RealGM Greatest Peaks Project (2022): #3 - 1999-00 Shaquille O'Neal

[Doctor MJ]

what makes you pick curry over the all-time bigs?

In patciular Kareem iirc was considered the best player itw in college and Hakeem has a bunch of seasons where he was arguably as or more impactful in the regular season(alsohas a higher average on playoff pipm despite playing much longer).

Duncan maybe not as good in the rs but idk curry was ever better than duncan was in 02 and 03.

I think you could argue hakeem and kareem were better regular season and playoff players while duncan was a better playoff player. Is your pick because you think they would struggle in the modern nba?

It's a really hard comparison to make and I don't want to imply that I think there's some obvious massive edge here, but yes, me thinking about these players in the state-of-the-art NBA very much influences me.

I find the Curry vs Olajuwon debate trickiest because I expect Olajuwon would be a DPOY candidate today, but all the other guys I'd expect to take a clear step backwards - though to varying degrees, with Shaq being the guy who'd be in the most trouble.

[OhayoKD]

what makes you pick curry over the all-time bigs?

In patciular Kareem iirc was considered the best player itw in college and Hakeem has a bunch of seasons where he was arguably as or more impactful in the regular season(alsohas a higher average on playoff pipm despite playing much longer).

Duncan maybe not as good in the rs but idk curry was ever better than duncan was in 02 and 03.

I think you could argue hakeem and kareem were better regular season and playoff players while duncan was a better playoff player. Is your pick because you think they would struggle in the modern nba?

It's a really hard comparison to make and I don't want to imply that I think there's some obvious massive edge here, but yes, me thinking about these players in the state-of-the-art NBA very much influences me.

I find the Curry vs Olajuwon debate trickiest because I expect Olajuwon would be a DPOY candidate today, but all the other guys I'd expect to take a clear step backwards - though to varying degrees, with Shaq being the guy who'd be in the most trouble.

I'm not saying it's a bias(because you might have a good reason), but it does seem like you have a preference for guards to bigger players because you also picked mj #1 over a bigger player with stronger impact signals who plays in the modern-state-of-the-art nba curry does.(curry arguably being a good emulation of what jordan would look like in the current nba actually)

Is it a portabiity thing?

[Doctor MJ]

what makes you pick curry over the all-time bigs?

In patciular Kareem iirc was considered the best player itw in college and Hakeem has a bunch of seasons where he was arguably as or more impactful in the regular season(alsohas a higher average on playoff pipm despite playing much longer).

Duncan maybe not as good in the rs but idk curry was ever better than duncan was in 02 and 03.

I think you could argue hakeem and kareem were better regular season and playoff players while duncan was a better playoff player. Is your pick because you think they would struggle in the modern nba?

It's a really hard comparison to make and I don't want to imply that I think there's some obvious massive edge here, but yes, me thinking about these players in the state-of-the-art NBA very much influences me.

I find the Curry vs Olajuwon debate trickiest because I expect Olajuwon would be a DPOY candidate today, but all the other guys I'd expect to take a clear step backwards - though to varying degrees, with Shaq being the guy who'd be in the most trouble.

I'm not saying it's a bias(because you might have a good reason), but it does seem like you have a preference for guards to bigger players because you also picked mj #1 over a bigger player with stronger impact signals who plays in the modern-state-of-the-art nba rn.

Is it a portabiity thing?

You're talking about MJ vs LeBron? If so, I don't feel comfortable concluding that the data puts LeBron's impact above Jordan's at this time. While I do think that '08-09 LeBron made have the GOAT impact season by some measures, I have reasons why I don't feel comfortable calling that LeBron's peak, and by and large LeBron hasn't been a guy leading ultra-outlier regular season teams the way Jordan was.

However, as I get more data, my conclusions may change.

[falcolombardi]

So I'm trying to think through this with Hakeem. In the regular season, from the limited +/- stats we have, he seems to consistently be a top 5-10 player in the league.

For the most part he elevates his level of play in the postseason, and looks more like a consistent strong MVP than an All-NBA guy.

Eye test wise, he looks like the most impressive player I have seen, with his combination of size, mobility, and skill.

Hard to reconcile.

Something I am curious about...for people selecting Hakeem (who I think has a case for GOAT peak):

(1) How do you treat his passing?
(2) How much better do you view his defense compared to Duncan's?

I think offensively what we're talking about is a scorer whose approach is less optimized for what's available in the regular season, and more optimized for having that extra counter you need against tough defense. He's not alone in this - I've talked about Kobe being like this too.

