RealGM Top 100 List #34

[drza]

Just to confirm drza, are Kidd's post Vince APM stats just as good? Because I could see the argument that he meant a special amount to the Kmart teams because they needed an offensive creator and his transition ability so much, but if Kidd's numbers are as good and better than Pierce and Tmac's with Vince there and improved APM methods, that's pretty impressive to me. Probably not enough to make me put him over Pierce (I just think Pierce is flat out the better offensive player), but enough to make me consider Kidd a top 50 guy which I don't think I currently do

Carter arrived in Jersey early in the '05 season, playing 57 games with them.

The " 4 years" I've been quoting so often in this thread for Kidd are 2003, '04, '05 and '06. So yeah, it includes Carter years.

Also, for what it's worth, the "Case for Dennis Rodman" website that DocMJ linked to several threads back included that author's analyses of how a player's absence affected a) his team's scoring margin and b) his team's win percentage based on the games that he missed in seasons. His calculations are very similar to the In/Out data that ElGee calculates, but it is done on every "eligible" player since 1986. For the scoring margin, the author used "at least 15 games both played and missed for a team in the same season" and at least 3000 missed minutes total in qualifying games to determine an eligible season (there were 164 total players on his list since 1986). He lowered the threshold a bit for win %, requiring still at least 15 games played/missed in a season to qualify but only 1000 missed minutes among qualifying games (470 eligible players since 1986).

Kidd had 7 "qualifying seasons" and measured out 9th among the 164 players in scoring margin difference ( http://skepticalsports.com/?p=1143 ), 9th out of 470 in win% difference ( http://skepticalsports.com/?p=1181 ), and 4th overall in the author's summary that seeked to combine the two methods ( http://skepticalsports.com/?p=1214 ). The rest of the author's summary top-5 were Rodman, Shaq, Barkley and Kobe.

I would expect everyone to take this with whatever grain (or boulder) of salt that they normally would such data, but I include it specifically in counter ElGee using his in/out data for offense only to suggest that Kidd's absences weren't so impactful to the team ORTG (so I suppose I really should include this in a quote-rebuttal to him, but for convenience I'm leaving it here and hopefully ElGee sees it). It ties back to the same story we see when looking at the team ratings in general...the ORTGs don't change much but the defense with Kidd looks much better according to the RTGs, but overall the team plays much better with Kidd than without him.

So, to me, something has to give. Every individual team-impact measure that I've seen suggests that Kidd had a huge impact. About the only rebuttal that I've seen is that, because the team impact usually appears on defense, it can't really be Kidd's impact. Yet, the individual +/- data suggests that Kidd was having one of the elite impacts in the league...but more offensive than defense. So either Kidd is some kind of one-in-a-billion statistical anamoly that always happen to luck up into teams and defenses that play better when he's around to make up for his poor offense that also the computations flukily think are very good...or else he had exactly the level of impact that my eye test and his accolades would have suggested, and that he really WAS having a huge impact, and that there is an alternative explanation (like the one that I offered on page 4 of this thread) for how his impact is being measured accurately overall but attributing it to offense/defense is just fuzzy.

[penbeast0]

Or his defense is very good and PGs do make a difference defensively

[Doctor MJ]

Doc, I'm not sure if you DR;TLed my post on the top of page 4 of the thread where I addressed you saying this before, but again, I'd be interested to see your rebuttal so I'll post it again:

There are two relevant multi-year APM studies that I am aware of for comparing Pierce and Kidd: Ilardi's 2004-09 study, and Englemann's "10 year ranking" (which I'm not sure is exactly 10 years, but is at least the most comprehensive long-range study he has). We could use Englemann's 05 - 10 or 06 - 11 studies as well, but since a) it's clear that Kidd's peak was earlier in the decade and b) the 2 shorter/later studies were done by the same person as the "10 year ranking" so presumably should be factored in, it seems to me that the Ilardi '04-09 and the Englemann "10 year" are the 2 to look at. And the results of these two studies are:

Ilardi: Pierce +6.5, Kidd +4.8 (note, this "difference" actually overlaps in standard error)

Englemann: Pierce +4.9, Kidd +4.7 (Englemann doesn't list the standard error, but I see no way these don't overlap)

So, the accurate statistical conclusion that one could make from these 2 multi-year APM studies is...that there is no statistical difference between Kidd and Pierce over these time periods. So, unless there are other multi-year APM studies that you were considering that I have not accounted for, the next step is to then go and look at the best shorter-term APM studies that are available to see if there is relevant info there. As I've alluded to in the past (and as I plan to post in a separate post here, time willing), Kidd consistently measured out higher than Pierce in the RAPM year-to-year data from 03 - 06, with Pierce finally catching him in '07 and passing him in '08.

Considering that, again, Kidd was in the league earlier and thus should be expected to decay sooner than Pierce, it seems very intuitive to me that Kidd being consistently higher than Pierce in RAPM during their early overlapping peaks with a Kidd decline while Pierce maintains would explain how both of the long-term APM studies (one ending in '09 and the other in '11) could have Pierce and Kidd fairly even over a long stretch. And thus, the fact that during their shared peaks Kidd was consistently measuring better tells me that, at least according to the best APM data available, Kidd peaked higher than Pierce. By a reasonable and repeatable degree. This is important to the way that I have ranking in this project based on "how great they were at playing the game of basketball" as my top criterion. And to come full circle, my analysis would seem to run exactly opposite to your statement that Pierce "beats Kidd fairly soundly"...if anything it looks to me like Kidd beats Pierce in APM pretty comfortably.

