we can at the macro level see Russell leaving having an impact. I can also see some other things having an impact. Yes, Russell was better than most 35 year olds (and Boston's replacements worse than typical 24 year olds) that doesn't mean he accounts for all Boston's above averageness or all their drop off.

-Re Detroit: Billups was better in '08 than '04 but that much better? The reading on Wallace certainly seems noisy, as looking at say '06 and '07, in '06 Wallace is the best player per possession in the league and, I think, more than explains Detroit's above averageness (net +9.1 per 200 possessions, 35.2mpg, pace 86.8, SRS+6.23) yet replaced by a washed up Chris Webber, Nazr Mohammed and Jason Maxiell (hardly remarkable players), Detroit remain significantly above average. And that in a relative down year by Billups. When Billups gets back to '06 levels of performance in '08, with the same core Detroit are back at better level (higher SRS) than they ever were with Wallace.

Re conclusions: I understand what you mean when you say he is "only" accounting for all above averageness. Given that .500 is pretty much the default and the substantial degree you need to be above that to win the title (excepting occasional flukes) I would pretty much say that doing so is pretty much winning titles single-handedly. Okay you might expect a team minus its best player to be a bit under .500 and so say his supporting cast is a touch above the average (that is of the average team, though, not the average contender), but it's close enough that I would say your interpretation of the dynasty fits well with the phrase: Russell won the titles single-handed.

Went with all new(er) players. Shocked no one else had a vote for Durant or CP3; KD will likely go down as the 2nd best SF of all time, and Chris Paul has the inside track as the GOAT pure PG. (Arguable if he passes Magic for career, depends heavily on postseason luck)

We aren't (supposedly) voting on what players might do or might have done. We're ranking their careers to date.

It's funny, in the first vote of the first thread -- pick top 25 from a group of 50 -- Stockton was near unanimous. In this round, he has 2 of 6 possible votes.

Elvin Hayes, Stockton, Wilkins, Drexler, Wade, Alex English, George Gervin etc. My last choice is too tough. I think I'll go with Gervin although he had never reached the finals.

Given Elvin Hayes and Isiah Thomas are already getting votes:

Elvin Hayes only once shot over 50% from the field for a season.
Elvin Hayes only thrice in his extended prime (i.e. everything but the last spell in Houston) had a TS% above .500
Elvin Hayes twice shot above 70% from the free throw line.
Elvin Hayes never managed a PER above 20, PER being a metric which's most noted criticism is that it overvalues volume scorers. For reference Chris Gatling's best season was a 23.9. Eric Murdock and Matt Geiger both had seasons above 20.
At least PER thinks he was mostly good, if never particularly special. WS/48 considered him only "above average" for most of his prime and most of that value coming in the more dubious defensive win shares).

And he was percieved as having poor entangiables and (fairly or otherwise) as being a choker.

He really excelled at playing big minutes and shooting a lot. He played on a team that won a title.

To those that have or would vote for him, what did he do that say Bob Lanier or Artis Gilmore (0 votes each) didn't http://www.basketball-reference.com/pla ... 01&y3=1988

I'll may do this with Isiah when I have the time, but if I don't the point will be: he was on team that won two titles, though they won them through D (and he wasn't a noted defender despite the steals), apart from strong though hardly historically unique 3 year stats peak he shot fairly poorly for his era and turned the ball over a lot; he only once finished top 5 in MVP voting (5th); his advanced stats and peaks are comparable to Tim Hardaway, Mark Price and Kevin Johnson but because he played in thinner talent era for point guards collared a few more awards and he should be in their ballpark, not this one.

MW00 wrote:The bullet point version of my response:

-Boston probably were better with Russell.

Er, yes...probably.

MW00 wrote:..But the difference is relatively marginal.

So, I'm guessing poor ol' Bill doesn't have your vote for top 5 player then?

MW00 wrote:...The important point to take is that they were good without him.

As in, good enough to get in the playoffs and lose...sure.

MW00 wrote:I don't think anything you have said or can say will refute this.

I am prepared to believe that there are folks out there immune to evidence and argument. Clearly my points have not been fully appreciated, what only suggests to me that I should type louder.

But before I repeat the main argument, let me deal more directly with the first of the chronological bookends: the interpretation of the before and after of Bill Russell's joining the Celtics. You brought to our attention the important point that the Celtics, in the first 24 games of the 1956-57 season, had a higher winning percentage than in the remainder, suggesting that this establishes that Russell (in at least the first year) added only marginal value. My reply was that, in fact, on a points scored/points allowed basis, that the Celtics actually marginally improved on both measures in the later 2/3 of the season AND that there is historical reason to believe that there is an anti-defense bias in such "nominal" measures as traditionally offenses improve relative to defenses over the course of the season. But you chose not to follow up on this factual point. So I did.

Sure enough, come the arrival of Bill Russell, here's what we see: As alluded to, PPG scored, increased from 105.2 to 105.6 and PPG allowed decreased from 100.6 to 100.0. But what of the league average? PPG increased from 97.6 to 100.6 in the latter 2/3 of the season. So, compared to average, on offense, the Celtics went from 7.5 PPG above to 5.1 (i.e. a relative worsening of 2.4 PPG) and on defense, the Celtics went from 2.99 below average to 0.6 above average (i.e. a relative improvement of 3.6 PPG). All during a time when the offense were seeing their anticipated seasonal improvement.

