Doctor MJ wrote:
As I'm out of town, this is my pre-recorded argument for:
Vote: Kevin Garnett
Basically, once Russell, Jordan, and Kareem are inducted, Garnett's my guy. This is a drastic statement no doubt, but some of you undoubtedly saw this coming. I made a thread basically about this on the Stats board a while back.
I'll say up front, that this argument is statistically driven, and driven predominantly by RAPM. I understand that that won't convince everyone, and I really don't expect Garnett to get inducted that soon, but I hope people will understand why I find he and the stats in this case so compelling.
First a little history:
As I'm sure everyone knows here, Garnett once won an MVP. At that time, while it wasn't a given that he was better than Tim Duncan by any stretch, the debate between the two of them was heated. But then it faded - why? Because Garnett's team faded. The team results dictated the perceptions of the individuals.
The unstated assumption here, if we are to assume a rational basis for this, is that the prior perception of Garnett's value was wrong, that it was instead about a team context he was lucky to have, and crucially: That that same type of perception about Duncan was and remains true.
Now, those with a serious grasp of NBA history already saw the issue with this: Garnett wasn't the first superstar to play on a weak team, sometimes bad luck is just bad luck. As such, the new perception of how much impact Garnett could give was actually RIGHT, the error is simply in overrating what Duncan or other superstars can truly do because you've only seen them in team success.
At the time though, I wasn't one of the wise folk. Though I'd followed the NBA basically my whole life, I was only just discovering RealGM, and it was this site that let me to really start applying my critical thinking to this entertainment. As such, I let my opinion of Garnett fall like so many others. I did however at least recognize the possibility of an experiment: If it was all truly about bad luck for Garnett, then all Garnett would need is a solid supporting cast, and the results would be phenomenal. And of course that's what we got in '07-08.
Some don't remember how impressive that actually was. They see it as a superteam, in the sense that they see 3 huge talents coming together as being something that obviously should be amazing. This is understandable because the team WAS amazing, and because right from the start Danny Ainge and company hype the team saying it would be amazing...but this is not what experts thought.
What experts saw was 3 stars, all used to being the alpha offensive threat on their team coming together. They also recognized Garnett as a serious defensive player, but the scouting on the team was that offense would be the strength, but that would not be enough to make them a serious contender. Add another 20 wins? Sure, but that's still not even 50 total wins.
Instead it was more like 40 wins worth of improvement, and the strength was defense, and of the stars, it was all about Garnett shifting with unprecedented versatility. And once people here really started analyzing the consequences of this, it became harder and harder to take seriously the notion that Garnett was a tier below Duncan as a player.
Still though, there's a clear tendency to acknowledge a similar tier, but give Duncan the tiebreaker. Even if you have a significant gap between the two on your GOAT lists, you could say stuff like "oh but the competition is just so fierce, I"m not saying it's a blowout, there are just so many good players!". What I found though over time, is that Garnett just seemed to come out looking better than Duncan.
So to some numbers. Here's the spreadsheet I made compiling together RAPM numbers over the years:
https://docs.google.com/spreadsheet/ccc ... _web#gid=3
And here are the top 5 guys of the RAPM age based on their average of their Top 5 RAPM years:
1. Kevin Garnett 10.87
2. LeBron James 10.47
3. Shaquille O'Neal 10.35
4. Tim Duncan 9.25
5. Dirk Nowitzki 9.13
Now, I don't rate guys just by these numbers. Among the more obvious things: We don't have complete data, the shape of one guy's 5-year prime is different from another guy, minutes played, team context, etc.
I actually rate the peak of both LeBron and Shaq ahead of Garnett...but that gap between Garnett and Duncan is just plain consistent. It's been there since we first looked at +/- numbers, and it's never gone away. At a certain point I just had to admit to myself that I didn't have any good reason to give Duncan a "tiebreaker" because it wasn't a tie.
Okay now, if we accept the values but debate what all they say, what does that bring up?
-Well, some have argued that perhaps Duncan playing on a better team with a better coach was somehow a disadvantage here. This has never really gotten traction in my mind though. What we've seen in the years following Duncan's peak is not that Pop & co are plucky and can minimize the falloff caused by Duncan's waning abilities...but rather that the offense just works better with Duncan in a lesser role. This implies that the RAPM numbers might actually be OVERrating Duncan.
And of course on the defense end, we've now seen Garnett put up better defensive RAPM numbers on a defensive team of at least comparable defensive prowess. The very argument brought up to perhaps support Duncan then ends up a knife driven right back around against his candidacy.
-What about the different types of APM? Don't they disagree? Well, something that confused me for a while is that NPI RAPM seems to prefer Duncan over Garnett while the PI RAPM I"m using prefers Garnett. What's up? Well, an issue with RAPM is that it essentially looks to dismiss outlier data, and the less sample and prior used, the easier it is for real trends to be dismissed as luck. Garnett has the advantage in pure APM - which doesn't have this issue, and in prior-informed RAPM which has less of this issue, so to me the best explanation I've heard is that where Duncan has the edge it's squarely due to a particular flaw in RAPM if you take a bad snapshot of it.
-What about XRAPM? Bringing this up in part because I've seen it's dismissal mentioned in this project.
I want to be clear I don't utterly ignore XRAPM, my real issue is that I want access to both RAPM and XRAPM. The latter isn't a replacement for the former. Fundamentally, statmakers should understand this. Practically there's just a matter that it takes me time to get a sense of a stat, and if you just throw a new stat out there, I may not feel comfortable doing much with it.
With XRAPM and it's use of box score stats, there's a bias in there that bothers me, and I'm still not entirely confident I have the hang of how it's contortions manifest. Maybe someday I will, and it will change my mind.
It is worth noting though that it's not like XRAPM here gives a mega advantage to Duncan here. Garnett has the edge in his MVP year, and he has the advantage when he goes to Boston, and the '90s data he lists isn't even real XRAPM. So even if you prefer XRAPM to RAPM, it paints a muddy picture here. Maybe the muddiness lets you keep giving Duncan the slight edge over Garnett though it doesn't for me, but it should not give any kind of blowout.
-You'll note my fixation on Duncan here. In part that's always because these two players are so closely linked. In part it's because Duncan's riding a new wave of enthusiasm. The pre-list Top 50 saw Duncan rise to the 4th spot while Garnett lingers at the 13th. I actually don't think that placement of Duncan is all that crazy...but the increased separation over Garnett is an issue.
As we all take in how awesome the Spurs culture is and give Duncan some credit for that where other stars wouldn't have let that be possible, it's not reasonable to say Garnett is one of those problematic egotistical stars. Garnett was a loyal soldier in Minneosta, made massive changes in his game that led to some wrongly thinking Pierce was more important in Boston, and was an inspirational presence for years there.
-Circling back to LeBron & Shaq:
Obviously Garnett's edge over LeBron is based on longevity. I don't need to go into that comparison to convince you, but Garnett's longevity is staggering by any standards. His PI RAPM numbers from year 3 to year 17 are clear cut superstar worthy. I doubt you'll find that from any other player in history. It's what also gives him an advantage over others who have arguably stronger peaks.
Shaq for example has very nice superficial longevity, but not to that extreme, and with all sorts of flare ups along the way.
So, please keep KG in your consideration. Cheers folks!
You're certainly entitled to your opinion, but so is Doc, as is everyone else.
