RealGM Top 100 List #8

[mtron929]

What needs to be accounted for in the statistical argument is the fact that the NBA is a long season with virtually half of the teams making the playoffs. Accordingly, top players realize that they can slack off in the regular season (e.g. Shaq in the 3 peat times), save energy, and exert more effort in the playoffs. Now, if the goal of the NBA was to win as many regular season games as possible with the largest scoring margins, then I would have to think that guys like Shaq and Lebron during the Heat era would have played a lot differently.

That is, they would have played like how KG plays.

KG is getting overrated in the statistical arguments because he always plays really hard. But that is not necessarily prudent. And I don't necessarily think that this is a positive attribute in the context of winning a championship. But it sure did wonders for his +/- stats.

[magicmerl]

Spoiler:

Garnett's postseason scoring (referenced to Duncan's)

Garnett's postseason scoring is a hot-button topic in this project. Colts18 and Ardee have asked me about it directly, and several others (including most recently magicmerl in this thread) have referenced KG's boxscore playoff numbers as a reason he doesn't gain traction with them. Fair enough. My preference would be to focus more on the entirety of Garnett's postseason impact (not just scoring), and I plan to do that eventually, but I don't think most are in the mindset to receive that until I address KG's postseason scoring. So, I'll go ahead and shine a light on that aspect of the game first. And I'll start off with a post from andrewww that articulated a few threads back what many believe, regarding the relative differences between Duncan's postseason scoring and Garnett's.

(This should be fun. On the one hand, as some of you have been asking for, I'm about to deliver a long post that doesn't reference RAPM a single time. On the other hand, I don't get the sense that my methods/conclusions will be well received by some of those that most loudly decry RAPM.)

For those in the KG boat and you've certainly made great points to expand on what I already knew of KG as a player...statistcs and especially advanced statistics really favour KG because of his all-around game on both ends of the ball including all-time great defense. No noticable weakness on offense minus volume scoring and it's a great recipe to build a team around, which was validated with the Celtics.

Fundamentally speaking, it is easier to build around a big man (PF/C) with great defensive skills blended in over a high scoring guard with a significantly lesser impact on defense (eg. Kobe). I get that.

But KG was always exposed when called up to be the focal point of the offense as a bonafide number 1 option. In other words, I think of him as a 6-11 version of Scottie Pippen who was one of the 2 best wing defenders ever imho (along with Bobby Jones). A great player especially on defense, but someone who was best served as a number 2 or 3 option on offense. Duncan was always someone while watching that I felt was more reliable when called upon to get some sure fire buckets.

This is why I ask myself, what if I threw KG into Duncan's shoes? I'm confident the Spurs wouldn't have as many as the 5 championships theyve currently won. There were times in those runs against the Suns, Lakers, Pistons and Heat where he delivered in the clutch, much moreso than when KG finally had a title-level team around him in Boston.

Duncan was on a higher level offensively and thus his standing as an overall player as well imho.

Default case of Duncan's postseason scoring over KG's, and initial rebuttal stance

[spoiler]This post was actually from a couple threads back, but I brought it into this thread so I could reply. Being in Vegas was like being in a time warp, so I feel like I've been missing some good conversations. But I wanted to address this, because I believe it's a common viewpoint in here. That Duncan and Garnett may be similar scorers in the regular season, but in the postseason that Duncan ramps his scoring up more than Garnett and thus that Duncan is on "a higher level" offensively. The oft-given basketball reasoning given to explain this difference is a mixture of:

Duncan maintains his efficiency because he's a more traditional low-post scorer.
He puts pressure on defenses that KG never could with his more high-post game.
KG is content to sit back and shoot fadeaways! He doesn't even draw fouls like Duncan.

With the conclusion that clearly Duncan's scoring style is just more valuable in the postseason, as evidenced by his efficiency, and that's why Duncan is the true alpha and KG is the glorified Pippen.

Obviously, I've never agreed with this line of reasoning. For more than a decade now I've been pointing out just how much KG had to do to both get his team to the postseason and then to give them any chance to compete once they get there. In an old post (I think from the Peaks project) I made the case that Duncan's slightly higher scoring efficiency a) was in the noise as far as determining their relative impacts and b) was NOT based on him playing more in the paint than KG. Instead, I argue that the difference between Duncan's postseason scoring efficiency and Garnett's comes primarily from having less overall help when playing against higher levels of competition.

Both Duncan and Garnett are the type of do-everything player that are ultra portable because they can fit their games into any team need. If a team needs more scoring, they can be primary scorers. If the team needs more rebounding, they can ramp that up. Need defense? They can be the best in the league. Etc.

But at some point, even for them, it becomes a zero-sum game. One player can't consistently have to maximize in every possible way without there being some degree of slippage somewhere. For both Duncan and Garnett, when their teams were grossly outmanned their slippage tended to show up in the scoring efficiency categories. Ironically, ElGee pointed out that small differences in individual scoring efficiency doesn't have a big affect on postseason odds even though it is at the foundation of every major box score stat that we have (e.g. PER, win shares, offensive rating, true shooting %).

Duncan and Garnett prime postseason scoring and "advanced" box score stats, no context

Spoiler:

So let's relate the above to Duncan and Garnett. If you look over their primes (1999 - 2008) in the postseason, you see these "advanced" box score stats:

Duncan: 26.4 PER, 55.2% TS, 110 O-Rtg, .215 Win Shares/48
Garnett: 23.9 PER, 52.3% TS, 106 O-Rtg, .172 Win Shares/48

"It's not a large margin, but it's a significant margin" is the common thought process, "and this difference in postseason effectiveness is the margin that Duncan has over Garnett"

Hmmm. Just to be sure of that, let's take a look at the source box score numbers that generated those "advanced" stats:

Duncan: 23.6 ppg (on 17.5 FGA + 8.8 FT = 21.4 shot possessions used), 13 reb, 3.6 ast (3.1 TO)
Garnett: 22.3 ppg (on 18.8 FGA + 5.8 FT = 21.3 shot possessions used), 12.7 reb, 4.5 ast (2.9 TO)

So that formerly significant seeming difference in the advanced box score stats comes from Duncan scoring 1.3 more ppg on almost exactly the same number of shot attempts (including possessions that end in FTs). That's it. That's the difference. And while of course, if given the choice anyone would take that extra point, that difference is minor at best. And you could argue that volume-wise, that 1.3 points is countered by KG averaging an extra .9 assists at the same turnover rate. But I'm hoping that looking at this, the point is clear: there WAS hardly any playoff scoring efficiency difference across their primes when taken as an average across years.