So, the bottom line is that the traditional narrative holds: Bill Russell came in and had an instant positive impact on defense. And if one contextualizes this, in term of his minutes played (less that 75% of available) and reasonable estimates of possessions per game, what one infers is the arrival of a defensive phenom, rookie year, probably about +4 points per 100 possessions as a DRtg.

But this is only a lead-in to the main argument. Though there is a statistical void what prevents "exact" estimates of team ORtg and DRtg, there is simply insufficient leeway in the range of possible values of the missing data that can overturn the conclusion that the Russell Celtics were at best a mediocre offense and at worst a well above-average defense. To disagree with this statement (barring explicit evidence and argument) is to venture into "opinions on the shape of the earth differ" territory.

Stipulating this, one then needs to confront the universally (?) accepted view that Bill Russell was the best defensive center of his era. But what does that mean? Well, since 2000-01, we have been blessed with both PbP data and their publicly available interpretation (continuing thanks, J.E.) so that we have a sense of what his impact was, and his name is Ben Wallace, a player whose statistical profile bear eerie resemblance to what we know (or are led to believe) of Bill Russell. And when we create a composite (Bill Wallace/Ben Russell?), taking the annual xRAPM numbers of Ben and multiply them times the share of Celtics minutes played by Bill, what we see is that this estimated defensive impact is eerily close to (any reasonable estimate of) actual Celtics performance. Quelle surprise!

MW00 wrote:-Re The the indirect response: Sure we can at the macro level see Russell leaving having an impact. I can also see some other things having an impact. Yes, Russell was better than most 35 year olds (and Boston's replacements worse than typical 24 year olds) that doesn't mean he accounts for all Boston's above averageness or all their drop off.

Having thought about the "Bill and Sam" identification problem a bit more, it occurs to me that the argument that "it really was Bill's leaving" can be significantly sharpened.

So, we agree that the large drop-off after 1968-69 was overwhelmingly defense? (I've tentatively got the relative decline as 1.8 points per 100 possessions on offense, and 4.4 on defense. Fiddle on the margins as you will.) The question then is how much of the defensive decline is apt to be accounted for by the retirement of a 35 year old shooting guard (G-F, according to basketball-reference) versus that of the aging defensive center of his generation?

Pretending that we don't already know that it's centers that primarily influence defensive performance and not shooting guards, I went to basketball-reference and got a list of guards and guard-forwards (Sam Jones' B-R designation) in the seasons 2000-01 to 2012-13, with characteristics roughly akin to Sam Jones in his final year: MPG>20, height<=6'5", age >= 34, and FGA/36min >= 15. This specification, to be clear, allows for players a bit taller, older, and much less offense-minded than Sam Jones was in his final season (where he shot almost 22 times per 36 minutes). This list of players was then cross-referenced with Jeremias' data.

And, what, do you think was found, regarding the resulting defensive ratings? Not one of said players had a positive DRtg. And this, of course, makes perfect sense. It is damn tiring being the focus of an offense (what he was) and one shouldn't expect such shooting guards to be able to be positive contributors on defense. So, then I checked FGA/36 minutes between 14 and 15. Same story. 13 and 14? Same story. 12 and 13? You guessed it. It was only when we got between 10 and 11 FGA per 36 minutes that we first find a player (the amazing John Stockton) who was able to muster a positive DRtg.

I could go on about how Jason Kidd was the only (low usage and) aged shooting guard with (relatively, in a positional sense) high defensive ratings, but that would only reinforce the point about the trade-off between offensive usage and defensive excellence. But the very clear point is this: there is no plausible estimate of the defensive impact of Sam Jones that would overturn the conclusion that the defensive decline of the Celtics in 1969-70 was owing (to the extent that it can be personified) to the retirement of Bill Russell.

It really, really was the Russell Dynasty.

Finally, a post script. This defense of the defense of Bill Russell is not simultaneously an endorsement of the view that he belongs in the top 5. For this, given the well-established competition, it needs to be shown that his contribution on offense was in fact positive. (A Ben Wallace-esque peak DRtg of 9.8 doesn't quite cut it.) And here, I don't think that there is any highly determinative evidence. Yes, he was a better scorer than Ben Wallace, relative to the centers of his day, and perhaps the best informed view is that this (plus all the other little things) would be sufficient to put him in +10 territory for several years (what is the cut-off in +/- criteria, anyway). But maybe not.

Finally...

MW00 wrote:-Re Detroit: Billups was better in '08 than '04 but that much better? The reading on Wallace certainly seems noisy, as looking at say '06 and '07, in '06 Wallace is the best player per possession in the league and, I think, more than explains Detroit's above averageness (net +9.1 per 200 possessions, 35.2mpg, pace 86.8, SRS+6.23) yet replaced by a washed up Chris Webber, Nazr Mohammed and Jason Maxiell (hardly remarkable players), Detroit remain significantly above average. And that in a relative down year by Billups. When Billups gets back to '06 levels of performance in '08, with the same core Detroit are back at better level (higher SRS) than they ever were with Wallace.