(1) I've given my thought on his passing - I think it was good enough to take advantage of spacing well, so I tend to think - at least for his peak - this wasn't a massive weakness. Great passers obviously have a great edge over him here, but in a typical debate, this wouldn't loom large for me.

(2) At his best, considerably better. I don't think Duncan was in the same league at all a quick-twitch athlete. Take Duncan and have him play for a lesser coach with bad defensive talent around him, and his lack of agility will become more noticeable, and even on the Spurs, while people tend to think "Duncan beat the SSOL Suns", the reality is that Duncan very much struggled defensively in those match-ups and it tended to be Ginobili that saved the day when the Spurs won out.

I do think though that Duncan had a hell of a defensive career (and overall career too, obviously) because he was able to keep doing his thing within Pop's context very effectively until he retired.

So doc, i hope you dont take this wrong as once more i find myself nitpicking your,well, picks.

But there are some areas where i feel you are taking them a bit differently with different pmayers to argue them up and down for example curry vs duncan

With curry you were very high on his consistency allowing warriors and kerr to develop a stable system/cultue around him. With duncan here you seem to be using the same thingh to diminish duncan merit (that he got to remain in a stable system with pop developed around him)

With curry you are critical of people using 16 curry struggles to anchor their view of hia other season, with duncan here you seem to be using struggles against phoenix against him even though the peak in question is 2002-2003

Here you mention ginobili being the best player against suns as somethingh against duncan when curry has his own impact metrics monster teammate who often had better plus-minus metrics than him (draymond)

In fact i think you give curry a lot of credit for draymond sucess cause he thruved under the focus on curry but dont do the same for duncan with ginobili

I remember you being critical of duncan never having an all time level team on the level of the 17 or 18 warriors when he never had ginobili, parker amd kawhi primes coinciding with his own (which curry had with durant,dray and thompson)

Not triying to attack your arguments doc just not sure you are giving a conpletely fair shake to duncan in a comparision like this

[OhayoKD]

It's a really hard comparison to make and I don't want to imply that I think there's some obvious massive edge here, but yes, me thinking about these players in the state-of-the-art NBA very much influences me.

I find the Curry vs Olajuwon debate trickiest because I expect Olajuwon would be a DPOY candidate today, but all the other guys I'd expect to take a clear step backwards - though to varying degrees, with Shaq being the guy who'd be in the most trouble.

I'm not saying it's a bias(because you might have a good reason), but it does seem like you have a preference for guards to bigger players because you also picked mj #1 over a bigger player with stronger impact signals who plays in the modern-state-of-the-art nba rn.

Is it a portabiity thing?

You're talking about MJ vs LeBron? If so, I don't feel comfortable concluding that the data puts LeBron's impact above Jordan's at this time. While I do think that '08-09 LeBron made have the GOAT impact season by some measures, I have reasons why I don't feel comfortable calling that LeBron's peak, and by and large LeBron hasn't been a guy leading ultra-outlier regular season teams the way Jordan was.

However, as I get more data, my conclusions may change.

Ig that's fair, though it seems hard to argue defense isn't a differentiator. Bulls became a way better defense in spite of jordan declining there by 91 while the cavs went from elite to bad defensively without lebron. Lebron's also anchored various elite playoff defenses while jordan hasn't anchored any.

I'm also skeptical of the "leading ultra-outlier regular season teams" argument because idk if i buy jordan being the reason the bulls elevated from 50 wins to the goat team in 91. His impact stats are lower in 91, but beyond that he made less defensive plays, made more defensive mistakes, shot less effeciently(vs worse defenses than he faced in 88, 89, and 90), and probably had worse turnover economy considering how much less he was handling the ball. You seem to attribute the improvement to jordan "playing within the system", but i think they just got better. It would be one thing if jordan's volume declined, but his effiency declined too

And, if 89/90 mj is peak mj, then it's hard for me to get past how the 15 cavs(without love or kyrie) were as good in the playoffs as the 88, 89, or 90 bulls. That shouldn't be lebron at his best, right? And that's a team without spacing. But lebron's basically acheiveing the same results just by protecting the paint and passing.

[DraymondGold]

Bill Russell discussion:

4.1962 Bill Russell (1964, 1963, 1965)

Now number 4 is arguably the most influential player ever with how he transformed the way defense is played in the league forever. The greatest defender ever, and the engine behind one of the greatest dynasties in sports history.