Good post.

One of the things I did, was get player rankings for various players in various many year studies. The ones I actually used were Ilardi's and then Engelmann's but from a version he's since tweaked. I haven't had time to really go back and review how much Engelmann's changed things, but when he came out with a 10-year study this struck me as not much more than a curiousity. When you make the sample size too big, then you simply aren't dealing with one player any more but many players. I like the half-decade-ish analysis because for the most part players don't change that much in a half decade (and when they do, it's obvious, and you know to take their APM less seriously).

So anyway, if we go by the '03-09 Ilardi model and the '05-011 Engelmann model I used, Pierce ranks 10th and 11th respectively, while Kidd ranks 29th and 25th respectively.

Now, the consistency encourages me, though I'll admit not all players have such consistent numbers, and you're right that if we allow for standard errors it's pretty easy to say that the gap between these two is not enough to have a confident difference.

More important is just the idea that in the metric I thought of Kidd as being well ahead, he's actually if anything behind. This doesn't make me say "Have to rank Pierce higher!", but it sure doesn't hurt his cause.

Re: Kidd older, so peaked earlier, decayed sooner.

This is very true, and a clear reason why one shouldn't say "Aha, so we know that Kidd isn't as good as Pierce".

Where I'm coming from is really looking into the years before APM where Kidd started getting accolade love and the fact that he seeming to get lucky in terms of having circumstances that make people give him credit that he probably shouldn't get. Timeline:

-Kidd goes to Phoenix, and in the first full year we see a major improvement with Phoenix winning 56 games. When his individual stats go up the following year, he leaps into MVP consideration and becomes essentially a lock for All-NBA 1st team for several years.

However that team improvement came with the offense getting worse and the defense becoming the team's strength, and the team actually regressed when Kidd took his statistical leap forward.

Also, strangely Kidd makes 1st team in '00 & '01 despite not being a top 2 guard by MVP voting. It seems clear that All-NBA voters like the idea of having a pure point guard on the 1st team. So they're giving him a strange nod here even considering that they've already raised his reputation based on team success on the back of defense.

-Of course then Kidd goes to New Jersey and becomes an official superstar, but as mentioned, Jersey's improvement is primarily on the defensive end, and Phoenix offense actually gets better with Marbury in place of Kidd.

None of this really matters if you're okay with saying Kidd's a defensive superstar...but I really don't think he is. When I factor that in, I think '98-99 might be the only year where I actually consider Kidd a Top 2 guard, and of course that was the strike year when everything was just thrown out of whack and in the short window between Jordan & Kobe.

Reflecting on all of this, I look back at my own RPOY Top 10s, and I find I'm kind of shocked that even as I was thinking of Kidd as being well ahead of Pierce, they look quite similar by that metric: Kidd has 5 Top 10 finishes, while Pierce has 4, despite the fact that Kidd peaked at a time where superstars were relatively few in number. Were I to go back and do it again, Kidd may even lose that edge.

Last, something I typically hold against players but have never really thought about tremendously regarding Kidd is the off court issues. Kidd changed teams twice leading up to his peak, both times because of his bad behavior. I am impressed when players stay with one team forever, and while I know there's some luck involved, this also doesn't hurt Pierce here.

[lorak]

Kidd several times missed more than 10 games during his prime, so we could look at his with/without numbers (team ortg and drtg).

Code: Select all

ortg   drtg   net   MOV   G missed   year   
+1,6   -0,3   +1,9   -5,8   60   1997   Kidd-DAL
+9,1   +2,4   +6,7   +4,7   49   1997   Kidd-PHO
+1,3   -0,9   +2,2   +5,9   15   2000   Kidd
-0,7   -4,7   +4,0   +3,6   15   2004   Kidd
+9,0   -1,0   +10   +0,3   16   2005   Kidd


(drtg with "minus" is good thing)

So we see small impact on defense, but sometimes great on offense. However for example in 2000 Kidd led league in assists and improved Suns offense only by +1.3 ortg.

[drza]

Good post.

One of the things I did, was get player rankings for various players in various many year studies. The ones I actually used were Ilardi's and then Engelmann's but from a version he's since tweaked. I haven't had time to really go back and review how much Engelmann's changed things, but when he came out with a 10-year study this struck me as not much more than a curiousity. When you make the sample size too big, then you simply aren't dealing with one player any more but many players. I like the half-decade-ish analysis because for the most part players don't change that much in a half decade (and when they do, it's obvious, and you know to take their APM less seriously).

So anyway, if we go by the '03-09 Ilardi model and the '05-011 Engelmann model I used, Pierce ranks 10th and 11th respectively, while Kidd ranks 29th and 25th respectively.

Now, the consistency encourages me, though I'll admit not all players have such consistent numbers, and you're right that if we allow for standard errors it's pretty easy to say that the gap between these two is not enough to have a confident difference.

More important is just the idea that in the metric I thought of Kidd as being well ahead, he's actually if anything behind. This doesn't make me say "Have to rank Pierce higher!", but it sure doesn't hurt his cause.

Re: Kidd older, so peaked earlier, decayed sooner.