I am confused. Are you somehow arguing with the results shown in http://stats-for-the-nba.appspot.com/ratings/2008.html? And if so what is the basis of the argument? Are you saying that the sum of minute/possession-weighted xRAPM estimates don't approximate actual team performances? Are you saying that the estimates of individual performances (offensive, particularly) diverge significantly from other reputable sources? What is the precise point you are making here? I am at a loss.

Elvin Hayes only once shot over 50% from the field for a season.
...
He really excelled at playing big minutes and shooting a lot. He played on a team that won a title.

Let's not forget that he was also a premier rebounder of his day.
And he was a shotblocker, so we might guess he was something of a defensive stopper. Twice all-D, in fact.

Among the 50 on the current ballot, he's the 14th best rebounder. Of the 13 who are higher, he's about in the middle in scoring proficiency.
Hayes' block rate is below those of Ewing, Dwight, and Artis, but doesn't include his first 5 seasons.

Check out Hayes, Lanier, Moses, and Parish: http://bkref.com/tiny/ySLqW
Advanced stats don't love Elvin. But look at their playoff numbers, and he quite holds his own.
Excelling in playoffs is generally the opposite of "choking".

Elvin Hayes pretty much exemplified the NBA PF in the '70s. Look at most WS in a season by F and F-C (basically forwards who did not play guard) from 1969 to 1979.
http://bkref.com/tiny/aWigv

Appearances in the top 100:
11 - Hayes
6 - Barry*
5 - Hairston, Hawkins*, J Lucas, Nelson, Silas, Tomjanovich
4 - Cunningham*, Haywood*, McGinnis*
3 - Mickey Johnson, Bobby Jones*, Kauffman, Kenon*, Love, McMillian,
2 - Baylor, J Green, Mix, Wicks

This is NBA only, and ABA vets are starred. In the NBA, no PF has half as many top seasons as Hayes in the interval.
And WS doesn't even 'like' Hayes.

With 7 voters and 70 votes, we have 7 players with 4 or more votes; and another 5 with 3 votes. These players are 'on the fence' : Barry, Frazier, Moses, Ewing, and Pippen.

How great was the career of Walt Frazier? Counting playoffs, he was in 915 games. Of those voted in, or ahead of him in this round, only Pettit was in fewer games (LeBron's about there).

He was heir apparent to Oscar and West -- rather outplaying West in the Finals a few times. But just 4 years after those guys quit, Frazier too was effectively done.

He put together a run of 7 allstar games, 6 all-NBA seasons, and 7 all-D 1st teams.
In the 7 years 1969-75, he has 92.5 Win Shares. Next best, among guards, are West (66), Oscar (59) ...
Among all players, he's behind only Kareem (115) in that span. Well ahead of Walker (77), Havlicek (74) ...

And in playoffs for that interval, his 15.8 WS easily outpaces Kareem (12.6), Wilt (12.1), West (10.9), Havlicek (10.1) ...

I dunno -- are 7 super-stellar seasons and postseasons enough?

schtevie wrote:

MW00 wrote:The bullet point version of my response:

-Boston probably were better with Russell.

Er, yes...probably.

MW00 wrote:..But the difference is relatively marginal.

So, I'm guessing poor ol' Bill doesn't have your vote for top 5 player then?

MW00 wrote:...The important point to take is that they were good without him.

As in, good enough to get in the playoffs and lose...sure.

MW00 wrote:I don't think anything you have said or can say will refute this.

I am prepared to believe that there are folks out there immune to evidence and argument. Clearly my points have not been fully appreciated, what only suggests to me that I should type louder.

But before I repeat the main argument, let me deal more directly with the first of the chronological bookends: the interpretation of the before and after of Bill Russell's joining the Celtics. You brought to our attention the important point that the Celtics, in the first 24 games of the 1956-57 season, had a higher winning percentage than in the remainder, suggesting that this establishes that Russell (in at least the first year) added only marginal value. My reply was that, in fact, on a points scored/points allowed basis, that the Celtics actually marginally improved on both measures in the later 2/3 of the season AND that there is historical reason to believe that there is an anti-defense bias in such "nominal" measures as traditionally offenses improve relative to defenses over the course of the season. But you chose not to follow up on this factual point. So I did.

Sure enough, come the arrival of Bill Russell, here's what we see: As alluded to, PPG scored, increased from 105.2 to 105.6 and PPG allowed decreased from 100.6 to 100.0. But what of the league average? PPG increased from 97.6 to 100.6 in the latter 2/3 of the season. So, compared to average, on offense, the Celtics went from 7.5 PPG above to 5.1 (i.e. a relative worsening of 2.4 PPG) and on defense, the Celtics went from 2.99 below average to 0.6 above average (i.e. a relative improvement of 3.6 PPG). All during a time when the offense were seeing their anticipated seasonal improvement.

So, the bottom line is that the traditional narrative holds: Bill Russell came in and had an instant positive impact on defense. And if one contextualizes this, in term of his minutes played (less that 75% of available) and reasonable estimates of possessions per game, what one infers is the arrival of a defensive phenom, rookie year, probably about +4 points per 100 possessions as a DRtg.