There are alot of reasons to believe Russell played a significant part in the Celtics team dominance and many have argued how he has a case for being the most valuable player of his era so I won't focus TOO much on that unless asked to.

Here are a few pretty strong indicators he has:

-We can see it on film and we can read/hear about the era in news articles and from others that have experienced the era.

-WOWY data(also looking at the team pre and post Russell and how the league changed over time).

-Team minutes distribution(how remained constant but everyone around him changed and played nowhere near the same amount of minutes in most years and they were still dominant), etc.

-

But i'll talk about why I believe their team net ratings still undersell how dominant they truly were like I did in the last thread for 2 main reasons.

1. Using the commonly used net ratings is not a true era adjustment - in lower scoring environments a team being worth +5 per 100 has more value, this can be seen when comparing the TS+ framework vs using rTS%.

Real life situations will never be this extreme but here is an example as to why we should use the former

In a league where the average TS% is 10, being +5 would mean you are scoring at a rate 1.5x(150% better) more effectively than league average

In a league where the average TS% is 50, being +5 would mean you are scoring at a rate 1.1x(110% better) more effectively than league average

When calculating net ratings using percentages rather than absolutes, the Celtcs would likely look even more dominant because the era they played in was a lower scoring environment and significantly harder for other teams to make up ground with less PPP available.

Great points! WOWY and related metrics definitely show Russell was the primary driver (you might say he was the Bus Driver... ) of the 60s Celtics. I also really appreciate your comment about absolute vs percentage net rating! That's something I've thought of before, and it always bugged me that we didn't change, since it's such a simple thing to fix (the various websites should just change their formula!). It's worth noting, this wouldn't just help Russell, right? For example, Wilt might also a similar boost for relative True Shooting percentage, team net rating, etc.

That Russell dribbling video you linked is such a joy to watch. The Giannis comparison just pops off the screen. The play where Russell jumps over that guy in transition reminds me of the similar play where Giannis jumps over Tim Hardaway.

I'm still fairly low on his scoring game though. His lack of efficiency and postgame of course pulls him down, but he also doesn't make full use of his athletic potential on rim finishes, offensive putbacks, roll rolling to the rim, etc. People often say Bill Russell would do worse in a time machine (and I do agree his defense would get worse in a later era, if only because it's hard to maintain the defensive volume he had with all-time high minutes and limited spacing). Still, I wonder whether he'd improve his scoring particularly in the more scalable areas I mentioned if he moved to a later era.

You mention he has late-clock bailouts that might lower his efficiency. I haven't heard that argument before! Do you have any metric or ballpark measurement of how often he did a late-clock bailout? Even a ballpark estimate would help us produce an adjusted TS% for him. I'm surprised they would go to him at the end of the clock... I would have thought they would go to their perimeter stars more, but perhaps that's the era difference showing up.

Wilt Chamberlain Discussion:

I haven’t seen as much back and forth discussion about Wilt vs Kareem or Shaq. I think 67 Wilt clearly is the better defender, rebounder, and passer of the three. I imagine it’s the scoring gap in 67 which pulls Wilt down for people? Are there any other reasons (e.g are people less high on Wilt’s playoff performance)?

Wilt is criminally underrated here at Real GM. That's why.

I'd like to once again, post this, just to see if anyone is listening/paying attention.

Note: You have to click the Tweet to see the entire thing.

Read on Twitter

Thanks for the response! It's definitely possible Wilt's underrated. In this board's defense, I do think there's more uncertainty with Wilt given we have less film and fewer statistics, at least compared to the more recent players. I see him as clearly in this Top Tier of peaks and clearly a top 10 peak of all time. If I were lower on his ability to "put it all together" (see below), he might end up on the lower end of this tier, while if I thought he successfully put it all together, he'd end up on the higher end of this tier.

One thing that stuck out to me after reviewing the tweet though: WOW, that's a difficult playoff run! I know you've been doing research on playoff opponent difficulty, where Wilt always is near the top of playoff opponent defense (viewtopic.php?f=64&t=2185164). Thanks for your great research with this! I haven't gone through the most recent update (yet!), but it's been a joy to read so far. But seriously... facing Bill Russell and Nate Thurmond in 11/15 playoff games.... By my account, that means 1967 Wilt played 73% of his playoff games against the GOAT defender or the GOAT Big Man man-defender.

