GoldenFrieza21 wrote:
You have established Garnett had very good RAPM numbers in 2004. I fail to see how that is pertinent to the argument we are having since we do not have RAPM numbers for Kareem.
It is intellectually dishonest to prop someone up in one criteria when there is no available data for someone else in the same criteria.
I don't see this as intellectually dishonest. We have 20-ish years worth of RAPM data, and among all the players that have come and gone in that span, Garnett's marks in '04 basically put him in the conversation for "GOAT"-level (or greatest of the last 20 years). bbref provides raw on/off numbers for the last 15 years; Garnett's in '04 are again GOAT-level. In other words, he's in the conversation for greatest single season impact of the last 20 years. The
implication is---and that's all it is: an implication; it's not proof, especially since as you say we don't have this data for Kareem; but it's nonetheless relevant---is that his impact is likely to rival (or at least be in the neighborhood of) that of peak Kareem.
Because he's right there in the neighborhood of Shaq, Lebron, and Duncan, it stands to reason that he's
probably pretty close to Kareem as well. I don't think that's an intellectually dishonest stance. As an approximate, that's fairly sound and logical reasoning.
And especially if we're attempting to qualify his defense vs. Kareem's, given the unique character of KG's defense, I would consider any argument that did NOT utilize some impact indicators as incomplete.
GoldenFrieza21 wrote:I hate RAPM with a burning passion but I am not going to get into that. I simply believe in being consistent and judging people by the same criteria. We do not have RAPM for Kareem, hence it is not fair to use it in this case (I think we should never EVER use it period because of the disaster that was the top 100 project but I digress on that).
That last statement is a bit off-side. You're labeling something a "disaster" presumably because the list bucks against conventional "truths" about the hierarchy of basketball greatness. But just because it [in some places] runs counter to convention (conventions which you, personally, may have held as relative "truths" for some time, thus making them harder shake), does not make it invalid.
Anyone can regurgitate a conventionally "accurate" all-time list. But a few dozen [mostly] very well-informed, very studious and thoughtful fans of the game put copious amounts of time into study, analysis, game-watching, reading, and debate to come to that result (by consensus). And that warrants consideration, even if it doesn't match the regurgitated status quo.
If I can be a touch melodramatic for a moment, for the purpose of making a point: once upon a time, conventional wisdom had it that the world was flat; and even later than that conventional wisdom had it that the Earth was the center of the universe. And the first people to suggest otherwise (particularly to the latter premise) were shouted down and perhaps even made outcasts or branded as heretics by the best scholars of the time.
Now I do not wish to compare the minds participating in the top 100 project to some of history's greatest minds (though maybe some of them are, idk), nor to compare the relative importance understanding the universe to that of an all-time greatest basketball players list.......but the basic circumstance does apply to both: humans are resistant to new information (especially if they can't easily understand it) which would attempt to disprove things they've held in their minds as "truth".
RAPM is just such a tool. It suggests things that run against the conventional "wisdom" about who's important out there on the court (e.g.
what? the guy who scores the most points isn't the most valuable player? Absurd!!). It's a tough thing for casual fans to wrap their brains around (mostly just because they're not willing or interested in putting in the time), that the guys with the biggest volume stats are not always the most important. Raw totals are easily counted and easy to understand. RAPM is not.
And I don't know how it could/should
possibly be seen as a worthless metric which should "never ever be used". How can I explain or qualify that statement......
OK, let's start with on/off stats. on/off numbers are not "mathematically manufactured".....they simply
are: Player A was on the court "a" number of minutes and off the court "b" number of minutes. During that "a" sample, his team averaged x number of points per 100 offensive possessions, and allowed y number of points per 100 defensive possessions. And during that "b" sample, his team averaged z number of points per 100 offensive possessions, and allowed * number of points per 100 defensive possessions. Those differences between the "a" sample and the "b" sample gives us his on/off lift (or drag) on both the offensive and defensive sides.
Not quite as easy to tabulate as points and rebounds, but still basically simple and directly quantifiable. This hasn't been plugged into some complex equation involving modifiers and standard deviations, etc, yet. As I said, on/off number simply
are what they are.
But wait, at any point in time when he's on the court, he's on the court with four teammates. How do we know how much (if any) of the lift (or drag) the player in question is responsible for, and how much is due to his teammates? Well, we don't know for sure yet (though it's interesting how relatively consistently the on/off results do tend to suggest what you'd expect for some of our recent greats like Lebron, Wade, Shaq, Duncan, Garnett, Chris Paul, etc). And that's what RAPM attempts to define.
I don't pretend to understand all the math by which RAPM attempts to do this. The basic premise (as I understand it) is that it looks at the on/off numbers for everyone and, based on minutes played, estimates the amount of time each potential rotation is on the court, how each rotation performs, and then thru the complex magic of mathematics attempts to filter thru all that line-up "noise"; by seeing which rotations perform the best and seeing who the common denominator(s) is/are in those most successful rotations----and in some models this might be partially/slightly informed by box metrics, though I'm not sure about that----doles out
individual credit.
Looking at the question RAPM attempts to answer---What portion of team's success (as measured by the point differential) is this player
directly responsible for?----and assuming that it's
even somewhat successful in answering that question.........How on Earth could that be considered anything other than a
fantastically relevant and useful metric?
Does RAPM do a perfect or exact job in answering that question? Is it a perfect stat? Almost certainly not. But the same is true of most of the other stats we have available. DBPM, DWS, and DRtg would have me believe that Joe Dumars was a below-average defender
for his entire career, for example. Individual DRtg is a HIGHLY unreliable stat in general, imo, except for bigs whose primary defensive value is as a rim protector (there it does a reasonably job). Anyway, there are multiple other examples of box-based metrics leading us toward something other than truth.
The results yielded in RAPM studies appear to be in keeping with "conventional wisdom" and a lot of our box-based metrics, eye-test, etc often enough to assume it's doing a reasonable job of what it attempts to define. And where we sometimes get "aberrant" results, you'll often find there's some fairly obvious explanations ("intangible" things which are barely perceptible on box-based stats): Shane Battier is extremely valuable despite barely mediocre volume stats because he can defend the perimeter or post, fights hard thru screens, does great at closing out on shooters, never falls asleep on defense or misses a rotation, etc; Amare Stoudemire has a lower than "expected" RAPM because the majority of his offensive gains are given back on the defensive end--->he doesn't fight his man off the block for post position (but rather just lets himself get posted up), he consistently misses defensive rotations, he's not a good pnr defender, he doesn't box out well and has generally been a mediocre rebounder, etc.
Is RAPM relied upon to an excessive degree and/or misused by some posters? Without a doubt. But then, that too is basically true of most other stat metrics. A backlash reaction of saying we should never EVER use RAPM is......well, frankly it's unreasonable. If the above plea hasn't convinced you that it has some valid application in player analysis, we'll have to just agree to disagree.
But something isn't so just because you say it's so. And you need more justification than merely pointing out that it gives you a different answer than PER or the boxscore, because.......because the Earth isn't flat.
"The fact that a proposition is absurd has never hindered those who wish to believe it." -Edward Rutherfurd
"Those who can make you believe absurdities, can make you commit atrocities." - Voltaire