I find this convincing. But if you don't, I still say that these numbers don't reflect Duncan being more effective than Garnett as a playoff scorer. Instead, I believe that this sample is skewed by Garnett's Minnesota teams consistently facing only stacked teams. And that Duncan's efficiency in similar situations mirrored Garnett's.

Before we dig deeper in the postseason situations, let's first look at Duncan and Garnett in the regular season so that you can get a feel for just how similar they were as scorers in their primes. In the regular season, from 1999 - 2008:

Prime Garnett and Duncan had almost exactly the same regular season scoring efficiency

Garnett: 22 ppg (8.6/17.5 FG + 4.5/5.7 FTA = 20 scoring poss), 12.2 reb, 4.8 ast (2.7 TO)
Duncan: 21.7 ppg (8.2/16.4 + 5.1/7.5 FTA = 19.7 scoring poss), 11.8 reb, 3.2 ast (2.8 TO)

Celtics Garnett (2008 - 2013) had almost the exact same scoring efficiency in the POSTSEASON as 2008 - 2013 Tim Duncan .

Next, let's look at their postseason numbers from 2008 - 2013 (Garnett in Boston vs Duncan in similar team caliber situation) in their late-prime/post-prime time periods:

Garnett: 17.5 ppg (7.3/14.8 FG + 2.9/3.5 FT = 16.3 poss), 53.3% TS, 9.9 reb, 2.6 ast (2.0 TO)
Duncan: 18.2 ppg (7.4/15.4 FG + 3.5/5.3 FT = 17.7 poss), 51.6% TS, 10.9 reb, 2.6 ast (2.1 TO)

So, KG and Duncan have almost the exact same scoring efficiency in the regular season in their primes and also a very similar scoring efficiency in the playoffs in their late prime/post-prime. So if there really is a difference in their postseason scoring in Duncan's favor, it would have to show up in their actual primes. So, let's look more closely at that time period.

In-context comparison method for postseason scoring for KG and Duncan

Spoiler:

(Note: Normally I would move forward with the understanding that Duncan's casts tended to be more talented/stronger than Garnett's. I think that's not just a supportable point, but a very clear one. However, there's been pushback in this project that Duncan's 01 - 03 casts were worse than Garnett's 2002 cast in particular, and that notion has gained enough traction (plus I know from experience debating with Baller 24) that if I argue that point here it will deflect away from the point I'm trying to make.)

So as we review their postseason performances for each year of their early prime thru peak (1999 - 2004), I'm going to not factor Duncan's or Garnett's team strength in. I'll instead compare them using a descending hierarchy of methods to get their performances on as similar of a plane as possible:

1) Performance against common opponent in the same postseason
2) Performance against the most similar caliber positional defense in the same postseason
3) Performance against highest quality opponent in the same postseason

1999 - 2001: Postseason performances in Early Prime against all-history defensive power forwards

Spoiler:

In 1999 and 2001, all eight of KG's playoff games had him facing one of the 2 best defensive power forwards in NBA history: Tim Duncan. Duncan also had those eight games against the other best defensive power forward in NBA history in Kevin Garnett...but in addition he had 22 other games in those two postseasons that were not against that level of positional defense. Let's compare:

Garnett against Duncan: 21.4 points (52.5% TS), 12 rpg, 4.1 apg (2.4 TO)
Duncan against Garnett: 20.7 points (51.4% TS), 11.9 rpg, 3.4 apg (2.2 TO)
Duncan against everyone else: 24.8 points (56.6% TS), 13.1 rpg, 3.2 apg (3.8 TO)

In 2000, Duncan missed the playoffs with an injury while Garnett faced off against the Portland Trailblazers, whose starting front line was Rasheed Wallace, Scottie Pippen and Arvydas Sabonis. If KG and Duncan are 1-2 on the defensive power forward list of this era, Rasheed Wallace is a comfortable third. And of course, he had Sabonis clogging the paint defensively with Pippen able to provide help as well.

In that series, Garnett averaged 18.8 points (44.1% TS), 10.8 rpg, 8.8 apg (3.3 TO)

Everything looks reasonable except for the awful scoring efficiency. KG was clearly (CLEARLY) the focal point of the Wallace-led Blazers defense that series, which really hurt his scoring efficiency overall. He did do a good job of taking advantage of the extra attention he was receiving to find teammates, as reflected by his excellent assist numbers, and Terrell Brandon was able to do well in the expanded space he had to work with. But the focus in this debate is KG's scoring efficiency vs. Duncan's. So let's go there.