But this is only a lead-in to the main argument. Though there is a statistical void what prevents "exact" estimates of team ORtg and DRtg, there is simply insufficient leeway in the range of possible values of the missing data that can overturn the conclusion that the Russell Celtics were at best a mediocre offense and at worst a well above-average defense. To disagree with this statement (barring explicit evidence and argument) is to venture into "opinions on the shape of the earth differ" territory.

Stipulating this, one then needs to confront the universally (?) accepted view that Bill Russell was the best defensive center of his era. But what does that mean? Well, since 2000-01, we have been blessed with both PbP data and their publicly available interpretation (continuing thanks, J.E.) so that we have a sense of what his impact was, and his name is Ben Wallace, a player whose statistical profile bear eerie resemblance to what we know (or are led to believe) of Bill Russell. And when we create a composite (Bill Wallace/Ben Russell?), taking the annual xRAPM numbers of Ben and multiply them times the share of Celtics minutes played by Bill, what we see is that this estimated defensive impact is eerily close to (any reasonable estimate of) actual Celtics performance. Quelle surprise!

MW00 wrote:-Re The the indirect response: Sure we can at the macro level see Russell leaving having an impact. I can also see some other things having an impact. Yes, Russell was better than most 35 year olds (and Boston's replacements worse than typical 24 year olds) that doesn't mean he accounts for all Boston's above averageness or all their drop off.

Having thought about the "Bill and Sam" identification problem a bit more, it occurs to me that the argument that "it really was Bill's leaving" can be significantly sharpened.

So, we agree that the large drop-off after 1968-69 was overwhelmingly defense? (I've tentatively got the relative decline as 1.8 points per 100 possessions on offense, and 4.4 on defense. Fiddle on the margins as you will.) The question then is how much of the defensive decline is apt to be accounted for by the retirement of a 35 year old shooting guard (G-F, according to basketball-reference) versus that of the aging defensive center of his generation?

Pretending that we don't already know that it's centers that primarily influence defensive performance and not shooting guards, I went to basketball-reference and got a list of guards and guard-forwards (Sam Jones' B-R designation) in the seasons 2000-01 to 2012-13, with characteristics roughly akin to Sam Jones in his final year: MPG>20, height<=6'5", age >= 34, and FGA/36min >= 15. This specification, to be clear, allows for players a bit taller, older, and much less offense-minded than Sam Jones was in his final season (where he shot almost 22 times per 36 minutes). This list of players was then cross-referenced with Jeremias' data.

And, what, do you think was found, regarding the resulting defensive ratings? Not one of said players had a positive DRtg. And this, of course, makes perfect sense. It is damn tiring being the focus of an offense (what he was) and one shouldn't expect such shooting guards to be able to be positive contributors on defense. So, then I checked FGA/36 minutes between 14 and 15. Same story. 13 and 14? Same story. 12 and 13? You guessed it. It was only when we got between 10 and 11 FGA per 36 minutes that we first find a player (the amazing John Stockton) who was able to muster a positive DRtg.

I could go on about how Jason Kidd was the only (low usage and) aged shooting guard with (relatively, in a positional sense) high defensive ratings, but that would only reinforce the point about the trade-off between offensive usage and defensive excellence. But the very clear point is this: there is no plausible estimate of the defensive impact of Sam Jones that would overturn the conclusion that the defensive decline of the Celtics in 1969-70 was owing (to the extent that it can be personified) to the retirement of Bill Russell.

It really, really was the Russell Dynasty.

Finally, a post script. This defense of the defense of Bill Russell is not simultaneously an endorsement of the view that he belongs in the top 5. For this, given the well-established competition, it needs to be shown that his contribution on offense was in fact positive. (A Ben Wallace-esque peak DRtg of 9.8 doesn't quite cut it.) And here, I don't think that there is any highly determinative evidence. Yes, he was a better scorer than Ben Wallace, relative to the centers of his day, and perhaps the best informed view is that this (plus all the other little things) would be sufficient to put him in +10 territory for several years (what is the cut-off in +/- criteria, anyway). But maybe not.

Finally...

MW00 wrote:-Re Detroit: Billups was better in '08 than '04 but that much better? The reading on Wallace certainly seems noisy, as looking at say '06 and '07, in '06 Wallace is the best player per possession in the league and, I think, more than explains Detroit's above averageness (net +9.1 per 200 possessions, 35.2mpg, pace 86.8, SRS+6.23) yet replaced by a washed up Chris Webber, Nazr Mohammed and Jason Maxiell (hardly remarkable players), Detroit remain significantly above average. And that in a relative down year by Billups. When Billups gets back to '06 levels of performance in '08, with the same core Detroit are back at better level (higher SRS) than they ever were with Wallace.

I am confused. Are you somehow arguing with the results shown in http://stats-for-the-nba.appspot.com/ratings/2008.html? And if so what is the basis of the argument? Are you saying that the sum of minute/possession-weighted xRAPM estimates don't approximate actual team performances? Are you saying that the estimates of individual performances (offensive, particularly) diverge significantly from other reputable sources? What is the precise point you are making here? I am at a loss.

