RealGM Top 100 List #11

[Doctor MJ]

With the Baylor thing it depends what's being argued. Did he take defensive attention away from West (I suspect yes), did he create open shots for West (again I'd suggest probably yes). Was he better than West, no. Was his larger role sub-optimal after those first three seasons, probably. Was their a duplication of skills and therefore a redundancy, I don't know. It would depend whether there was sufficient shot creation skills amongst teammates that they could reduce Baylor's load (changing year on year), or on whether West could have shot more and Baylor less at an tradeoff that would have been positive for the team. I sense you think that since they didn't fall off, West could take on Baylor usage and so was under-utilised? But if so isn't that somewhat on West? There's a swings and roundabouts element to this argument (If Baylor wasn't giving West open shots, then wouldn't LA holding up largely okay be to be expected, or conversely if Baylor was so great and not duplicating an somewhat existing skillset, less efficiently, how come LA didn't get much worse without him).

Lots and lots of thought here. It's good. Not sure if I have a lot to say in response because things go in many different directions.

I think some of the threads your picking at just seem like hypotheticals that are unlikely. If you've got two volume scorers and put them both on the court together, you have some redundancy. This is just a fact. You can argue that if they have enough other skills going that the net synergy of the two overwhelms that issue sure, but when you have a team that everyone wonders why they seem to always stay excellent with injury but never get over the top, to me you've got a really work to have another argument make sense above the redundancy one.

Is it on West he didn't take more control? Sure. Factor that in as you please. Not the easiest thing in the world to do though when someone casts as big a shadow as Baylor did.

Whether or not getting the keys to the team might of helped West, he didn't warrent them in '61. He should have started more (though his minutes were ulitmately fine, but as it transpired it unnecessarily p****d West off). Baylor at that point versus West is no contest. I don't mind hypotheticals but it's hard to see West getting the keys to any team at that time.

I don't know what your basis for saying this is. In that season Baylor shot 2100 FGAs, and other than Wilt no one else shot even within 400 of that. It was very clearly not a typical situation that West was dropped into. The Lakers had their horse, and they were going to ride him. Makes no sense to judge West based on what we saw there, and think it represents what would happen most places.

Let's also note that West wasn't some out-of-nowhere guy. He was the #2 pick in a draft with Oscar, which is essentially the same thing as being a normal #1 pick. It was plenty common back then for teams to hand their team over to their star rookie, and West was fresh off a 29/16/4 season with two years of deep tournament runs and an MOP in the tournament. Were he put in Oscar's place in Cincy, you think they aren't going to be looking to reformat themselves around him?

You want to say he was less ready than Oscar immediately? Okay fine, but don't exaggerate it. He shot up to 30 PPG once Baylor missed some time his second year while seeing his TS shoot up by 6%. He was learning very quick as soon as opportunities presented themselves.

Baylor sans West is just a hypothesis. But looking at Basketball-Reference his top 5 scoring outings in '68, West's longest absence were all in games without West. So I think part of the impression that Baylor was taking West's shots in the late 60's might be illusory based on games in which Baylor had to (try to) take up the slack.

Right so you noted before. I don't think you're crazy, but I don't think that's the complete explanation.

Regarding longevity. Criteria might differ here. Some are okay if you miss parts of the season if you get into shape for the playoffs, and obviously that's better for your team than playing through stuff and hurting yourself longer term. But you're not adding value and fwiw you're possibly costing HCA (the value of this varying depending on length of series etc). There's a degree of the hypothetical about it, but with Robertson I feel confident in a given game he'll be available and performing at or close to full strength. West was gutsy so I believe he'd try, but there are a few absences. Even if I was a longevity is only a tie-breaker guy, Oscar is at least I'd suggest close enough to warrent a look at it. And even with a system that penalizes Oscar for bad teammates he has a not insignificant advantage. I'm not someone who writes off longevity. Whilst I'm less consistent than I'd aspire to be my criteria is something like championship value added (not quite the title probability stuff which has a specific meaning on here and has a significant playoff weighting doesn't really mind missing RS, nor WARP or Win Shares type idea where each amount of goodness added is of the same value, even though adding say .200 WS or WARP per 48 gives you a much higher team ceiling than two guys giving .100) or value added with high end bonuses or something like that.