Duncan didn't face the Blazers (or anyone else) that postseason, but he did face them three times in the regular season. Let's look at the four Garnett postseason games, and the three Duncan regular season games:

Garnett 1: 12 points (6/20 FG, 0/0 FT), 10 reb, 11 asts, 2 TO
Garnett 2: 23 points (8/16 FG, 7/10 FT), 10 reb, 5 ast, 4 TO
Garnett 3: 23 points (11/22 FG, 0/0 FT), 13 reb, 10 ast, 2 TO
Garnett 4: 17 points (5/20 FG, 6/6 FT), 10 reb, 9 ast, 3 TO

Duncan 1: 9 points (2/9 FG, 5/6 FT), 6 reb, 1 ast, 0 TO
Duncan 2: 12 points (5/9 FG, 2/2 FT), 2 reb, 2 ast, 1 TO
Duncan 3: 36 points (13/21 FG, 10/11 FT), 15 reb, 6 ast, 1 TO

Of course one was in the playoffs and one was in the regular season so it's not a full apples-to-apples comparison, but I see a similar trend. Both postseason KG and Duncan had two very underwhelming scoring performances against the 2000 Trail Blazers. Duncan had one berzerk game as well, while postseason KG had 1.5 reasonably good scoring efforts. But looking at this, I see no reason to believe that in the postseason Duncan would suddenly have been putting up his normal scoring output against the Blazers. Especially if he was called on to also have much more complete floor games in the postseason than he did in the regular season.

Conclusion: If you look at the overall scoring eficiency/"advanced" box score stats of playoffs KG to playoffs Duncan for 1999 - 2001, it appears that Duncan gives you a solid advantage. However, if you actually go into the match-ups and compare Duncan's performance (or expected performance) against the same level of power forward-specific defense that KG faced in those postseasons, I see no evidence that Duncan scored any better or more efficiently than KG did when in KG's circumstances.

My conclusion - early prime KG was just as good of a postseason scorer as early prime Duncan by both volume and efficiency. It wasn't Duncan's interior offensive style that led to the better numbers, it was instead the relative level of competition difference.

2002: Postseason performance at (Duncan) & near (KG) peak against elite team

Spoiler:

In 2002 Garnett's Timberwolves faced the 57-win Mavs in the postseason. In 2002 David Robinson was injured in the playoffs, so when Duncan's Spurs faced the 58-win Lakers Duncan's squad was overmatched. So, let's look at the scoring efficiencies and overall output in these two series as well as Duncan's opening round series against the 45-win Supersonics.

KG vs Dal: 24 ppg (51.3% TS), 18.7 reb, 5 asts (4 TO)
TD vs LA: 29 ppg (51.7% TS), 17.2 reb, 4.6 ast (4.6 TO)
TD vs Sea: 25.8 ppg (60.5% TS), 11 rpg, 5.5 apg (3.5 TO)

My conclusion: against the stronger teams where they were called upon to do everything, again, Garnett's and Duncan's lines look very similar. Both put up video game counting stats but at poor efficiency. Against Seattle Duncan didn't have to do as much and both his scoring and passing efficiencies were much better.

2003 - 2004: Postseason performance at peak against the same team both years

Spoiler:

In both 2003 and 2004, Duncan's Spurs and Garnett's Wolves matched up with the same Lakers' squads as their toughest tests. Here's how their numbers compared:

KG vs LA: 25.4 ppg (52.9% TS), 14.6 rpg, 4.9 apg (3.4 TO)
TD vs LA: 24.4 ppg (55.6% TS), 12 rpg, 4.1 apg (3.5 TO)

This is the first of these comparisons where Duncan has slightly higher scoring efficiency, but it is very close. On the flip side, Garnett had better passing efficiency (as measured by assists vs turnovers). On the whole, these are exceedingly similar performance levels on offense.

My conclusion: again, against (in this case the exact same) competition, Garnett and Duncan look remarkably the same in the postseason.

Overall conclusion: Duncan has already been voted in several threads ago, so this post isn't intended to make you place KG over Duncan on your lists. That ship has sailed. Instead, I just want you to consider what it was about Duncan that made you trust him as a playoffs focal point more than Garnett. My belief is that as scoring options (in the regular or postseason), Duncan and Garnett are so similar it's ridiculous. Stylistically there are differences (Duncan's approach is more low-post while Garnett's is more high-post), but there's no inherent advantage of one method over another...it's all about results. A quick glance at the basketball-reference stats pages would seem to indicate that Duncan was a better scorer than Garnett in the postseason to a small (but noticeable) degree. However, as I illustrated above, if you really look at the context of their postseason scoring, when KG and Duncan were in similar situations they scored in an almost eerily similar levels.

So. Duncan's and KG's offensive styles stay pretty consistent from the regular season to the postseason. It seems to me that the primary "advantages" that Duncan had as a postseason scorer were because traditionally low-post offense was thought to be more efficient and effective at warping defenses... and then that the Basketball-reference stats indicated that Duncan DID have a postseason advantage. However, historical trends indicate that, overall, the spacing effect of a big man with shooting range and a big man able to run the offense through the high-post are very conducive to good team units. And Garnett's scoring efficiency was almost exactly the same as Duncan's in the regular season over their primes, in the postseason in their late/post primes, AND (when put into context) in the postseasons of their actual primes and peaks.[/spoiler]

Thus, if your reasoning for feeling that Garnett shouldn't be ranked in the top-10 is in any way due to your belief that his postseason scoring isn't strong enough to warrant that, I ask that you just examine again what I laid out above. And then really consider whether you were giving Garnett a fair review in light of the way that you view Duncan.

Thankyou, this is a very valuable post.

I remember back in about 2004 looking at shot charts on NBA.com, and would you believe it? Garnett shot a higher fg% than Duncan from EVERYWHERE on the floor. Everywhere. Yet because Duncan took 2/3 of his shots at the rim, he had a higher overall fg%. What can you make of that? Is Garnett the clearly better player because he litterally is better from everywhere? Or is Duncan the better player because his brain is directing him to take more of his most effective shots?

I think part of the thing that is so hard to fight against is the 'championships or bust' mentality. Championships are really important to a players legacy, yet they are won by teams, not individuals. I personally find it slightly credible to give Duncan a disporportianate level of credit for winning before, during and after his prime, with different rosters. He's happy to do the things that the team needs to win, whatever that might be. Garnett *wants* to win, but he wants to do it more on his own terms (i.e. jumpshooting). It seems like a little thing and I'm uncomfortable thinking of Garnett dropping 10 spaces behind Duncan because of it.