This is very true, and a clear reason why one shouldn't say "Aha, so we know that Kidd isn't as good as Pierce".

Where I'm coming from is really looking into the years before APM where Kidd started getting accolade love and the fact that he seeming to get lucky in terms of having circumstances that make people give him credit that he probably shouldn't get. Timeline:

-Kidd goes to Phoenix, and in the first full year we see a major improvement with Phoenix winning 56 games. When his individual stats go up the following year, he leaps into MVP consideration and becomes essentially a lock for All-NBA 1st team for several years.

However that team improvement came with the offense getting worse and the defense becoming the team's strength, and the team actually regressed when Kidd took his statistical leap forward.

Also, strangely Kidd makes 1st team in '00 & '01 despite not being a top 2 guard by MVP voting. It seems clear that All-NBA voters like the idea of having a pure point guard on the 1st team. So they're giving him a strange nod here even considering that they've already raised his reputation based on team success on the back of defense.

-Of course then Kidd goes to New Jersey and becomes an official superstar, but as mentioned, Jersey's improvement is primarily on the defensive end, and Phoenix offense actually gets better with Marbury in place of Kidd.

None of this really matters if you're okay with saying Kidd's a defensive superstar...but I really don't think he is. When I factor that in, I think '98-99 might be the only year where I actually consider Kidd a Top 2 guard, and of course that was the strike year when everything was just thrown out of whack and in the short window between Jordan & Kobe.

Reflecting on all of this, I look back at my own RPOY Top 10s, and I find I'm kind of shocked that even as I was thinking of Kidd as being well ahead of Pierce, they look quite similar by that metric: Kidd has 5 Top 10 finishes, while Pierce has 4, despite the fact that Kidd peaked at a time where superstars were relatively few in number. Were I to go back and do it again, Kidd may even lose that edge.

Last, something I typically hold against players but have never really thought about tremendously regarding Kidd is the off court issues. Kidd changed teams twice leading up to his peak, both times because of his bad behavior. I am impressed when players stay with one team forever, and while I know there's some luck involved, this also doesn't hurt Pierce here.

I guess, when we see the same thing happen repeatedly for over a decade it seems hard to chalk it up to some strange coincidence happening repeatedly.

In 1997 Kidd goes to a 40-win Suns team with a poor defense, and in 1998 they win 56 games with a strong defensive rating. They proceed to win 50+ games in every non-strike shortened season between 1998 and 2001 with Kidd at the helm, all of them on teams sporting good defenses. Kidd leaves for the 2002 season, and suddenly the Suns revert back to a 36 win team with poor defense.

In 2001 the Nets win 26 games with a bad defense. In 2002 Kidd arrives, and the Nets win 52 games with a great defense.

In 2003 we get access to +/- data, and lo and behold Kidd measures out as one of the better players in the NBA in 2003. And 2004. And 2005. And 2006. In 2007 Kidd's RAPM starts to slide a bit, and the team record falls to .500 for the first time in his tenure (this is the first time Pierce equals Kidd in RAPM). In the 2008 season that ends with Kidd 35 years old, Pierce finally beats Kidd by a reasonable amount in the yearly RAPM and the struggling Nets trade Kidd away to be a subordinate piece in Dallas.

Also, in the 7 seasons that Kidd both plays and misses at least 15 games for a team in the same season, his impact on both the team's scoring differential and win % are among the very best in the NBA among candidates over the last 25 years.

At what point does this stop being a coincidence and start indicating that Kidd was really having a huge individual impact? Granted when Kidd moved from team-to-team he wasn't necessarily the only change, but when the same pattern replicates itself over and over and over again for a full decade across multiple teams and with 3 different ever-more-specific ways to measure impact...I mean, at some point can't we just call a spade a spade? How can ALL of that be narrative based on winning bias? When the common denominator in all of the winning, to a very specific degree, seems to be Kidd himself...isn't that an indication that HE is the winning bias? I know his impact is unorthodox. But how can anyone look at him having the same positive impact everywhere he goes over and over again and just conclude that its some bizarre repetitive fluke?

[Doctor MJ]

At what point does this stop being a coincidence and start indicating that Kidd was really having a huge individual impact?

You're overestimating the size of the trend and ignoring a glaring counterpoint.

1) Twice Kidd changed teams and we saw clear improvements in his first full season with the new teams.
2) Both times, the key to the improvement was defense.
3) Kidd's primary +/- impact according to stats was offense.
4) In many year league-wide data, we've yet to see any point guard have the kind of defensive impact to justify the improvements we saw with Kidd's 2 team changes.

So all we've got are two general data points supporting the Kidd rainmaker narrative, but the specific data points about Kidd go against that narrative, and the more reliable league-wide data goes against the narrative even more strongly.

By far the most likely scenario is that Kidd got flipped heads on a coin twice, and everyone ate it up because people (myself included) tend to want to believe that a high IQ player who racks up assists has intangible impact well beyond the stats.

[drza]

At what point does this stop being a coincidence and start indicating that Kidd was really having a huge individual impact?

You're overestimating the size of the trend and ignoring a glaring counterpoint.

1) Twice Kidd changed teams and we saw clear improvements in his first full season with the new teams.
2) Both times, the key to the improvement was defense.
3) Kidd's primary +/- impact according to stats was offense.
4) In many year league-wide data, we've yet to see any point guard have the kind of defensive impact to justify the improvements we saw with Kidd's 2 team changes.