Re: MVP voting. Robertson has a clear edge in MVP shares. That 2nd place in MVP run from '66 goes
66: 2nd
67: No votes
68: No votes
69: No votes
70-72: 2nd

So to yes to me that does mean that over the sum of their careers Oscar was consistently called better since he was still getting All-NBA first teams during that span (West was 2nd team in '68 and '69). MVP voting from that era in particular had a particular best player on the best team flavour (hell Johnny Kerr did better than Wilt in '63). And still Oscar got some mild consideration through to '68. I'd take later model West though I'm not sure how much Oscar was just fitting into a role in Milwaukee and sublimating his game (it's hard to argue with the team level results, though off the top of my head I recall with/without probably isn't in his favour here). But that's 3-5 years (depending on whether you're talking about MVP consideration, West being better per minute, or West having a larger overall impact). Oscar had the same spell of clear advantage earlier in their careers and has the better numbers and hasn't missed as many games. Even if you go in for "Well, West would have been in the MVP running without injuries and injuries don't matter if you make the playoffs" that doesn't necessarily put him anything more than even for that spell '67-'68 in terms of who was considered better. So I don't think people thought Oscar was better every year, year on year. But I think they thought he built enough of a gap for himself by '68 (or '69) that despite what West did thereafter it wasn't closable. And that's what pretty much every GOAT list has said since.[/quote]

Good thoughts.

Regarding this little snippet "along with Oscar's extremely gaudy early '60s stats when things were inflated like crazy". I've seen this reasoning before and I don't like it. Because West played in the same years and had the same chance to put up inflated, extremely gaudy stats. He didn't. It's particularly galling when used as Bill Simmons does manipulatively in terms of his numbers dropping in an expansion era. But West is celebrated for happening to peak later in the apparently weaker league (and not penalized for not putting up numbers when he had the chance to). Okay I'm arguing with someone who isn't here now. Back on topic, if it's just one player was elite first and so it's hard to switch order, fine I can see there's an argument there, I don't know whether its true. At the margins I can see a case his peak is slightly over celebrated because of boxscore inflation (though his team pace was on the slower side, and his best, MVP, got a book written about it year was considerably slower than the pace explosion peak - 109.4 fga per game in '61 and 107.7 in '62 when Wilt got 50, Oscar got his triple double etc; but only 99.1 fga in '64 AND Robertson's teams were inevitably slower than the league average AND the triple double stuff wasn't talked about until the 80s but people always seem to have had Robertson ahead). I just don't see that West did enough to close that gap.

Anyway as before I can see why people would go for West based on slightly different criteria than my own.

I don't understand why you're confused.

The guys with huge primacy in the fastest pace years were the guys who were going to put up the biggest numbers in those eras. West going onto Baylor's team made that a lot less likely to get. Is this really something you're debating?

[Doctor MJ]

Well let's take the qualitative conclusion out of it then.

Deep into their respective careers:
-Malone was the one who continued to act as the volume scorer superstar.
-Malone was the one who continued to play big minutes.
-Garnett was the one whose team continued to show night & day difference based on his presence.

As long as you accept all that and factor that in without dismissing it, I'm fine with it, and I would hope you'd be fine with my competence as long as I did the same.

I wonder, though... Malone playing such big minutes and basically never missing a game means there isn't really a meaningful sample of data for concluding anything about how his team performed in his absence, no? Whereas Garnett did miss a lot of games, so we got a far clearer picture of what things looked like with out him (and, to his credit, it really delineated his impact as strong).

Malone played 81, 49, 82, 81 and 80 games from 98-02 (the 49 being out of 50). That leaves 5 games in which he didn't play.