[magicmerl]

But your analogy doesn't quite work, because if he does think KG is better than Duncan, and he thinks Duncan is better than Robinson - then it does make perfect sense to say that KG is better than Robinson. A>B>C logic is applicable in that scenario.

In your scenario you admit that you threw your match, so naturally your daughters statement doesn't account to much.

My daughter *thought* she was better than me. She knew I was better than my dad. Therefore, she was better than him. But her conclusion was based on a flawed assumption.

Just like assuming that Duncan is worse than Garnett is a flawed assumption.

[HeartBreakKid]

What needs to be accounted for in the statistical argument is the fact that the NBA is a long season with virtually half of the teams making the playoffs. Accordingly, top players realize that they can slack off in the regular season (e.g. Shaq in the 3 peat times), save energy, and exert more effort in the playoffs. Now, if the goal of the NBA was to win as many regular season games as possible with the largest scoring margins, then I would have to think that guys like Shaq and Lebron during the Heat era would have played a lot differently.

That is, they would have played like how KG plays.

KG is getting overrated in the statistical arguments because he always plays really hard. But that is not necessarily prudent. And I don't necessarily think that this is a positive attribute in the context of winning a championship. But it sure did wonders for his +/- stats.

I think that has to be the first time I've seen someone get punished for playing too hard. That's gotta count for something.

[mtron929]

What needs to be accounted for in the statistical argument is the fact that the NBA is a long season with virtually half of the teams making the playoffs. Accordingly, top players realize that they can slack off in the regular season (e.g. Shaq in the 3 peat times), save energy, and exert more effort in the playoffs. Now, if the goal of the NBA was to win as many regular season games as possible with the largest scoring margins, then I would have to think that guys like Shaq and Lebron during the Heat era would have played a lot differently.

That is, they would have played like how KG plays.

KG is getting overrated in the statistical arguments because he always plays really hard. But that is not necessarily prudent. And I don't necessarily think that this is a positive attribute in the context of winning a championship. But it sure did wonders for his +/- stats.

I think that has to be the first time I've seen someone get punished for playing too hard. That's gotta count for something.

lol. I think you have to pick your battles to win the war.

[magicmerl]

IThe question now is, what makes KG better than Dirk, Robinson, Malone, Barkley, Bird, Hakeem, Magic, Kobe ext

A post comparing Garnett and Duncan doesn't expand on that in any way

Actually, I kinda disagree. If he can prove that Duncan and Garnett are basically equal, then either
a: It's a mistake that Duncan was voted in as highly as he was, or
b: Garnett should be voted in immediately after
(or some combination of the two).

[drza]

You're implying that Boston wasn't the favourite to win that year, that people were in some way surprised. That's not the case. I thought they would win it all before the season started, and I *think* the Vegas oddsmakers did too.

FWIW, I got the Celtics at 15 - 1 to win the title at Mandalay Bay in September or October of 2007. Comparing that to this year's odds, Vegas has five teams at 10 - 1 or better. Based on this, I'd say that Vegas oddsmakers had the Celtics as a good team, but not one of the primary contenders.

This also matches with the projections of sites like ESPN, who had the majority of their panelists not pick the Celtics to even make it out of the East, let alone win the title. There's a link somewhere to the old predictions, but I want to say that most projected the Celtics to win their division bow out in either the conference semis or the conference finals.

[HeartBreakKid]

But your analogy doesn't quite work, because if he does think KG is better than Duncan, and he thinks Duncan is better than Robinson - then it does make perfect sense to say that KG is better than Robinson. A>B>C logic is applicable in that scenario.

In your scenario you admit that you threw your match, so naturally your daughters statement doesn't account to much.

My daughter *thought* she was better than me. She knew I was better than my dad. Therefore, she was better than him. But her conclusion was based on a flawed assumption.

Just like assuming that Duncan is worse than Garnett is a flawed assumption.

Hm..I still don't see how the analogy is any good. I mean you throwing a match and tricking your daughter into thinking she is better than you, thus better than her grandfather is not the same thing as someone thinking Kevin Garnett is better than Duncan, thus Robinson - I mean it's really not even close.

I'm not gonna harper on this any more, but to get the gist of it, it isn't a fact that Duncan is better than Garnett, it is a fact that you threw your match though. It's really not the same

[magicmerl]

I think that has to be the first time I've seen someone get punished for playing too hard. That's gotta count for something.

No, I had the same criticisms of the Jazz in the 90s era. Back then all of the top teams would rest their starters once their playoff seeding was locked in, but the Jazz played those meaningless end of season games with the same intensity because dammit, that's how Jerry Sloan plays basketball.

I think that history has proven that strategic resting during the regular season leads to better playoff results.

[Purch]

IThe question now is, what makes KG better than Dirk, Robinson, Malone, Barkley, Bird, Hakeem, Magic, Kobe ext

A post comparing Garnett and Duncan doesn't expand on that in any way

Actually, I kinda disagree. If he can prove that Duncan and Garnett are basically equal, then either
a: It's a mistake that Duncan was voted in as highly as he was, or
b: Garnett should be voted in immediately after
(or some combination of the two).

I disagree, just because you state that at one point Duncan and Garnett put up similar numbers, it doesn't mean that Garnett now has a case to be ranked over Hakeem, who most people stated had a higher peak than Duncan

[HeartBreakKid]

But your analogy doesn't quite work, because if he does think KG is better than Duncan, and he thinks Duncan is better than Robinson - then it does make perfect sense to say that KG is better than Robinson. A>B>C logic is applicable in that scenario.

In your scenario you admit that you threw your match, so naturally your daughters statement doesn't account to much.