So all we've got are two general data points supporting the Kidd rainmaker narrative, but the specific data points about Kidd go against that narrative, and the more reliable league-wide data goes against the narrative even more strongly.

By far the most likely scenario is that Kidd got flipped heads on a coin twice, and everyone ate it up because people (myself included) tend to want to believe that a high IQ player who racks up assists has intangible impact well beyond the stats.

But how does that in any way address his high in-season in/out winning percentage and scoring differential impact? That come from 7 different qualifying situations? That measures out as among the best in the last 25 years?

And how does that in any way address him having such high RAPM values in 03, 04, 05 and 06 consecutively?

He didn't just flip heads twice and have to be explained. He flipped heads in every year that we have those stats for, then the "rainmaker" years happened to be the 2 times that the in-team impact could be tested going from one team to another...and in both cases he aced the exam. What do you know, he actually made it rain.

This is a really weird situation, as usually I can at least see where you and ElGee are coming from even if I disagree. Here, I really don't get your logic. It's like you just decided that his accolades must have been undeserved, and we should just ignore that in every way we can measure it he was one of the highest impact players of his generation. I don't get it.

[lorak]

But how does that in any way address his high in-season in/out winning percentage and scoring differential impact? That come from 7 different qualifying situations? That measures out as among the best in the last 25 years?

According to with/without Kidd's impact on defense was small and on offense sometimes great and sometimes also small. Overall his net result isn't anything special if we compare him with other superstars.

[Laimbeer]

Just got back from the weekend, and I'm pretty surprised Cousy still isn't nominated. Anybody got a vote count handy?

[ElGee]

At what point does this stop being a coincidence and start indicating that Kidd was really having a huge individual impact?

You're overestimating the size of the trend and ignoring a glaring counterpoint.

1) Twice Kidd changed teams and we saw clear improvements in his first full season with the new teams.
2) Both times, the key to the improvement was defense.
3) Kidd's primary +/- impact according to stats was offense.
4) In many year league-wide data, we've yet to see any point guard have the kind of defensive impact to justify the improvements we saw with Kidd's 2 team changes.

So all we've got are two general data points supporting the Kidd rainmaker narrative, but the specific data points about Kidd go against that narrative, and the more reliable league-wide data goes against the narrative even more strongly.

By far the most likely scenario is that Kidd got flipped heads on a coin twice, and everyone ate it up because people (myself included) tend to want to believe that a high IQ player who racks up assists has intangible impact well beyond the stats.

But how does that in any way address his high in-season in/out winning percentage and scoring differential impact? That come from 7 different qualifying situations? That measures out as among the best in the last 25 years?

And how does that in any way address him having such high RAPM values in 03, 04, 05 and 06 consecutively?

He didn't just flip heads twice and have to be explained. He flipped heads in every year that we have those stats for, then the "rainmaker" years happened to be the 2 times that the in-team impact could be tested going from one team to another...and in both cases he aced the exam. What do you know, he actually made it rain.

This is a really weird situation, as usually I can at least see where you and ElGee are coming from even if I disagree. Here, I really don't get your logic. It's like you just decided that his accolades must have been undeserved, and we should just ignore that in every way we can measure it he was one of the highest impact players of his generation. I don't get it.

I'll respond to your other post in a second, but I wanted to say I had the *exact* same thought as Doc, who summed it up nicely. On one hand you point to +/- numbers to tout Kidd -- those suggest a good offensive player. Then you look at his team's success, which is defensive, and tout Kidd. You can't have it both ways -- the information doesn't line up.

I can't speak for Doc, but I think Jason Kidd was a very good player. But that doesn't mean Garnett, Walton or Russell good (yikes!). For perspective, here are my rough RPOY standings for Kidd for his best years:

99 6th
00 Top-10
01 Top-11
02 Top-10
03 Top-8

He was never a top-5 guy to me...and I do hold him as one of the best defensive PG's of all-time. But I just never thought he was a great offensive player. He was great to play *with* and made the right pass, and was a QB, but so was Brevin Knight and now Rajon Rondo (and KIdd is better than both). That's why he's arguably a top-50 player of all-time and one of the 10 best players of his decade. But he's not this good, and I don't really see how any of the information presented here suggests he is.

[therealbig3]

My count:

Vote:

Gilmore-5 (penbeast0, mysticbb, Doctor MJ, lukekarts, drza)

Gervin-3 (ElGee, DavidStern, RoyceDa59)

Pierce-3 (therealbig3, Fencer reregistered, Dr Mufasa)

Wilkins-1 (JordansBulls)



Nominate:

Paul-3 (ElGee, therealbig3, Doctor MJ)

Cowens-2 (JordansBulls, Dr Mufasa)

Moncrief-1 (penbeast0)

Miller-1 (mysticbb)

Cousy-1 (Fencer reregistered)

Iverson-1 (DavidStern)

Walton-1 (RoyceDa59)

Mourning-1 (drza)

[Laimbeer]

Gervin is the most deserving of that lot. I prefer Cousy, but Cowens is the one with a chance against Chris Paul(!) for Gawd's sake.