He played 3030, 1832, 2947, 2895 and 3040 minutes in those seasons, and averaged 36.8 mpg when he did play. That really isn't a significant sample of performance. 5 games, plus an additional 4,160 minutes over a half-decade? That's the sample we're using here to try and indict Malone's team performance.

One of the fundamental premises you just described was that Malone's teams didn't show the same impact from Malone's absence, but you're basing that off of a fairly unreliable sample, no? Perhaps RAPM someone attempts to counter this, but there's a considerably larger sample for Garnett's absence than for Malone's.

Well, what you're largely asking here is whether it makes sense to use RAPM to analyze an NBA season. Are the numbers meaningful if a starter is totally healthy? Those who use it think so. I think so.

Is it possible there's some noise in there? Sure. Fine to question specific values for specific reasons. Right to factor other things too...

But the data is there. We have data from '97-98 until Malone retired. That's quite a lot actually.

[acrossthecourt]

"A 7-point knock to a player's ORTG is just as significant as the RAPM stats which form the basis of KG's argument, you know?"

I just wanted to respond to this little bit: a seven point drop in efficiency given a player's minutes and usage is like a drop of 1 to 1.4 points per game, which is dwarfed by the RAPM results in most cases. Plus, you're not taking into account any increases in efficiency or adjustments in schedule strength.

As I've said, the changes in his efficiency are not statistically significant, and ElGee has posted the similar stats Garnett and Duncan have versus similar defenses. There's just this weird bias people have about Garnett's offensive game, and I don't know how it can ever be destroyed despite mountains of evidence and data.

edit:

Well let's take the qualitative conclusion out of it then.

Deep into their respective careers:
-Malone was the one who continued to act as the volume scorer superstar.
-Malone was the one who continued to play big minutes.
-Garnett was the one whose team continued to show night & day difference based on his presence.

As long as you accept all that and factor that in without dismissing it, I'm fine with it, and I would hope you'd be fine with my competence as long as I did the same.

I wonder, though... Malone playing such big minutes and basically never missing a game means there isn't really a meaningful sample of data for concluding anything about how his team performed in his absence, no? Whereas Garnett did miss a lot of games, so we got a far clearer picture of what things looked like with out him (and, to his credit, it really delineated his impact as strong).

Malone played 81, 49, 82, 81 and 80 games from 98-02 (the 49 being out of 50). That leaves 5 games in which he didn't play.

He played 3030, 1832, 2947, 2895 and 3040 minutes in those seasons, and averaged 36.8 mpg when he did play. That really isn't a significant sample of performance. 5 games, plus an additional 4,160 minutes over a half-decade? That's the sample we're using here to try and indict Malone's team performance.

One of the fundamental premises you just described was that Malone's teams didn't show the same impact from Malone's absence, but you're basing that off of a fairly unreliable sample, no? Perhaps RAPM someone attempts to counter this, but there's a considerably larger sample for Garnett's absence than for Malone's.

Well, what you're largely asking here is whether it makes sense to use RAPM to analyze an NBA season. Are the numbers meaningful if a starter is totally healthy? Those who use it think so. I think so.

Is it possible there's some noise in there? Sure. Fine to question specific values for specific reasons. Right to factor other things too...

But the data is there. We have data from '97-98 until Malone retired. That's quite a lot actually.

It's not as simple as "oh look at how his teams do without Karl Malone." There are hundreds of variables, and we estimate Karl Malone not just with his own impact but everyone else's.

It's not like he only sits out when the game doesn't matter. He still rests at normal intervals during the game, and over the course of a season there's enough there to do something with. But his teammates sit out a lot more, so it's easier to estimate their impact and thus estimate his. RAPM is not just looking at how a player does with*out* a player but when he's in the game and what the scoring margin is.

[Texas Chuck]

Well, what you're largely asking here is whether it makes sense to use RAPM to analyze an NBA season. Are the numbers meaningful if a starter is totally healthy? Those who use it think so. I think so.

Is it possible there's some noise in there? Sure. Fine to question specific values for specific reasons. Right to factor other things too...

But the data is there. We have data from '97-98 until Malone retired. That's quite a lot actually.