My daughter *thought* she was better than me. She knew I was better than my dad. Therefore, she was better than him. But her conclusion was based on a flawed assumption.

Just like assuming that Duncan is worse than Garnett is a flawed assumption.

Hm..I still don't see how the analogy is any good. I mean you throwing a match and tricking your daughter into thinking she is better than you, thus better than her grandfather is not the same thing as someone thinking Kevin Garnett is better than Duncan, thus Robinson - I mean it's really not even close.

I'm not gonna harper on this any more, but to get the gist of it, it isn't a fact that Duncan is better than Garnett, it is a fact that you threw your match though. It's really not the same

Actually magicmerl, I think your analogy would work better if you didn't throw your match. Ie, your daughter beating you once legitimately wouldn't make her better than her grandpa. I think that would fit more with the discussion.

[HeartBreakKid]

I think that has to be the first time I've seen someone get punished for playing too hard. That's gotta count for something.

No, I had the same criticisms of the Jazz in the 90s era. Back then all of the top teams would rest their starters once their playoff seeding was locked in, but the Jazz played those meaningless end of season games with the same intensity because dammit, that's how Jerry Sloan plays basketball.

I think that history has proven that strategic resting during the regular season leads to better playoff results.

Yeah, I understand what you're saying, though I think we need to realize that even the best teams "coast" through a short number of games. The best teams still play high impact basketball hence why they get amazing records, I doubt a star player taking a few off games here and there or them lollygagging against inferior competition explains the gigantic difference in stats between that individual and Kevin Garnett.


On a slightly related note, I dont think any star player could coast on the Timberwolves, not without missing the playoffs - which actually did happen to Garnett.

[D Nice]

What needs to be accounted for in the statistical argument is the fact that the NBA is a long season with virtually half of the teams making the playoffs. Accordingly, top players realize that they can slack off in the regular season (e.g. Shaq in the 3 peat times), save energy, and exert more effort in the playoffs. Now, if the goal of the NBA was to win as many regular season games as possible with the largest scoring margins, then I would have to think that guys like Shaq and Lebron during the Heat era would have played a lot differently

I've brought this exact point up in the past discussing 2010 Kobe. Shaq actually doesn't seem to get knocked for the trait even when it got to the point of exacerbating issues so significantly his teams would under-perform by, you could argue, double-digit win totals. Also really haven't heard anybody knock '14 Lebron for it to the degree that he gets ranked behind Durant (it should and rightfully did however factor into him being MVP).

But yeah, people have a tendency to over-compartmentalize some things when looking at certain guys.

[drza]

Okay, but could you address the parts about Robinson and the 90's RAPM data? I want to know not only for this debate, but for general use, like comparing Marbury to Kyrie Irving (pretty similar RAPM numbers from that site through their first three seasons).

The 90s "RAPM" data on Engelmann's site that you linked are all estimated data, and not RAPM at all. They are, I believe, a version of SPM that uses box score data to try to estimate what the RAPM might have been. Not the same thing.

As far as I know, the only site that has any actual RAPM data from the 90s is AcrossTheCourt's blog. The play-by-play data is available back to 1997, so the first prior-informed RAPM data (which is what is generally the standard for non-boxscore measures) is for the 1998 season.

Doc MJ's now much-cited spreadsheet uses AcrossTheCourt's RAPM values from 98 - 2000; there is no publicly available 2001 RAPM; the 2002 RAPM data is from about the last 1/3 of the season + the playoffs (Englemann's site), and then all of the other RAPM data on that spreadsheet come from Englemann's prior-informed RAPM data up through 2012.

[magicmerl]

On a slightly related note, I dont think any star player could coast on the Timberwolves, not without missing the playoffs - which actually did happen to Garnett.

He got the best of both worlds: playing hard out for 40 minutes a game while not even making the playoffs.

On one level that's tragic. On another level it's stupid.

Actually magicmerl, I think your analogy would work better if you didn't throw your match. Ie, your daughter beating you once legitimately wouldn't make her better than her grandpa. I think that would fit more with the discussion.

Agreed.

[An Unbiased Fan]

I would agree that evaluating defense is harder. The question is why RAPM is the answer. Vlade Divac is rated higher than Ben Wallace, Bogut, and Dwight by it.

97-14 DRAPM:
Divac - 2.60
Big Ben - 2.39
Bogut - 2.35
Marc Gasol - 2.09
Dwight - 1.96

To me, its clear that Divac was just part of more defensive rotations due to the King's lack of size in their roster. Which illustrates why RAPM is fine for coaches to analyze their lineups, but not for individual impact. No two players have the same role, roster, team system, rotations. There is no mechanism to extract the individual from the group.

Spoiler:

This helps me understand a little more your issue with "individual impact". To me you're building the word up into some thing more than it is.

If I have a player who I can insert into a lineup and the result will be an improvement of a certain amount, "impact" is a perfectly reasonable choice of words for the effect his insertion will give me. I also tend to use the word "lift" a lot, which I don't recall if it bothers you, but really the point is that it's utterly reasonable to use some word here, and it hardly makes sense to to coin some random word for it. The process of creating new word senses for more common words as we begin to talk in more granular detail in a discipline is what it is, and really the only time to object is if there's a fundamental problem with the words being used because other words would be better.

I see a pattern here where you - and many others - don't allow things to fail gracefully. If I link RAPM and impact, being fully aware that correlation is not causation, I don't expect the connection to be free from noise. I push forward with it nonetheless as a thing I expect to be imperfect, whereas you're inclined to show the counter example to prove they are not the exact same thing, and then not use it to anywhere the extent I do.

Again, lineups themselves have separate functions. There are defensive rotations, offensive rotations, small ball, big rebounding lineups, crunchtime lineups, and a ton more depending on the coach, his roster, and his system.