Vote: Gervin
Nominate: Cowens

[ElGee]

No one has still addressed my issues with early RAPM. Ridge regression is a squeezing technique -- it will lump the population closer together. The math being used in the early single-year RAPM, ignoring that the one-year sample size is always an issue, seems to be squeezing the whole league into a tight pocket.

Am I to believe that in the middle of the decade, individual impact suddenly became much better according to Englemann? The 2002 leader was Shaq at +3.4 (peak Shaq, fwiw), Doug Christie (3.2, eek) and peak Duncan (3.0). In 2006, they would be ~ the 35th best players in the league. No one finds a problem with this??

Finally, I've always said this about the +/- family of stats, but how good the team is needs to be accounted for as well. I've written about it in detail, and I just don't see any way around the fact that both empirically and logically it's easier to improve a bad team by a bunch than a good team by the same amount. As far as I can tell, no APM/RAPM number adjusts for that, so when you see Tim Duncan and Elton Brand have the same number, but one makes his team +10 and the other +3, I don't see those as equatable values.

I've mentioned it before, but I don't think your oft-repeated bolded stance is true. I think you are somewhat talking about a "diminishing returns" effect, but that you're applying it incorrectly. The diminishing returns aren't so much a factor of "the team is as good as it can be so can't be improved more" as much as it is a factor of "this player is contributing something to the pot that none of his other teammates can". Which isn't the same thing.

For example, the 2011 Heat were not optimized to the greatest extent possible based upon the talent upon the team. Logically, the Wade/Bosh-led cast was better (with LeBron removed) than any of the Cavs casts without Bron. Yet, the Heat actually won fewer games than the last couple of LeBron-led Cavs, and LeBron's APM was lower. For the diminishing returns concept to be based purely on how good your team is, this couldn't be unless LeBron suddenly just got a lot worse. But when you look at it through the lens of redundancy, we see how the ways that LeBron could be maximized overlapped in large part with the way that Wade (and to a lesser extent Bosh) needed to be maximized. As such, LeBron's value wasn't purely additive because he was bringing things to the table that others already did.

On the flip side, Nash joined a Suns team in '05 with a lot of talent on it led by Stoudemire, Marion and Johnson and shot the team up a bunch of notches because his skill set WASN'T replicated by his talented teammates. When you put his skillset together with theirs, it was a strongly additive effect, more so than what we saw with LeBron this year.

A player's ability to be strongly additive regardless of the team situation is definitely a plus for that player, but it isn't easily quantified by the data we have now. Likewise, it is fair to talk about that as a potential weakness in APM to the extent that with players on strong teams with redundant talent it may be difficult in the short term to separate that player from his teammates. But this just reduces to the "colinearity" issue with APM studies that is a well-vetted issue that is best addressed by looking at multiple years at a time.

In the specific case of Kidd, as I've pointed out several times, we don't HAVE a multi-year APM study that covers his peak without also covering his decline. Thus, the best that we have to look at are multiple single-season studies taken during his peak with the best single-season APM studies that we have access to. But the fact that Kidd consistently measures out among the better in the league (and better than Pierce by a solid margin) for four consecutive years before he obviously started on an age/injury downside in his mid-30s, to me, is very convincing. I look at that, and see no way that a multi-year 2003 - 2007 APM study wouldn't conclude that Kidd had a clearly higher value than Pierce over that period.

I'll let someone else tackle the issue of Englemann's compressed years...when it was brought up before, someone (probably Mysticbb) explained that the year-to-year compression was due to him choosing different lamdas for different seasons and that the take-home was to analyze differences within-seasons more-so than trying to compare across seasons. If that's not enough to satisfy you on that front, I'll have to leave it to Mystic (or someone) to address that for you.

But as for the improving from bad to average being harder than from average to great, despite how many times you've repeated it I still see nothing to support your stance. At least not nearly to the disqualifying degree that you've been attempting to use it. And I'm really not sure about the Duncan/Brand example you gave, as my quick perusal had Duncan comfortably ahead of Brand in each individual year between 2003 and 2011 in RAPM as well as in both the Ilardi 03 - 09 and the Englemann "10 year APM" studies...so as far as I can tell, the RAPM studies we've been heavily discussing agree with you.

Yeah, and you didn't address it before when I explained it in detail and provided an example (and there are countless examples). Many players can move a team from 0.95 to 1.10. Basically no one has ever moved a team from 1.10 to 1.25 (pts/pos). That would have to be one heck of a coincidence

Call it diminishing returns, but it's more than that because it doesn't take much to get an offense respectably over a point per possession in the NBA. You have to have a total paucity of offensive talent and a non-offensive coach basically to lurk in that putrid territory. And I've explained why -- one ball, 5 players, and rules totally catered to the offense.

But as easy, in one sense, as it is to put pieces in place to make teams respectable on offense, it becomes equally challenging to get them elite because, well, there's 5 players and one ball. You have to add the right parts, as you point to, and those parts have to interact properly. That's hard...unless you have offensive geniuses like Magic Johnson and Steve Nash.

Funny you mention the Heat first, since they had a 112.7 ORtg w/James on the court. Let's not act like that's bad -- we were just hoping with the volume of superstars they would hit 115+. Hasn't happened yet. Your description of roles/additiveness is spot on, and that's the crux of the issue with getting 99% of NBA teams from say, 113 ORtg to 120 ORtg. A 106 team would love a shooter to space the floor, and as such see efficiency steadily improve (eg Ray Allen). Adding a shooter, probably redundantly, to a team already scoring 113 points won't just push them to 120! We literally have almost no evidence of this and just a boatload of evidence confirming the former.