Well I agree with what thserkin is saying. And I agree that it makes sense to use RAPM as part of an analysis of an NBA season, but if we look at Malone playing 750-1000 more minutes than KG and being more productive than KG and somehow conclude based on RAPM that those years count towards longevity for KG, but not Malone....well that just doesnt make sense.

[tsherkin]

Well, what you're largely asking here is whether it makes sense to use RAPM to analyze an NBA season. Are the numbers meaningful if a starter is totally healthy? Those who use it think so. I think so.

Is it possible there's some noise in there? Sure. Fine to question specific values for specific reasons. Right to factor other things too...

But the data is there. We have data from '97-98 until Malone retired. That's quite a lot actually.

Yes... in principle. But again, you're talking about a miniscule amount of time without Malone as a point of comparison. it's not actually that much time, really.

I mean granted, he was getting older, so he wasn't going to always remain as effective as he was in his earlier days, that seems obvious. Again, you referenced some +9 number and a fall-off to +5 that I don't see in any of the sites I use as a reference for RAPM; could you drop that link?

Also, why do you think the stat you're referencing disagrees so much with the other RAPM stats and a basic look at his individual performance?

[RebelWithACause]

Well, what you're largely asking here is whether it makes sense to use RAPM to analyze an NBA season. Are the numbers meaningful if a starter is totally healthy? Those who use it think so. I think so.

Is it possible there's some noise in there? Sure. Fine to question specific values for specific reasons. Right to factor other things too...

But the data is there. We have data from '97-98 until Malone retired. That's quite a lot actually.

Yes... in principle. But again, you're talking about a miniscule amount of time without Malone as a point of comparison. it's not actually that much time, really.

I mean granted, he was getting older, so he wasn't going to always remain as effective as he was in his earlier days, that seems obvious. Again, you referenced some +9 number and a fall-off to +5 that I don't see in any of the sites I use as a reference for RAPM; could you drop that link?

Also, why do you think the stat you're referencing disagrees so much with the other RAPM stats and a basic look at his individual performance?

https://docs.google.com/spreadsheet/ccc?key=0Am56xyn_resAdHV2M185TGhoazBTcm5WMFV0eXZqaFE#gid=3

This is the spreadsheet of DocMJ, from 98-12. Normalized and scaled RAPM results.
Those are prior informed numbers.

Still hoping you to update the sheet to 14 rather quick Doc

For older RAPM data, late 90s you can also vist acrossthecourts site

98: http://ascreamingcomesacrossthecourt.blogspot.de/2013/12/1997-98-rapm-non-prior-informed.html
97:http://ascreamingcomesacrossthecourt.blogspot.de/2013/10/introducing-1990s-rapm.html

97 is the first year, so those scaled results have more noise, since its non prior informed.
98 has both prior informed and non prior informed

The other years you may find yourself.

Colts18, shutupandjam and the guys from gotbuckets also provide RAPM data.

[Doctor MJ]

Well, what you're largely asking here is whether it makes sense to use RAPM to analyze an NBA season. Are the numbers meaningful if a starter is totally healthy? Those who use it think so. I think so.

Is it possible there's some noise in there? Sure. Fine to question specific values for specific reasons. Right to factor other things too...

But the data is there. We have data from '97-98 until Malone retired. That's quite a lot actually.

Well I agree with what thserkin is saying. And I agree that it makes sense to use RAPM as part of an analysis of an NBA season, but if we look at Malone playing 750-1000 more minutes than KG and being more productive than KG and somehow conclude based on RAPM that those years count towards longevity for KG, but not Malone....well that just doesnt make sense.

Well as I was saying, if attaching "better longevity" as a label to the data makes you queasy, I'm fine to back away from it, just so long as you see the data involved here, both the part that favors Malone and the part that favors Garnett.

[disenfranchised]

Kobe most ever first team ALL-NBA selections.

Kobe most ever first team ALL-DEFENSE selections.

Kobe most consecutive all-star games ever.

31,000+ points and counting.