Use my quotes in context or not at all. Boston "probably being better with Russell is made in response to a post, which by chopping my response (without said context) and answering it, makes it look as though I'm saying Russell's career impact, rather than on that first season, is marginal. You further do so by saying they were "good enough to get in the playoffs and lose...sure" when, again the discussion laid out is clearly within the context of the '57 Celtic team which was way out ahead of any other team, prior to Russell's arrival.

This is not the first time that you have (deliberately?) misrepresented my position earlier you posited that my argument was that "Coincident with Russell's arrival and departure, the Celtics took a discrete and large step up and down in defensive performance". Clearly no one ever has or will argue this. What have argued is his net influence wasn't all of the Celtics above averageness, and that multiple events occured between seasons which you can't control for. It's insulting that you make and knock down straw men in order to try to "win" an argument/debate on forum.

I chose not to follow up on what you apparently wanted me to because I didn't find it relevent. My position was (a) Russell was absent for a third of the season, (b) Boston were very good without him, (c) this does not fit with your suggestion that "What I am stating, however, is that it is very likely that Bill Russell's +/- profile fully "explains" the Celtics' performance in each and every year of the dynasty."

You then appeared surprised to learn that Russell missed a substantial portion of that season, "This is true (as I discovered, owing to your intervention"), and yet try to shift the debate to an argument about Russell's defensive rather than net impact, I debate in which I am uninterested. It certainly isn't relevent to the discussion taking place, which is about Russell's net impact, not his impact on a specific end.

As for the Sam Jones points, my main response is that you have ignored my points regarding Bailey Howell's decline and a change of focus from contention to player development being important aspects of the dropoff and are thus again fighting straw men in trying to say "Well it wasn't Sam Jones who helped the defense."
Incidentally you said you believed Sam Jones was a good defender before.

as for Sam Jones, my understanding is that he was a very good defensive guard

Why back away from that now by (a) using Drtg, hardly a flawless form of defensive measure, (b) using analogy which is hardly strong evidence, and indeed assuming you don’t believe teams should play 5 centers, would suggest Jones being “a very good defensive guard” would be a substantial upgrade on what other teams had at the position (and thus be an example of Bill Russell having a good supporting cast). Do you believe he was good on D or not? It is very much peripheral to the question at hand but it might serve you well to have a consistent position.

As for RAPM, I'm not as au fait as I'd wish to be with it. All I was saying above is that it's predictive powers seem to have gone somewhat astray in the case of the impact of Ben Wallace. If I had to venture an reason I'd suggest maybe extreme players cause it problems. Or perhaps it is only good at describing contextual impact rather than absolute value (though that would only explain why they didn't fall off as much as expected, not how they got better without him and in any case would But I don't know (x)RAPM well enough. If it is accurate I'd be surprised as to why no team has built around defensive big men of the type that are cheaply available (Bo Outlaw, Jason Collins) or the sort that teams tried to get rid of because of big contracts (Ostertag, Bradley, Ratliff).

Still I would have been interesting to have the debate, and your argument at least has the internal consistency that defense is undervalued across the board thus Ben Wallace or Mutombo comparisons aren't shrunk from, and presumably defensive specialists across the board would be rated higher. That would have some merit over those that argue that Russell is the greatest but wouldn't put any other defensive specialists (i.e. players whose value is derived almost wholly from D) anywhere particularly high. Unfortunately it won't be with you as you persistently misrepresent my arguments to score cheap points.

Mike G wrote:

Elvin Hayes only once shot over 50% from the field for a season.
...
He really excelled at playing big minutes and shooting a lot. He played on a team that won a title.

Let's not forget that he was also a premier rebounder of his day.
And he was a shotblocker, so we might guess he was something of a defensive stopper. Twice all-D, in fact.

Among the 50 on the current ballot, he's the 14th best rebounder. Of the 13 who are higher, he's about in the middle in scoring proficiency.
Hayes' block rate is below those of Ewing, Dwight, and Artis, but doesn't include his first 5 seasons.

Check out Hayes, Lanier, Moses, and Parish: http://bkref.com/tiny/ySLqW
Advanced stats don't love Elvin. But look at their playoff numbers, and he quite holds his own.
Excelling in playoffs is generally the opposite of "choking".

Elvin Hayes pretty much exemplified the NBA PF in the '70s. Look at most WS in a season by F and F-C (basically forwards who did not play guard) from 1969 to 1979.
http://bkref.com/tiny/aWigv

Appearances in the top 100:
11 - Hayes
6 - Barry*
5 - Hairston, Hawkins*, J Lucas, Nelson, Silas, Tomjanovich
4 - Cunningham*, Haywood*, McGinnis*
3 - Mickey Johnson, Bobby Jones*, Kauffman, Kenon*, Love, McMillian,
2 - Baylor, J Green, Mix, Wicks

This is NBA only, and ABA vets are starred. In the NBA, no PF has half as many top seasons as Hayes in the interval.
And WS doesn't even 'like' Hayes.

He was a good rebounder, sure, just not a special one, I was talking about the things that he was exceptional at (and that was playing a lot and shooting a lot). For defense it's hard to get a really accurate gauge on how good he was. He got two 2nd team all-D but sometimes such accolades go to stars and or those with good stats, it could be argued that competition at forward (especially if looking at/voting for a power forward) was weak after DeBusschere (though Silas had a defensive rep, but not the stats). In the earlier years the reviews seem to indicate he had the ability but not always the effort/consistency. Later (1976) it is noted that "Hayes has become tenacious" (contrasted with Unseld, who "always was") in that department. The Hollander handbooks never lead the team defense section with him or give any indication that he was an anchor, though he does seem to have been a positive.