Nash is who he is because you can put so many different combinations of foursomes together with him and he makes the offense so good (by making shots for these players so easy). In 2006, Phoenix was 106.4 ORtg without Nash and 114.8 with him. A 115 ORtg with those parts is pretty staggering, especially when you say "let's give him a mediocre spot-up 2-guard, a slasher/defensive 3, a PnR offensive star at the 4 and a lumbering center who can make an open 15-footer...and the ORtg of that team (including bench permutations) with Nash will hit 119!

Take Dirk, the great flavor of the year with his sexy spacing and whatnot. In 2004, he was +6.2 (115.6 on). McGrady, in Orlando was +8.1 (107.2). Do you think what McGrady did was 31% better?

Duncan and Brand were made-up numbers. Didn't mean to suggest otherwise. I've been looking at 82games on/off data for years now and don't see where you're coming from. Quick, name someone who lifted a sub-100 offense by 10 points. Now name someone that lifted a 110+ offense by 10 points.

(Vlad Rad in 08 comes close...I can't think of another.)

[ElGee]

Well I still just can't get behind the 1-2 year superprimes in Walton, Tmac, Paul and the argument that you have a better chance of winning a title with 1 or 2 years as a top 3 player than 10 years as a top 10 one.

I think it's very important to have at least like... 5 years. To win a title you need a lot to go right... you need teammates to fit, chemistry to fit, luck in the playoffs, health, to be able to avoid an unbeatable competitor (great example: longevity allowed Olajuwon to have bad luck by facing a GOAT team in the 86 Finals and having his teammate's career collapse - but then have luck to get a Jordan retirement in 94 and 95), you need a team who can win without taking lumps and experience hits in the playoffs first (I would say the 08 Hornets and 09 Cavs are GREAT examples of this), etc.

Yes, Walton was able to beat the odds. He also wouldn't have done it without landing Maurice Lucas on his team from the start, or a relatively weak competition in 77, or the Sixers kind of self combusting.

I think there's a lot of proof in the pudding that disproves the Walton theory. You've got Paul and Tmac themselves, peaking at best in the league level and not even getting within a stone's throw of a title team. Bernard King had a crazy year and didn't come close. You've got Lebron having a first 8 years as good as anyone's in the regular season and going titleless. KG repeated the same in Minnesota. You've got Jordan playing 6 magnificent years and not winning on. Kareem went from 72 to 79 as by far the best player in the league and didn't get one. Wilt, Oscar, West lost a lot.

Likewise, you CAN win a title or make the finals with Paul Pierce, Ray Allen, Reggie Miller, Alonzo Mourning, Willis Reed, Dave Cowens, Pau Gasol, Jason Kidd. Because all of them either have a title or have been closer than Tmac and Paul. They were all closer than freaking 87 and 88 Jordan or 2009 and 2010 Lebron, much superior players.

I think if you take Tmac or Paul over contemporaries, you're putting a ton into the hands of LUCK. That everything will come together just for that year or two and not the ones where they aren't there. Yeah it's hard to get a player who's best on a title team caliber like those guys were, but it's also hard to get a supporting cast good enough for them, too. Just mentioning Paul first, I don't think he's proven that with merely good players he can win a title. The 08 and 09 teams were pretty solid by my standards. As good as 77 Walton's? Not that far off. Tmac had his 05 and 07 good teams... it wasn't his superprime, but how much worse was he that it'd make the difference between Round 1 knockout and title caliber with that team? Even the GOATs can't pull a 77 Walton or 94 Olajuwon or 11 Dirk on a dime. Not even Walton or Olajuwon or Dirk most of the time

There are very few shortcuts in the league IMO. The majority of title winners and Finals teams had best players in the 2nd half of their careers, who'd already seen close to it all in the playoffs. Even the teams that won with under 25 stars like Wade, Duncan, Magic, Kareem, Russell had someone else on the team that'd been through it: Shaq, Robinson, Kareem, Oscar, Cousy on those teams. I think most of the time if you take the Tmac and Paul route of a young star beginning his career with a franchise and having that early peak only, you're going to lose. IMO. Virtually the only exception is Walton, who won right before the widely accepted weakest title teams (Seattle, Washington) in the druggy, ABA transition, parity league, against a chemistry nightmare team, where they fell down 0-2 in the Finals and just had everything go right from that point on and had help from the GOAT basketball Finals crowd. I just don't think you can dependably ape the 77 Blazers.

I don't understand why you're thinking about it this way. It sounds like a different form of the Rings argument. It's the Hypothetical Rings argument!

Players can only increase the chances a team wins a title. The more random team a player can push closer to a crown, the better. It's probabilistic, not dichotomous. There are also plenty of examples of new teams coming together and winning early (57 Celtics, 71 Bucks, 80 Lakers, 99 Spurs, just off the top of my head). Obviously, Paul, Walton and King would be better options than almost everyone else if they too were put on one of the best teams in the league...