I wish we had better metrics for the 60s! Maybe one day we could get Squared2020 enough games (and time!) to go through and supply RAPM. Pasting my Wilt statistics here (using Curry as the barometer, as I've been doing throughout this project):

Spoiler:

5. Curry vs Wilt: [spoiler]Plus-minus based stats:
Ai. AuPM (no data available for Wilt. 2016 Curry 2nd all time).
Aii. Postseason AuPM (no data available for Wilt. 2017 Curry 2nd all time).
Bi. Goldstein RAPM / Historical Square2020 RAPM: (miniscule sample available for Wilt, far below Curry’s 7th all time).
Bii. Goldstein Playoff PIPM (3 years for sample size): (no data available for Wilt. Curry’s 8th all time)
Additional plus minus stats: C. on/off: (no data available for Hakeem. Curry’s 1st all time)
Additional plus minus stats: D. WOWY: 65-70 Wilt >= 2016-2017 Curry (not sure about full prime WOWY. I brought in 16 Curry because Ben hasn't finished publishing Curry's mid/post-2017 WOWY numbers yet).
Additional plus minus stats: E. ESPN’s RPM: (no data for Wilt. 16 Curry 2nd all time)
Additional plus minus stats: F. Backpicks’ CORP evaluation: 1967 Wilt > 2017 Curry (healthy 2016 Steph Curry 4th all time > Wilt.)

Box score-based data
Gi. Backpicks BPM: 2017 Curry > 1967 Wilt (and healthy 2016 Curry is 2nd all time)
Gii. Postseason Backpicks BPM: 2017 Curry (4th all time) > 1967 Wilt
Hi. BR’s BPM: (no data for Wilt. healthy 2016 Curry’s 4th all time)
Hii. BR’s Postseason BPM: (no Data for Wilt).
Additional box score stats: Ii. WS/48: 1967 Wilt > 2017 Curry (but healthy 2016 Curry 3rd all time > Wilt)
Additional box score stats: Iii. Postseason WS/48: 2017 Curry > 1967 Wilt

We barely have any metrics to compare vs Wilt. Of the limited sample we do have, Wilt and Curry are tied 3 to 3. Curry leads in the postseason-only stats. If we add a neighboring year (2016 for Curry and either 1966 or 1968 for Wilt), Curry comes out on top in 5/6 stats. Still, we’re missing the majority of my preferred stats, so there's more uncertainty here.

[EDIT: may post Wilt vs Kareem and Duncan statistically if I have time here].
Wilt does show some postseason decline in WS/48, though he retains just as much value in Backpicks BPM. A few of the years he's competing against are behind in the regular season and creep ahead in the playoffs... but after reviewing the competition in your tweet, I wonder how well the postseason metrics we do have for Wilt capture the opponent difficulty. There's no way any of Kareem, Shaq, Duncan, Hakeem, Curry faced playoff defenses as good as Bill Russell and Nate Thurmond in 73% of your playoff games!

Wilt: Inconsistency year to year gives me a bit less confidence in him, in 1967 it seems like he put things together, and that 76ers team was for sure one of the more dominant teams of all time, as well as him putting one of the more dominant playoff runs ever, but how sustainable is his value? Was that team just a perfect fit and I should have less confidence when picking him to lead my team in a vacuum? It wasn’t too long after his 64 and 67 seasons where he just had a flat-out questionable impact from the WOWY data we had(1965 and 1969)

I'm not the biggest fan of Wilt as a scorer in 1967 and don't think he blended his scoring attack with creating tor teammates at the same time quite as well as those two so it made me hold back on his offense a bit, there were also the weird WOWY indicators across his prime(where some just look straight up unimpressive) where i'm not sure if it was just a best fit situation or not and if I should have the same confidence in him as my other picks who had consistent very high impact indcators throughout their entire prime in different situations/team constructions. I would still give Wilt all 3 of those advantages and personally I could've seen myself voting him as high as like 3 if I viewed him a bit more optimistically, hopefully we can get more discussion going on him in later threads.

Thanks for responding to my question! I see where you're coming from. Wilt changed his play style more than many of the all-time greats over his career. With this stylistic change, the source of his value also changed. Earlier on, his value of course came more from his scoring volume. Later on, it came more from his defense and passing.

This is what I meant before by Wilt "putting it all together." In theory, if you think Wilt could put all these sources of value together at once (his scoring, efficiency, defense, passing, rebounding, screening, etc.), he'd have a tremendous argument for the greatest peak in this tier. A lower interpretation would posit the he could only choose a few areas at once, and thus be lower in this tier. It sounds like we're in agreement on this!