Regarding his playoff performance, I would concur that the choker label is probably unfair which is why said "percieved". That said it's still hardly remarkable. Apart from Parish whose playoff numbers are alarming (I wasn't aware of this), he compares unfavourably in career playoff numbers to Lanier (whose numbers are 3/5ths from Milwaukee) and Moses.

Hayes was certainly consistent (and consistently good-but-not-great if you value scoring as PER does). Win Shares may not 'like' Hayes but they don't dislike him. As used here they reward him greatly for playing big minutes. If talking WS/48 then yes, it's not a fan. Even then though he does fairly well out of DWS (you know, the less trusted part). It's OWS (and OWS/48), which account for less than half of his career win shares (I wonder how many of our prospective top 50 or 100 we could say that for) that finds him wanting.

And we've already been through the lack of NBA power forward competition in the 70s.

So yeah Hayes was good, but he wasn't ever special. I just wonder how much he moved the needle in terms of a team's title odds.

Regarding his playoff performance, I would concur that the choker label is probably unfair which is why said "percieved". That said it's still hardly remarkable. Apart from Parish whose playoff numbers are alarming (I wasn't aware of this), he compares unfavourably in career playoff numbers to Lanier (whose numbers are 3/5ths from Milwaukee) and Moses.

Lanier was great but not so durable. Moses and Parish hung on forever with reduced minutes. Hayes averaged 38.4 minutes for his career vs 34 or less for these others.

In RS, Lanier and Moses were at .175 WS/48, Hayes .116
In PO, Hayes rises to .135 while the others maintained their .175
So I was really pointing out that Big E did not have a history of playoff flops. When you're the main man (unlike Parish) people notice these things.

Hayes shot better in playoffs (.501 vs .491 TS%) . While this is still not great, it closes the gap somewhat vs those others.
In PO, he becomes the best shotblocker of the bunch, and those guys were strictly centers. Only Moses is the better rebounder, but he was better than everyone.

Summary of those 5 players who look like they may or may not be elected to our top 25 - Regular seasons on the left, playoffs on the right http://bkref.com/tiny/Nb3Mk
About 1/4 of Barry's RS minutes were in the ABA, and about 1/3 of his playoffs. His PER and WS/48 were much higher in ABA.

It's always an open question, how much a player's postseason should count, relative to regular seasons.
Here are playoff winshares multiplied by increasing numbers, then added to RS winshares:
In 'straight' combined WS (column 1), Moses has a healthy lead over everyone.
With a playoff multiplier > 6, Pippen takes the lead, and Frazier passes Barry and Ewing.

Finally, there's the value of championships. Pippen 6, Frazier 2, Barry 1, Moses 1, Ewing 0

MW00 wrote:Use my quotes in context or not at all. Boston "probably being better with Russell is made in response to a post, which by chopping my response (without said context) and answering it, makes it look as though I'm saying Russell's career impact, rather than on that first season, is marginal. You further do so by saying they were "good enough to get in the playoffs and lose...sure" when, again the discussion laid out is clearly within the context of the '57 Celtic team which was way out ahead of any other team, prior to Russell's arrival.

Apologies. With the exception of the first two snippets (which was just for fun - sorry) I actually wasn't attempting to misrepresent your views in any way, and in fact - not that I am precisely sure what your views are - I don't think these remarks can be terribly far off the mark, given the factual context that I have been at pains to describe. This is to say that if Bill Russell's is not the player I (approximately) estimate him to be, but more along the lines of where I see your argument going, he simply cannot have been a really good player (in the context of "Best of..." lists.)

Anyway, more to the direct point, I knew you were referring to a notional marginal effect of his arrival, and my substantive point in reply was that this simply wasn't true (and without Russell, in 1957, they would have been good enough to get in the playoffs and lose, just like 1956.) And then I illustrate the point with calculations that the defense (what Russell was reputed to have brought) improved to the tune of about 3 points per 100 possessions once he arrived (again, in the teeth of seasonally-improving offenses, hence biased low in terms of what a full year average would be expected to be). And if one makes the parsimonious assumption that his addition to the roster was the only relevant change, then the fact that he played only 72% of the minutes, this implies that he entered the NBA providing a +4 on defense. And such a performance level is quite simply not marginal in any way shape or form, never mind from a rookie. (Then as for the apparent offensive deterioration upon his arrival, that is a story consistent with lore that he wasn't a very good offensive player.)

MW00 wrote:This is not the first time that you have (deliberately?) misrepresented my position earlier you posited that my argument was that "Coincident with Russell's arrival and departure, the Celtics took a discrete and large step up and down in defensive performance". Clearly no one ever has or will argue this. What have argued is his net influence wasn't all of the Celtics above averageness, and that multiple events occured between seasons which you can't control for. It's insulting that you make and knock down straw men in order to try to "win" an argument/debate on forum.

At a loss on this one. I was not attributing my view (what is in quotes) to you. That was represented as my point, because...well...it is my point. See above.