Because you're using lineup data as a base, and not individual numbers, everything produced is reflective of rotational trends, nothing more. Rashard out did Dwight in 2009 because of how Stan used him, NOT because "he was inserted into a lineup and the result was an improvement"

That's the problem, the very premise of RAPM as an individual measurement is flawed. Its fine for rotational data, but that's it. There's nothing in its calculation, that separates the individual from the group. You instead get lineup trends

Spoiler:

Which is the right way to do things? Well, my way naturally. Obviously I don't expect you to simply accept that, and we probably won't come together on this, but I'll say a couple things:

1) Think on the concept of the "prior" being used in RAPM. Basically the R in RAPM amounts to infusing the sample set with data that will probably smooth out the fluky data. In doing so we lose some validity, but gain reliability, and in practice in many contexts that still makes for a more powerful tool.

Now, in our lingo here "non prior" actually means a prior where we assume pure neutrality. No biases at all toward any player! Add in zeroes everywhere! Why is this not the most effective way of doing things? Because true "no bias" would mean putting in the right value, and a pure neutral infusion of data deviates from what's actually right. If we can use some other method that will probably lead to something that deviates less from what's actually right, then most of the time we'll get a better result.

And this is what I'm doing when I'm using RAPM as a starting place for impact analysis. I'm starting with some level of confidence that the data is meaningful, and going from there. I look at plenty of other things to - orthogonal data, observations, reputations, I look at aspects of the lineup allocation that could influence it, etc. The point is though, there's never a point where I throw the thing out simply because it disagrees with something else. To do that is to rationalize one's existing opinion, and if one's willing to do that in one place without qualm, then one may end up doing it everywhere.

So you bring up Divac with regards to Kobe Bryant, but Divac has nothing to do with the fact that year after year after year we can't find any major correlation between Kobe and his team's defensive success. You bring up Divac only as a way to essentially say "So it might just be coincidence", and that's just not good enough. Maybe there's something in there that essentially makes Kobe unlucky again and again and again by this metric, but dayum, that's astonishing if true. So astonishing in fact that it's pretty clearly not the most likely "prior" if you're following my analogy.

Now really focusing on Kobe a bit here with regards to this stat: There is indeed a specific issue with separating offense and defense with +/-, and if you really want to focus on that as the explanation, I'd be interested to see where you went with it. The reason why it's tough to take that argument so seriously though is that it very clearly to a "same difference" kind of thing. If the lineup focus is deflating Kobe's apparent defensive impact, it's also inflating his offensive impact accordingly, and people who scoff at the notion that Kobe's defense is overrated are no less likely to scoff at the notion that his offense is overrated. In the end, this stat makes Kobe look fantastic, just not as fantastic as some are inclined to think.

Hmm, I have to wonder if you read my thoughts on RAPM fully, because I was pretty clear about why Divac did better defensively in RAPM than the other centers. I never inferred "So it might just be coincidence", quite the opposite. I specifically laid out why he was on top, and picked that example on purpose.

1) The Kings lacked size on their roster, so whenever Adelman had defensive rotations on the floor...he needed his best bigman. By the very nature of his heavy participation in defensive rotations, Vlade put up gaudy DRAPM numbers. This phenomenon is seen in others too. Defensive specialists will rate high, as will offensive specialists. 6th men like Manu/Odom will rate high. There is no mystery with RAPM for me, I fully understand the stat. My point is that it clearly doesn't quantify individual impact...at all.

2)Kobe's DRAPM is deflated because his backcourt running mates were Derek Fisher, Chucky Atkins, Smush Parker, and they spent a good amount of time with him on offensive rotations. You also have the effect of Phil always wanting either Kobe or Pau on the floor, so Kobe spent good stretches with weak defense up front. Conversely La went through Pau/Bynum a bit more when Kobe was off court, and had big lineups more equipped for defense. in the 3peat days it was the same except you had guys like Slava/Samaki in Kobe's rotations more.

RAPM is nothing but rotational trends. Nash will do great because Phoenix had no floor general like him, and a unique system that depended on his skill. It's not "noise" that screws up the results, based on the methodology, all that 'noise" is supposed to be there. You're just expecting the data to be something its not, instead of what it is.

Spoiler:

2) Everything I'm talking about here in terms of accepting and using an imperfect tool, it all is basically taken as a starting point by people are real data analysts. The whole notion of regularization, the use of priors, the fact that regression analysis provides nothing but correlation, and the fact that we are forever attempting to bridge correlation and causation...that's what data analysis is. That's what science is with any kind of complicated data source.

So we're in this awkward situation when analytics come to basketball and we've got people having it thrust upon them not in some structured educational setting but just as it comes. People bring up objections which are GREAT philosophy of science questions, which is cool, but then it gets weird, because others in the room have the answers at the ready, and those posing the questions often don't really listen to the answer.

And with all of this, someone like me comes off as arrogant. How dare I say I have the answer here? I'm just some pseudonymous avatar, what right do I have to have "the" answer? But of course from my perspective, I'm not an avatar. I'm me. The same me who is used to being the guy in the actual brick & mortar room and giving people answers every day. I'm not considered arrogant when I do it there, but y'know, I could be. There's a huge presumption to it.

In the end its ethos/pathos/logos stuff. In person, in the places where people normally see me, I typically don't have to work hard to convince them I know stuff. Out in cyberspace, clearly it can be much harder.

Doc, there's nothing advanced about RAPM. No offence, but i don't understand why everytime it comes up, you act like people who don't subscribe to it..."just don't get it". We get it, the problem is that its very process revolves around lineup based datasets that can't be used to find individual influences. The function of RAPM as it's conducted is to find trends in rotations/lineups through the course of a NBA season. That's literally what it does. What people call "noise' is supposed to be there, and all they have to do is analyze the specific team's lineups and players to see what caused it.