[penbeast0]

I'm a bit iffy on that last idea. Walton could fit into almost any team, Paul not quite so much, Bernard King would not work well on a lot of good teams, especially those built on chemistry and defense . . . for that matter, Gervin, much as I loved him when he played and great as his impact was when the team was built around his skill set. He isn't a guy that you can just plug into the 2 guard spot on a team like this year's champion Dallas although that was probably their biggest hole. The impact might very well hurt Dallas rather than helping it (though having a PG like Kidd might make it work -- you couldn't add Gervin to Miami add that successfully either to use the other finalist).

[ElGee]

I'm a bit iffy on that last idea. Walton could fit into almost any team, Paul not quite so much, Bernard King would not work well on a lot of good teams, especially those built on chemistry and defense . . . for that matter, Gervin, much as I loved him when he played and great as his impact was when the team was built around his skill set. He isn't a guy that you can just plug into the 2 guard spot on a team like this year's champion Dallas although that was probably their biggest hole. The impact might very well hurt Dallas rather than helping it (though having a PG like Kidd might make it work -- you couldn't add Gervin to Miami add that successfully either to use the other finalist).

Wasn't trying to equate their level of play. I was saying that be it Walton, Paul or King, that they are so much better than almost everyone else (largely increasing the chance to win a title) is what matters.

I have no idea why you don't think Bernard King in 1984 wouldn't work well on a lot of teams. Same with Gervin...?? Simply because you can come up with radical examples doesn't change their average value to random teams.

[Doctor MJ]

At what point does this stop being a coincidence and start indicating that Kidd was really having a huge individual impact?

You're overestimating the size of the trend and ignoring a glaring counterpoint.

1) Twice Kidd changed teams and we saw clear improvements in his first full season with the new teams.
2) Both times, the key to the improvement was defense.
3) Kidd's primary +/- impact according to stats was offense.
4) In many year league-wide data, we've yet to see any point guard have the kind of defensive impact to justify the improvements we saw with Kidd's 2 team changes.

So all we've got are two general data points supporting the Kidd rainmaker narrative, but the specific data points about Kidd go against that narrative, and the more reliable league-wide data goes against the narrative even more strongly.

By far the most likely scenario is that Kidd got flipped heads on a coin twice, and everyone ate it up because people (myself included) tend to want to believe that a high IQ player who racks up assists has intangible impact well beyond the stats.

But how does that in any way address his high in-season in/out winning percentage and scoring differential impact? That come from 7 different qualifying situations? That measures out as among the best in the last 25 years?

And how does that in any way address him having such high RAPM values in 03, 04, 05 and 06 consecutively?

He didn't just flip heads twice and have to be explained. He flipped heads in every year that we have those stats for, then the "rainmaker" years happened to be the 2 times that the in-team impact could be tested going from one team to another...and in both cases he aced the exam. What do you know, he actually made it rain.

This is a really weird situation, as usually I can at least see where you and ElGee are coming from even if I disagree. Here, I really don't get your logic. It's like you just decided that his accolades must have been undeserved, and we should just ignore that in every way we can measure it he was one of the highest impact players of his generation. I don't get it.

Okay, good stuff to bring up.

First though: I'm with LG about the distinction between offense and defense. It doesn't make any kind of sense to me to credit a player with a team's defensive turnaround because his offensive RAPM looks good. Now I can imagine someone saying "I have much more faith in overall +/- metrics than offense vs defense, and so a turnaround plus strong individual overall +/- to me adds up", but the deal breaker to me is that we know again and again that point guards just don't have massive defensive +/- impact. To have Kidd's offense/defense split to come down exactly where we'd expect it to given his position, and to have that be inexplicably incorrect is just too much for me. That's simply getting into hand waving that might as well wave away +/- stats altogether.

But then you say "Okay, so it's just a total coincidence that Kidd puts up huge overall RAPM every year of his time in Jersey where he had a reputation as a rainmaker?" and you've got an excellent monkey wrench.

I won't claim to have everything worked out, but a key thing is that I just don't have that much confidence in Engelmann's one year numbers as true snapshots of what a player did in a given year.

Go and look at the Top 5's for each of his year. They are incredibly consistent. Typically of the 5 top guys one year, 4 will be in the top 5 the next year. That's crazy. You don't see that kind of consistency in box score stats, and let alone in the vanilla APM stats.

What he's doing is using data from outside of the year in question to help come up with a rating for the player for the year in question that has as little noise as possible. It's understandable why he does what does, but I think it's a mistake to look at each year's numbers as independent experiments.

What's also interesting to note is that we have Ilardi's 6 year APM numbers from '03-04 to '08-09, and now Engelmann's 1 year RAPM numbers for each of those years. How do they compare?

By Ilardi, Kidd ranks 29th.
By Engelmann, Kidd ranks 5, 4, 9, 26, 25, 14.

Literally, every year of the study, Engelmann has Kidd higher than Ilardi's cume.

Okay, so who do we believe? Well, I don't believe anyone fully, but I give Ilardi's study significantly more weight for a few reasons:

-You can't just add totals from each year and get a cume score. It's always more accurate to put them all in the same data set and compute it together.
-RAPM's advantage over APM only exists in small scale studies. I don't believe a 6-year APM study requires the fluff RAPM puts into to stabilize matters, and Engelmann's RAPM basically agree (though they put it in terms of "In longer term studies, APM comes to more closely resemble RAPM data").
-Engelmann's 6 year study from '05-06 to '10-11 doesn't look terribly different from Ilardi's study.

