AEnigma wrote:DraymondGold wrote:His statistical profile is low for a GOAT-tier candidate. Let's take the Top 12 peaks from the latest project (Jordan/LeBron, Shaq/Kareem/Wilt/Duncan/Hakeem/Russell/KG, Magic/Curry/Bird)...
-RAPM: We have actual RAPM data from 59 games pre-97 (in 85, 88, 91, and 96), thanks to the great work of Squared2020
. Small sample, non-peak years, but still... Hakeem appears in the top 30 in the league only
once, and never gets in the top 15. Despite the small sample, it's still large enough for players like Jordan, Magic, Kareem, Bird, Moses, Robinson, and Barkley to all clearly have Top-10 level seasons.
I have quibbles with weaponising those other metrics against him, but I doubt that would be productive (although I did find it funny how specifically this time around you decided to bail on those CORP evaluations you usually value so much
).
It's a shame you so often start debates by assuming intellectual dishonesty in your opponent... I've done my best to set the tone better than that, but you seem set in your ways on this habit....
Regardless, if you'd like me to add CORP, I'd be happy to!
-CORP: Hakeem looks better here! Unlike the other 4 metrics, where he universally is in the bottom half (and frequently at / near the very bottom), here his 1/2/3 year peak is around #4/12!... while still having clear separation from the GOAT tier. So if his best stat is sub-GOAT level, and all other stats are noticeably worse.... then there's still not much argument statistically for Hakeem being GOAT-level.
But using RAPM in this way feels much more severely objectionable.
Say I take a “random” sample of Lebron’s NPI RAPM.
2005: slots in right around #60 even though he is averaging 27/7/7.
2008: #11
2014: #20
2018 (no postseason): #36
… damn, Lebron’s all-time impact suddenly looks pretty questionable.
Now imagine if we only had samples of 20% of games. How bad could I make him look?
Hakeem does not seem to show absolute top of the line regular season impact, no. At best, he might have a speculative one season at the top of the league in the regular season (1993) and a somewhat less speculative and possibly generous second place in 1994. But this is not a genuine way to frame that argument, relying on single-season RAPM samples from random fifths of seasons that might not even crack his personal top five.
I was pretty clear about the caveats of the sample size and year choice. Nothin much new here, but yes the sample size and year choice is poor.
So let's give a bit of consideration to this counter. If there's limits to including RAPM (when there's a poor sample), why include it at all? Why consider it?
The logic goes something like this: if we're to argue Hakeem's peak is GOAT level (and even moreso if we're to argue his career is GOAT level), we would expect (but not require) him to have GOAT-level or near-GOAT-level impact in a stat like RAPM, which is pretty much industry-standard good stat to use for judging impact. If we have limited samples from prime but not peak years (as we do), we would prefer (but again not require) him to have at least all-nba / MVP level impact in RAPM, to say nothing of strong MVP level, all-time impact, or GOAT-level impact.
The fact that he doesn't show anywhere near GOAT, All-time, or even strong-MVP level impact in the RAPM data we do have should give us pause. It doesn't invalidate the entire argument for Hakeem! But it should raise some eyebrows, make us consider whether there are biases in the sample, make us question whether there's biases in our previous judgement of Hakeem (e.g. if his reputation as an offensive player overstates how good he actually is), and most importantly... it should make us dig deeper.
For example, we might look at his other impact metrics (e.g. WOWY, PIPM-estimate, BPM) to see if those rate him higher. If Hakeem rates quite poorly (for a GOAT player) in all the stats we have, even if there's flaws in each individual one, that should give us much greater confidence that there is indeed something flawed in the idea that Hakeem is GOAT level. If we want to argue that Hakeem is GOAT level in both peak and career, we should be able to explain why all the stats across the board undervalue him.... for example, we'd have to find film analysis that makes Hakeem look so much more impressive than the other GOAT candidates that it justifies putting Hakeem in contention with the players who appear like GOATs both in the film and in the stats.
Personally, I just can't see interpreting the film of Hakeem that favorably to put him in contention with Jordan/LeBron/Kareem. But if you'd like to make the case, I'd love to see it! Always happy to see more film analysis on this board
Once we have full-season RAPM data in 1997, Hakeem's not even in the Top 50 in the league.