And sorry, but at no point in your reply did you address WHY Divac was rated above the other centers. Of course, I gave you the answer in this post, but that leaves you with two choices. Either you're stuck with it just being "noise", or you admit that RAPM is nothing but a reflection of rotations. Either way, using it for individual comparisons becomes void.

[acrossthecourt]

You're implying that Boston wasn't the favourite to win that year, that people were in some way surprised. That's not the case. I thought they would win it all before the season started, and I *think* the Vegas oddsmakers did too.

FWIW, I got the Celtics at 15 - 1 to win the title at Mandalay Bay in September or October of 2007. Comparing that to this year's odds, Vegas has five teams at 10 - 1 or better. Based on this, I'd say that Vegas oddsmakers had the Celtics as a good team, but not one of the primary contenders.

This also matches with the projections of sites like ESPN, who had the majority of their panelists not pick the Celtics to even make it out of the East, let alone win the title. There's a link somewhere to the old predictions, but I want to say that most projected the Celtics to win their division bow out in either the conference semis or the conference finals.

http://www.basketball-reference.com/blog/?p=6859

Hollinger said 51 wins and Neil Paine said 48 wins.

edit: Bill Simmons said 49-33

Apparently statistical plus/minus was fairly accurate in predicting how good Boston would be. There's another point in the plus/minus camp. It's not a useless stat and you can do a lot of useful things with it....

Okay, but could you address the parts about Robinson and the 90's RAPM data? I want to know not only for this debate, but for general use, like comparing Marbury to Kyrie Irving (pretty similar RAPM numbers from that site through their first three seasons).

The 90s "RAPM" data on Engelmann's site that you linked are all estimated data, and not RAPM at all. They are, I believe, a version of SPM that uses box score data to try to estimate what the RAPM might have been. Not the same thing.

As far as I know, the only site that has any actual RAPM data from the 90s is AcrossTheCourt's blog. The play-by-play data is available back to 1997, so the first prior-informed RAPM data (which is what is generally the standard for non-boxscore measures) is for the 1998 season.

Doc MJ's now much-cited spreadsheet uses AcrossTheCourt's RAPM values from 98 - 2000; there is no publicly available 2001 RAPM; the 2002 RAPM data is from about the last 1/3 of the season + the playoffs (Englemann's site), and then all of the other RAPM data on that spreadsheet come from Englemann's prior-informed RAPM data up through 2012.

He approximates RAPM by using quarter by quarter scoring margin and minute totals, I believe, as well as who started the game.

[acrossthecourt]

Okay, I'm pretty tired of this "RAPM isn't very useful because it's only finding rotation trends."

First of all, what does that even mean? It's adjusted for strength of opposition and who you're playing with. If you play with a defensive unit, those defensive players will have good defensive values and it won't assume you're a good defender either.

If you're talking about collinearity, then yes, that's a problem, but that's why we use priors and RAPM.

"6th men like Manu/Odom will rate high. "

Manu started every game in 2005 and was first in prior-informed and NPI RAPM in 2005. He wasn't a sixth man then. He was a part of their crunchtime lineups (he has been for a long time) and was often the primary ballhandler in close games.

Yet advanced plus/minus metrics perform very well at predicting future wins, particularly when there's a lot of roster turnover.

You see what that means? That even with a new context the player values attached to RAPM are useful? It's not rotation trends. If it was rotation trends, then when we tested the metric out of sample with future seasons it wouldn't perform well. But it does well.

It's not a relic of rotational weirdness. But you know what? Other metrics have problems like that. Jason Collins sets a screen, gets Vince Carter open, doesn't get credit for it. Jason Collins plays Dwight Howard well, holding him to 16 points on middling efficiency without much help. He does this without blocked shots, so he doesn't get credit for it in the box score. PER ignores him.

JJ Hickson crashes the boards, picks up rebounds, sometimes steals them from teammates, hits a lot of shots inside, plays terrible help defense but still picks up steals, and looks great on PER. Yet when he's replaced with Robin Lopez Portland jumps up 20 wins (it wasn't just Lillard's improvement.)

Carmelo outscored Durant in 2013. Who was the better scorer?

Alex English led the 80's in points. Who was the best scorer in the 80's? And best player?

Adrian Dantley looks comparable by PER to Bird from like '80 to '86. Who was the better player?

Chris Andersen led the 2013 playoffs in WS/48 minutes. Was he the best player?

Zach Randolph had more Defensive Win Shares than Noah in 2014. Who was the better defender?

Boozer was 13th and Rudy Gay was 17th in Defensive Win Shares. Were they great defenders?

All stats have problems. But we shouldn't ignore every stat because they're not perfect.

You can nitpick all you want, find the weirdest results and make fun of them, but that is no way to rate a metric or system. You do that systematically with EVERY result (if you can.) No one's saying RAPM is perfect. There are flaws. There's noise.

But to play down some consistent patterns on a guy who won an MVP makes it seem like some people are just not open to the idea of Garnett as a top player or RAPM in general. It won't matter how accurate the stat is or how pertinent it is to rating players. And it won't matter what new information people will unearth about Garnett.



By the way, let's not be so confused by Magic's rating here. ElGee's championship odds are illustrative here: after Jordan/Kareem/Russell, it's really close between the next group of guys. The difference between the fourth guy and the tenth guy is smaller than you'd think.

[Doctor MJ]

Now, my first quibble here is that the 'proof' about Dirk's superior focused entirely on scoring, which is only one part of offense. I certainly don't deny that in this case that one facet of offense is enough to give Dirk the overall nod on offense, but it's the imbalance of the argument in striking because Garnett's not only better on defense than Dirk, he's also better at all the other facets of offense, yet the weight of the argument seemed to lump the 90% of the game that's not individual scoring into something that can be waved away.