5 rings.

Not top ten? The incompetence in that notion is bewildering. There's no question he's top 10 all time, but the real question is, is he top 5? I say YES.

[ShaqAttack3234]

I'm going to vote for Oscar actually. I use to think Oscar was overrated because of the triple double season, and I still think people completely miss the point when they bring that up as the case for him being great, especially since his best season appeared to be '64, but looking at both Oscar's extraordinary offensive impact in his own era, which he have a better idea of thanks to the estimated offensive ratings for pre-'74, and how he was regarded by his peers, with some claiming he was the best, even over Wilt and Russell, I've become A LOT higher on Oscar. He's also the guard from that era who I have the least problem seeing translate to the modern NBA, possibly the player who I have the least trouble seeing period post-Kareem. The reasoning is simple. From the footage I've seen, Oscar was great at protecting the ball and using his size and strength to get off mid-range shots, where he had a great touch, much like Sam Cassell, except Oscar was bigger, seemingly more athletic, a better passer, more dominant and he wasn't known to be the defensive liability Cassell was. Most of the time, I'd have gone with Kobe, but I've really become much higher on Oscar, and it just seemed like the right call to me at this time.

[Basketballefan]

Kobe most ever first team ALL-NBA selections.

Kobe most ever first team ALL-DEFENSE selections.

Kobe most consecutive all-star games ever.

31,000+ points and counting.

5 rings.

Not top ten? The incompetence in that notion is bewildering. There's no question he's top 10 all time, but the real question is, is he top 5? I say YES.

I don't see an argument for top 5...i think you are emphasizing accolades too much and without context. Top 10? Agree but just barely.

[tsherkin]

Kobe most ever first team ALL-NBA selections.

Kobe most ever first team ALL-DEFENSE selections.

Kobe most consecutive all-star games ever.

31,000+ points and counting.

5 rings.

Not top ten? The incompetence in that notion is bewildering. There's no question he's top 10 all time, but the real question is, is he top 5? I say YES.

Your certitude in combination with your contempt for the alternative and your lack of legitimate analysis need a gut check. Differig opinions are to be welcomed and discussed, not denigrated.

You'll want to mind your tone and be more productive if you want to participate in discussion. There is no "incompetence" involved in choosing to believe Kobe isn't a top 10 player, and suggesting otherwise is unproductive and insulting to those who hold that opinion.

[Doctor MJ]

Still hoping you to update the sheet to 14 rather quick Doc

I suppose this is as good a place as most, I'm open to people's thoughts:

What do you think are the most reliable sources of RAPM are now?

I'll say up front, I picked single year PI RAPM for a reason in my spreadsheet, so I'd prefer to stick with that, and I want consistency. At some point though I've got to start updating the spreadsheet, and realistically I'm not going to be generating my own numbers right now.

Folks in here generating their own stuff, I'd be particularly happy to hear your thoughts on the subject. I rather doubt that being used by "my spreadsheet" is a goal of yours, and maybe we'll end up circumventing that shortly, but it would be nice if we could gain more confidence in our statisticians again.

[ElGee]

also without any coherent thought or discussion on what counts as "greatness.

Can you tell me what the difference is between "greatness" and how good someone was ("goodness")? And then can you tell me why it should usurp goodness (i.e. what's the value of that criteria)? I've heard you reference this for years and never understood it...

[ElGee]

RR is extremely useful for regularization, but the problem you're trying to solve requires the data to first be individualized to an extent

Given the possibility that you might be some kind of math genius that is way beyond me...Can you please explain a situation when you would use Ridge Regression so I can understand it? What you're saying is incompatible with everything I know. Thanks.

[ElGee]

f I'm understanding this correctly, when most of the criticisms of Garnett pertain to his playoff offense, which is considerably worse than his regular-season offense, yes? There's a gargantuan difference in his performance on that end of the floor once the playoffs begin

That makes me question the stat even more, though, because he's demonstrably worse on offense in the playoffs, and irrefutably so.