And next you succinctly state the (your) position that I think is untenable, unless one is prepared to attach all sorts of inelegant epicycles to the story. Mine is a simple story. If Russell was the great defender he is reputed to have been, we know what that must have looked like and what he must have brought to the table. And it so happens that in 13 years of +/- data there is one clear example (and I am sure if it were able to be extended back a few more years one or another would also be identified.) So, the point is not to fixate on Ben Wallace; the point is to understand what the effect of a great defender (and irrelevant offensive player) would be and how much of actual Celtics performance this would explain. And, what turns out to be the case is that such a hypothetical essentially explains everything, both in levels and trajectory.

Is there a missing fact that might make the story more persuasive? I've already related that the Celtics by any reasonable measure were offensively average and defensively great. But it is also the fact that in so being the Russell Dynasty was not historically strong on an annual basis. As I interpret the data, the Russell Celtics, on average, were worse than the average NBA champion (for the 1974 plus years, when the ORtg an DRtg are officially listed). So, the argument returns, if the Russell Celtics were defined by average offense and great defense but on net they weren't historically great, what alternative stories can you tell to Russell's profile "explaining" the Celtics' historical performance?

There are basically three options, for the adding-up constraint to hold, and it must hold:

(1) We're both right! Russell is as I describe, but there were good Other Celtics as well. But this then necessitates that there were also some Other Unspecified Celtics playing significant minutes who had to have been some of the worst players ever to lace up sneakers. Not plausible, right?

(2) What I perceive to be your story, where All Other Celtics were a net positive (either on defense or offense or both). This necessarily implies that Bill Russell was something akin to Marcus Camby, a very good defender (and player) but pretty much a "marginal" player, in terms of the topic of this larger conversation. And it also imposes a significant burden on the storyteller to explain why All Other Celtics who came in phases throughout the Russell years exhibited a defensive trajectory akin to that of an individual.

(3) Bill Russell was the Bill Russell of lore. There! That was easy.

MW00 wrote:I chose not to follow up on what you apparently wanted me to because I didn't find it relevent. My position was (a) Russell was absent for a third of the season, (b) Boston were very good without him, (c) this does not fit with your suggestion that "What I am stating, however, is that it is very likely that Bill Russell's +/- profile fully "explains" the Celtics' performance in each and every year of the dynasty."

What can I say at this point except that, of course, it is very relevant, but on we go.

MW00 wrote:You then appeared surprised to learn that Russell missed a substantial portion of that season, "This is true (as I discovered, owing to your intervention"), and yet try to shift the debate to an argument about Russell's defensive rather than net impact, I debate in which I am uninterested. It certainly isn't relevent to the discussion taking place, which is about Russell's net impact, not his impact on a specific end.

You very much misunderstand the point I was making. I absolutely knew that he had only played two thirds of the season. What I hadn't realized was that the win-loss record was better in his absence. But then again...I dealt with that very issue.

And also you are confused if you think I have been or am trying to shift the debate to an argument about Russell's defensive rather than net impact. And I suspect the confusion arose because it is almost certainly the case that his net impact was his defensive impact. Per the Ben Wallace analogy and per known box score stats, there is not much reason to believe that he had much of an impact on offense in a +/- sense. As such, his net impact is his defensive impact, and that is the story I was telling. (And if his impact on offense was significantly positive, then that just bolsters my larger point, right?) As such, you should be interested as it is not just of great relevance to the discussion taking place, it is the discussion taking place.

MW00 wrote:As for the Sam Jones points, my main response is that you have ignored my points regarding Bailey Howell's decline and a change of focus from contention to player development being important aspects of the dropoff and are thus again fighting straw men in trying to say "Well it wasn't Sam Jones who helped the defense."

This is only partially right. I did not ignore your remarks regarding Bailey Howell's decline. To the contrary, I looked at the facts, and you may be right that some significant fraction of the offensive decline post-Russell could be attributed to this. (I don't know and didn't bother to try and come up with a back of the envelope guess at this e.g. the drop in his TS% was also accompanied with a drop in minutes that were apparently taken by Don Nelson with a higher TS%, etc.) But the point, again, is that the big drop wasn't an offensive decline, it was a defensive decline, and that is what needs to be explained. And for that reason I paid attention to Sam Jones, given that his was the other departure, hence a more proximate explanation for the decline.

MW00 wrote:Incidentally you said you believed Sam Jones was a good defender before.

as for Sam Jones, my understanding is that he was a very good defensive guard

Why back away from that now by (a) using Drtg, hardly a flawless form of defensive measure, (b) using analogy which is hardly strong evidence, and indeed assuming you don’t believe teams should play 5 centers, would suggest Jones being “a very good defensive guard” would be a substantial upgrade on what other teams had at the position (and thus be an example of Bill Russell having a good supporting cast). Do you believe he was good on D or not? It is very much peripheral to the question at hand but it might serve you well to have a consistent position.

Ah, you are careful to note an apparent contradiction, and there is one. After having hit "submit" on my initial comment, I wished that instead I had used the word "impression", and even that is perhaps too strong. The impression in particular arose solely from an observation when monkeying around with the data. In particular, there was a bump up in DRtg by my estimates between 1960 and 1961, what coincided with an increase in Sam Jones' MPG (but what also coincided with a bump up in MPG for K.C. Jones and Bill Russell).