I don't agree that Garnett is better at 'all the other things' that go to make up offense other than shooting. Dirk turns the ball over less. When Dirk sets a screen the defense basically can't guard that because of the threat that Dirk is going to pop out to the three-point line, regardless of whether that actually happens on that play or not. The THREAT of that happening completely distorts how defenses have to play Dirk on the pick and roll. Not so with Garnett, who is a fine midrange shooter, but didn't have Dirk's range. Garnett does pass the ball more than Dirk, but I think that's partly a function of Garnet''s unselfish temperament, and partly a function of Dirk being such a good shooter that it's the correct play for him to take the shot more of the time.

Fair enough.

My assessment of the +/- data we have going back into the late '90s is that there are 3 guys clearly above the rest: LeBron, Shaq, and Garnett. If Garnett doesn't actually belong in such rare air, the question then becomes to explain what the bias is. Simply calling it "luck" is not sufficient, when we talk about this much information.

So does this mean that we all made a terrible mistake voting Duncan in as high as we did?

Well I don't know if I'd characterize anything as a 'terrible mistake' here.

Also my pre-list had Duncan 6th, so him getting voted in 5th really, really doesn't bother me.

That brought us to the acid test though: If we could see Garnett succeed overwhelmingly with talent around him that didn't justify such lofty expectation, then it wouldn't make sense any more to assume that the Minny supporting cast was just another normal supporting cast.

And of course that's what happened in Boston.

You're implying that Boston wasn't the favourite to win that year, that people were in some way surprised. That's not the case. I thought they would win it all before the season started, and I *think* the Vegas oddsmakers did too.

Here's the NBA preview for the year from ESPN:

http://sports.espn.go.com/nba/preview2007/

The average pick of the experts there is for them to be the #3 seed in the East, and Bill Simmons the Celtics homer basically doesn't even seriously consider them in his title prediction.

Note also the talk of the Celtics as an offensive team.

If you saw a championship coming for the Celtics based on their defense becoming huge, then you're far, far, far from the norm.

If you go and look at the Boston roster the the previous season, and ordered the players by DRtg, the Celtics literally kept every good defender, and traded every bad defender away. Again, this is analysis that you can do before their championship season even starts.

So I don't agree with what you are saying. YOU might have assumed that the Celtics were going to win with offense, but I certainly didn't.

I'll reiterate my last line: If you saw this all coming then you are vastly smarter than all the guys at ESPN put together to say nothing of myself, and I just can't figure out how you can be that smart but not remember that all the talking heads were disagreeing with you when I remember it plain as day.

Now that's not entirely fair. It's certainly possible for one to know basketball and ignore ESPN guys, it just seems a bit unlikely to me that this was as clear cut for you as you make out.

Certainly the Boston Big 3 was a big story coming into the year, and certainly the question of whether they would be contenders was a debate. It's just that the cool heads in every room were saying it was pretty unlikely, and I don't recall even once someone saying anything like, "You don't understand, the defense you're going to see when Thibodeau gets ahold of Garnett with the guys that are already there is going to be a whole new thing. It's going to set a new standard that teams will be looking to emulate all over the league."

While that seemingly hyperbolic talk might seem a lot for me to ask from prognosticators, the thing is that the Boston team doesn't do what it did without that. This isn't a by-hook-or-by-crook situation. The Celtics found one particular method toward greatness, and if you had ask the prognosticators ahead of time, "Say Thibs & KG make a world beating defense, do you think the Celtics can win the title?" - basically spoon feeding them the epiphany, they'd have said "Well sure, but why would you think that?".

It's no shock that the Celtics could be what they were given the defensive success, but that also means that any prediction that expected the overall success without going hyperbolic on that specific almost certainly just got lucky.

If you hand't realized this before, it should make your jaw drop. Again, it wasn't supposed to be this way. No one thought it would be this way. It happened because Garnett was able to slide over and let his inferior teammates do their thing, and create a juggarnaut whenever he was healthy.

I think are applying an over-correction with regards to Garnett. When you say 'nobody' you're clearly wrong, because, well, I'm a counterfactual.

It seems to me that you underrated Garnett prior to him joining the Celtics, realised that he was better than you thought he was, and have now applied an overcorrection to him.

So I"m wrong both times while you've been correct the entire about him and also predicting a defensive juggarnaut that would win a title where no one at ESPN saw it coming.

Again, my hat's off to you.

For the life of me, I just can't fathom how one can acknowledge this, and think there was some issue with Garnett in his ability to lead a champion. Between that and the epic overall lift he provided Minny, what's there left to address?

Look, Garnett was awesome, incredible, all of those things. But the thing is, so was Karl Malone. So was Charles Barkley. So was David Robinson. So was Hakeem Olajuwon. Some of those guys lucked into good situations like Garnett did, and won a ring. Others ran into MJ.

The thing is, how do we work out which of these worthies are worth voting in at #8?

A post lauding Garnett doesn't help me choose between these guys because ALL of them have gaudy resumes that when looked at make them look amazing.

Oh I see. Well then magicmerl, I apologize for wasting your precious time.

[Texas Chuck]

But to play down some consistent patterns on a guy who won an MVP makes it seem like some people are just not open to the idea of Garnett as a top player or RAPM in general. It won't matter how accurate the stat is or how pertinent it is to rating players. And it won't matter what new information people will unearth about Garnett.

I haven't been one of the people trashing RAPM or KG. But my issue regarding KG and RAPM as it relates to other players already being discussed or soon to be(Dream, Mailman, Admiral, Chuck to name but a few) is this:

We have this RAPM data for KG that we simply don't have for a lot of the players from a prior generation. And that RAPM data for KG is outstanding and thus I think there is a real danger of starting with that knowledge(that we don't have for other great players) and then finding reasons to argue him above them because we have "proof" of his greatness. Does anyone really think that Admiral in his prime isnt going to have amazing RAPM numbers if we had them? Note: Im not accusing anyone of doing that intentionally, but we would be naive to not at least ask the question of what influence this information is having on how we see him in comparison with those we don't have the data on.