Woah. I know acrossthecourt already addressed this, but am I missing something here? How is he clearly worse of a scorer in the PS? Have you seen the data that I've posted about him? Feel free to lump this in with a response to ATC...

[PCProductions]

Saying RAPM can't find individual value is like saying we can't prove the existence of black holes because we can't see them. It's utterly ridiculous to toss aside an entire subset of mathematics as if it's up some "opinion" to accept it.

[tsherkin]

also without any coherent thought or discussion on what counts as "greatness.

Can you tell me what the difference is between "greatness" and how good someone was ("goodness")? And then can you tell me why it should usurp goodness (i.e. what's the value of that criteria)? I've heard you reference this for years and never understood it...

I can't; that is precisely why I keep raising the issue. We do not have a coherent, consistent idea of what it is we are rating/evaluating, which is a constant stumbling block in our community discourse.

It is my subjective opinion that we have a mixture of evaluations which are attempting to assess career value and player ability, which are inherently different things. We do not have any kind of consistency in which we evaluate from one person to the next, which makes for a weaker debate environment, since not everyone is debating the same thing.

I don't want to suppose one is more important than the other, but it'd be nice to know about which we are talking, as it changes things.

[Doctor MJ]

RR is extremely useful for regularization, but the problem you're trying to solve requires the data to first be individualized to an extent

Given the possibility that you might be some kind of math genius that is way beyond me...Can you please explain a situation when you would use Ridge Regression so I can understand it? What you're saying is incompatible with everything I know. Thanks.

Yeah, this is something I keep bringing up too, and it's going to keep happening until AUF or someone else gives an answer, because I really think the issue is probably that AUF doesn't like regression analysis period rather than him having a real basketball issue. But he can prove me wrong with some pro-regression analysis points.

Also, not to be a jerk, but saying "RR is extremely useful for regularization" is just a weird thing to say and that drives back into linguistic issues. AUF, and others I'm sure, object to using the term "impact" to describe what RAPM does, but it's just a way put what it does into language. "Regularization" is similarly just away to describe the type of thing that ridge regression (and other things) is doing using language that's a bit more intuitive to outsiders, it is meaningless to describe how useful something is at regularization, at least so far as I understand it.

[RebelWithACause]

Still hoping you to update the sheet to 14 rather quick Doc

I suppose this is as good a place as most, I'm open to people's thoughts:

What do you think are the most reliable sources of RAPM are now?

I'll say up front, I picked single year PI RAPM for a reason in my spreadsheet, so I'd prefer to stick with that, and I want consistency. At some point though I've got to start updating the spreadsheet, and realistically I'm not going to be generating my own numbers right now.

Folks in here generating their own stuff, I'd be particularly happy to hear your thoughts on the subject. I rather doubt that being used by "my spreadsheet" is a goal of yours, and maybe we'll end up circumventing that shortly, but it would be nice if we could gain more confidence in our statisticians again.

Unfortunately our most knowledgeable poster about RAPM is gone, mystic.

GotBuckets is my top choice for RAPM right now.

[ceiling raiser]

Still hoping you to update the sheet to 14 rather quick Doc

I suppose this is as good a place as most, I'm open to people's thoughts:

What do you think are the most reliable sources of RAPM are now?

I'll say up front, I picked single year PI RAPM for a reason in my spreadsheet, so I'd prefer to stick with that, and I want consistency. At some point though I've got to start updating the spreadsheet, and realistically I'm not going to be generating my own numbers right now.

Folks in here generating their own stuff, I'd be particularly happy to hear your thoughts on the subject. I rather doubt that being used by "my spreadsheet" is a goal of yours, and maybe we'll end up circumventing that shortly, but it would be nice if we could gain more confidence in our statisticians again.

Unfortunately our most knowledgeable poster about RAPM is gone, mystic.

GotBuckets is my top choice for RAPM right now.

Don't forget acrossthecourt and shutupandjam.

mystic is super sharp though, I really like what he brings to the table in terms of not only discussing results, but the theory/mathematics behind his thinking/methods.