Yeah you should source that because mine has him in the top thirty and roughly on par with Penny Hardaway, and that is of course with all the usual role-player filler (Terry Mills!).
Happy to share the source. I used Goldstein RAPM, which is pretty standard version of RAPM:
https://docs.google.com/spreadsheets/d/1eK0i6L0q2Brih5nKOKZLGHVofY0JWKOlnnEaSMu1LTM/edit#gid=0. I searched for Hakeem, found his 1997 year, then searched for 1997 to find how many seasons rank above Hakeem's. When I search the spreadsheet, I find 61 years better in 1997 -- let me know if I miscounted!
Again, it's a small sample and in non-peak years, but Hakeem is last by a large margin in the RAPM data we have for the Top-12 peaks (which is only missing Russell/Bird)
Oh is it missing Bird? Because I could not help but notice that 1985 and 1988 Bird come across as quite underwhelming in those RAPM samples, even though those samples are much more complete than Hakeem’s and much more tied to his actual peak.
I respect Squared’s work in trying to put that together, but I wish he had held back on sharing any data that was far short of usual sample standards, because this scattering of games has so far pretty much only worsened discourse.
Good catch! I meant to type Wilt. It's missing Wilt/Bird. Edit: Russell/Wilt!! Wow, I just couldn’t type it right.
But to your point, yes Bird does rank worse than expected... still better than Hakeem at his best, but worse than expected. It'll be interesting to see if Squared2020 has any more games to add!
Like you mentioned, his team performance is also lower (likely partially because of worse teammates). Bt if you were to try to convince me that he had a GOAT-level peak in 93-95, I'd also probably want better team dominance. We have to be careful not to equate the team's performance to value of their best player, but peak Hakeem's teams (93-95) are last in postseason co-Net Rating (the new Backpicks metric) among this tier of players, near the bottom in postseason relative Net Rating, and at the very bottom in overall SRS (regular season + postseason). And this is if we only look at years where he had a better cast.
… Better casts that were still well shy of the standard of everyone else in that top eleven group, while facing overall brutal SRS postseason competition. Getting the average of those rosters performing at a, what was it, 7.6 SRS postseason level is more impressive than the “floor-raising” I have seen of basically anyone shy of Lebron.
I have a different interpretation of the postseason success... I'll try to post if I have time.
AEnigma wrote:LukaTheGOAT wrote:Weird framing because Hakeem was still excellent in 1988-92 even if you prefer Jordan in those seasons, while Jordan was absent for all of 1994 and basically all of 1995.
Yeah and Jordan was another level from 88-92 lol. I said all-time level years, and I'm not eager to push 88-92 Hakeem into the same territory as Jordan. I didn't mention 96-98 Jordan where Jordan was clear of Hakeem either because I didn't think that stretch was in the same tier.
And as I said, I think those all-time level years are exponentially more valuable than just MVP level years. Hakeem's extra years don't do it for me. Same is true when comparing MVP years to All-NBA years, etc.
Except we are comparing two all-time years to one nothing year and another near nothing year. If you and you specifically are grading 1988-92 as like a 100, no complaints, whatever, and say 1988-92 Hakeem is on average only like a… what… 70… okay, that gives your individual assessment of the two a solid 150 cumulative value difference, or something like that. Arbitrary numbers but illustrates a point if you sincerely think that Jordan over that period was close to 1.5 times as valuable (which I would not agree with in the slightest, but that is tangential to this exercise).
That cumulative value should be pretty much entirely wiped away by 1994 and 1995, and that is without throwing 1986 in there too.
This would be taking a "linear" approach to career value (i.e. weighting longevity >> peak). This is far from what most people do (most people give extra weighting to having a higher peak), and there's strong evidence that we should actually not take career value linearly.
The best article I've seen on the topic can be found here:
https://thinkingbasketball.net/2018/04/13/goat-meta-thoughts-and-longevity/. It's a bit out of date from Thinking Basketball's recent rankings (e.g. since the article, he has LeBron passing Kareem), but it should explain the idea in more detail.