But you are otherwise rather very off base. (a) Defensive xRAPM is the acknowledged best defensive measure out there. And by a very long shot. (b) Reasoning by analogy is hardly strong evidence? Egads. Over 13 years, being not able to identify one aging, shooting guard with major scoring responsibility having made a positive contribution on defense? What actually only accords with common sense, there only being so much energy to dispense on the court. Don't know what to say about this.

But more generally, it is important to realize a structural feature about basketball in this regard. Players of the approximate stature of shooting guards are not key determinants of defensive success. That is a potential privilege of those who are tall and those protecting the paint (who for obvious reasons tend to be tall). This is just the way it is. And no, this doesn't mean that you can play five centers. It means that there is only so much a 6'4" guy can contribute. And when he is 35 and gassed from running around shooting the ball a lot, no, I don't think that such a player has much chance of being a net positive on defense.

MW00 wrote:As for RAPM, I'm not as au fait as I'd wish to be with it. All I was saying above is that it's predictive powers seem to have gone somewhat astray in the case of the impact of Ben Wallace. If I had to venture an reason I'd suggest maybe extreme players cause it problems. Or perhaps it is only good at describing contextual impact rather than absolute value (though that would only explain why they didn't fall off as much as expected, not how they got better without him and in any case would But I don't know (x)RAPM well enough. If it is accurate I'd be surprised as to why no team has built around defensive big men of the type that are cheaply available (Bo Outlaw, Jason Collins) or the sort that teams tried to get rid of because of big contracts (Ostertag, Bradley, Ratliff).

At once you say you're not terribly current but you're confident that Ben Wallace's xRAPM numbers must be off, never mind that the actual existence of Ben Wallace isn't necessary for the original argument. OK.

Then next you say you're surprised that no team has been built around defensive big men of the type that are cheaply available, but then again we've got the 2004 Pistons (though that building may not have been entirely conscious).

But more generally, this failure remains an enduring mystery.

... a really good player (in the context of "Best of..." lists.)...
...(3) Bill Russell was the Bill Russell of lore. There! That was easy.

At his retirement, he was a consensus alltime Top 5. Wilt and Oscar were widely believed to be better "players", though not greater "winners". Mikan, Cousy, Pettit, Baylor, and West had their supporters, too.

Fast forward 45 years, and is he still a de facto alltime Top 5 player? Well, that would be pretty extraordinary. A lot of greats have come and gone.

If the vote had been for the alltime Top 15 players, Russell would be a near unanimous pick. In an undifferentiated "Top 15", Russell fans should be satisfied.

People tend to have their 'favorite' guy, for better or for worse, amen: Oscar, Magic, Erving, Iverson, whomever, such that one just can't stand to see anyone ranked above them.

Rather than construct arguments designed to put your guy at the top, and convince yourself that it's infallible; maybe try the exercise of picking Any player from the list and making such an argument. It's entirely do-able.

Mike G wrote:Rather than construct arguments designed to put your guy at the top, and convince yourself that it's infallible; maybe try the exercise of picking Any player from the list and making such an argument. It's entirely do-able.

You talkin' to me?

If so, let me try to be even clearer. My spirited defense of Bill Russell isn't because he is my guy (nor because I tend to recoil at the anti-defense bias so prevalent here, there, and everywhere); it is because someone made the provocative suggestion that his reputation owed primarily to team success and that there was no evidence that he was the guy.

And at the end of my blathering on this topic, I even explicitly made the point, what should have made my perspective crystal clear:

schtevie wrote:Finally, a post script. This defense of the defense of Bill Russell is not simultaneously an endorsement of the view that he belongs in the top 5. For this, given the well-established competition, it needs to be shown that his contribution on offense was in fact positive. (A Ben Wallace-esque peak DRtg of 9.8 doesn't quite cut it.) And here, I don't think that there is any highly determinative evidence. Yes, he was a better scorer than Ben Wallace, relative to the centers of his day, and perhaps the best informed view is that this (plus all the other little things) would be sufficient to put him in +10 territory for several years (what is the cut-off in +/- criteria, anyway). But maybe not.

And I could have gone a bit farther. Having a notional, modern-day, statistical composite Russell pass the http://www.basketball-reference.com/ and stats-for-the-nba.appspot.com/ gauntlet, akin to the exercise undertaken for Sam Jones, we also find that there is no modern (2001 to 2013) precedent for a Bill Russell-type offensive player having positive value (i.e. offensive xRAPM >0).

And what this implies is that unless Bill Russell over-performed the entire Celtics defense (i.e. the Other Celtics were net negative contributors on defense) to the tune of at least +1 or unless his contributions to "all the other little offensive things" has no modern-day precedent, he cannot be thought of as having had multiple seasons of +10 net impact and hence cannot realistically be thought of as a top 5 player, especially given that the 2000s already identify three such players.

As for undertaking this same type of exercise for most of the other contenders for "Best Of" (centers in particular) I in fact did so, following the exact same methodology, in trying to better understand Bill Russell's contributions. The Russell story, however is one of the simpler ones and delivers the clearest message.