Last I'll note that by Lewin's 1 year APM numbers from '04-05, Pierce actually finished 1st. I don't use that to say "See, Pierce was better.", but just to show that there is data clouding the issue on both sides.

[Fencer reregistered]

I'd be interested in a recap of how Kidd's apparent impact on defense could be a misreading of what was actually offensive prowess.

The main mechanism I can think of for that is that you hurt your defense less with an offensive make than a miss, and with a miss than a TO. But I'm not clear on how that would map to Kidd's situation in such a numerically remarkable way.

I don't remember the exacts of my previous post, but the gist was that having Kidd as a point guard changed the way that his teams approached the game and that while the net gain was a positive in the overall team product, it may be difficult to decide whether the improvement was more on offense or defense.

The example I used was the way that those early 00s Nets played, and how that might differ with a Nash type PG. Those Nets played an aggressive, attacking style on offense/defense that was made possible primarily because of Kidd's strengths and also played to the strengths of the personnel. The games were a track-meet, and stylistically probably closer to the old-school 60s style of low-efficiency get-out-and-run-and-stop-your-man-more-than-he-stops-you than we've seen in recent years. Martin and Jefferson, at that stage of their careers, were hyper-athletic finishers offensively that also played get-in-your face defense. They were able to play that style, in large part, because they had a system and point guard that ended an abnormally high number of defensive possessions with the ball. Between his guard-leading rebounding and his near league-leading steals, Kidd was making the defense/offense shift very quickly throughout games. As such, the Nets naturally became more of a fast-breaking team on offense. Because Kidd was able to gather those rebounds, Martin and Jefferson were free to consistently spend more time and energy on their man without worrying about crashing the defensive boards. And the offense that the Nets did generate was strongly dependent upon Kidd, whether it was in initiating the breaks or even in creating whatever shots they did get on offense. But on the whole, it would certainly appear that the Nets were putting more focus and energy into their defense than their offense, and that a significant portion of their offense was coming from the break.

Now, were Nash the point guard instead of Kidd, I would expect that the team O-Rtg probably goes up. Very likely the Nets bigs have better rebounding numbers. They still could be a fast-breaking team, but it would be more of a traditional big-man-outlet-pass style, And in the half-court, Martin and Jefferson likely shoot the best percentages of their career at their best volume, with a much larger offensive bent to their games. And by the same token, Martin and Jefferson are likely spending less energy and focus on defense because they are concentrating more on their offensive responsibilities. Instead of pressing up into their man and just boxing out on defense, they now have to leave their assignments more often to crash the boards. And the Nets defense is no longer the top in the league.

Now in these two scenarios, the team O-RTG and D-RTG numbers would suggest that Nash is an offensive wizard while Kidd isn't, and that Kidd just lucked into being on such strong defensive teams. But, at least according to the offensive RAPM numbers from 03 - 06, the individual offensive RAPMs for Kidd and Nash look similar. And their overall team results look similar as well with comparable levels of supporting talent. To me the logical inference is that, as I alluded to above, the way that Kidd played allowed the Nets (and before that the Suns) to run a system that was more defensive in tenure but that still relied upon Kidd's offensive and defensive abilities and overall had a similar OVERALL rating/impact to what a team would have with Nash in his place...but that when trying to break down that impact into offense-vs-defense team ratings the difference is going to show up in the defense. But when taken from team level to individual level, as the +/- stats are designed to do, we can see that Kidd's impact is clearly there and large, on a similar order to Nash's.

Thank you.

It sounds as if you're suggesting that Kidd taking the rebounds let everybody else play better DEFENSE. I'm not totally sure how that works. I absolutely agree that the rebounding scheme involved the bigs blocking out while Kidd actually got the ball, but I'm not sure how that turns them into better DEFENDERS until such time as the ball goes up.

It also sounds as if you're suggesting the games turned into a track meet, with everybody getting tired and being sloppy on offense accordingly, and Kidds' opponents suffering even worse than his teammates. If that's what you mean, do possession counts or something back it up? Those aren't totally dispositive; the idea of an old-style running offense is to sprint down, hope you get a good shot, and set up in the halfcourt if you don't. Still, that would be the obvious place to look for evidence.

[Doctor MJ]

Yeah, and you didn't address it before when I explained it in detail and provided an example (and there are countless examples). Many players can move a team from 0.95 to 1.10. Basically no one has ever moved a team from 1.10 to 1.25 (pts/pos). That would have to be one heck of a coincidence

I often marvel at how we can't seem to get past 1.15. Pretty amazing.

Your point about how no +/- stat can truly adjust for the fact that it's easier (typically) to lift a bad team than a good one is correct. I also don't think there's any objective way to make a stat that appropriately adjusts for this.

I do advocate factoring this in when rating the impact +/- indicates a players...but...it's important not to be too carried away. The dominant APM guys are typically on winning teams, and in the most glaring counterexample (Garnett), we didn't see a huge drop off in +/- when he moved over to a winning team.

[penbeast0]

Gervin is the most deserving of that lot. I prefer Cousy, but Cowens is the one with a chance against Chris Paul(!) for Gawd's sake.

Vote: Gervin
Nominate: Cowens

I agree and will change my nomination from Moncrief to Cowens . . . I love Paul but he just hasn't proved enough yet.