The question then becomes: did 1967 Wilt put it all together at once (i.e. did he successfully balance and combine scoring value, the playmaking value, and the defensive value)? You say you're less convinced on his ability to blend scoring value with playmaking, which I think is understandable -- like you say, if you were higher on his scoring value at his peak, you'd probably bump him in your ranking. To me, I think he still maintained enough scoring value in 67 to be in the conversation, which probably explains why I have him higher. I think he clearly balanced defense and playmaking in 1967. The question is scoring. Per Backpicks' ScoreVal, 67 Wilt is higher than peak Hakeem in the regular season by a lot and in the postseason. He's just behind Shaq and a bit more behind Kareem, with the gap larger in the playoffs. But as I mentioned earlier, I wonder how much the lost value is underrating his opponents defense (he faced Bill Russell and Nate Thurmond 73% of the time).

Why did Wilt put it altogether in 1967, but not in other years (or at least not as effectively)? You mention his inability to balance Scoring, Playmaking, and Defense in other years lowers your confidence in 1967. To me, this comes down to coaching and mindset.
-In the early 1960s (peaking in 1962), there are tons of stories of young Wilt intentionally focusing on scoring, both because of his own preference (he though focusing on scoring would produce the most value) and due to coaching.
-In the mid 1960s (peaking in 1967), Wilt and his coach saying they intentionally wanted to balance scoring, playmaking, and defense. In 1968, Wilt said he intentionally wanted to focus on playmaking over scoring.
-In the early 1970s, Wilt said he further de-emphasized scoring with the approval of his coaches.
-The Argument that Wilt DIDN"T "put it all together": 1) If you're lower on Wilt, I think you could argue his intentional decision to hyper-focus on scoring in the early 60 and playmaking in 68 limited his overall value. Whereas Russell saw the benefit of a balanced offense, Wilt's inability to see the benefits lowers his in-game value. 2) While Wilt was known as a smart player, his IQ advantage tended to be more slow and methodical. You could argue his in-game quick decision making wasn't quite as high. For example (like Hakeem, Duncan, and Durant), Wilt could sometimes get into the mindset of "Okay, now I'm in scoring mode. Okay, now I'm in playmaking mode," without being able to fluidly switch between scoring and playmaking (and even leverage them against each other) using quick decision making, like the best offensive players do. I think there's some merit to both of these counters.
-The Argument that wilt DID "put it all together:" Still, it gives me pause that many of these decisions were approved and even encouraged by the coaching. It's much harder for Wilt to balance scoring and playmaking in 1962 if his coach is actively encouraging him to try to average 50 ppg. I wonder how much worse analytics and strategy in the 60s enabled this behavior. This is partially why I'm actually high on the time-machine argument for Wilt. What if, rather than having a coach encourage only scoring, Wilt was taken to a time where coaches valued more offensive balance (and had the perimeter spacing to better enable that balance)? Wilt was clearly very statistics driven... what if, rather than playing in a time where the best stat to maximize were points per game, he was taken to a time when the best stat to maximize was plus minus per game? Wilt might have had quite a few more years where he "put it all together."

[Proxy]

You mention he has late-clock bailouts that might lower his efficiency. I haven't heard that argument before! Do you have any metric or ballpark measurement of how often he did a late-clock bailout? Even a ballpark estimate would help us produce an adjusted TS% for him. I'm surprised they would go to him at the end of the clock... I would have thought they would go to their perimeter stars more, but perhaps that's the era difference showing up.

Right now I don't have any estimates unfortunately. It was really just something I feel i've seen occasionally when watching the film from that era we had over the years. Maybe when I have way more time I could go back and try tracking something like that if someone hasn't already. I think it's some weird era thing though yeah lol. I appreciate your response and maybe we can get more people to look at that sort of thing for other players as well. I also thing you brought up some great points about Wilt and I could also see how he is a plsyer that may benefit from the time machine argument.

[DraymondGold]

Kareem Discussion:

.

.
I think you both voted for Shaq first in this ballot. He definitely had a great peak! I also appreciate the reasons you gave, and I’ll try to give a more in depth response if I have time.

Do either of you have any counters for the statistical case I made on page 2? Looking across a variety of all-in-one metrics (which use actual plus minus data, plus minus data that adjusts for teammates, and box score estimatjons of plus minus data), Shaq looks like he’s below Curry, playoff Kareem, and arguably Duncan. If you’re less swayed by the data, 70sFan also did a film analysis in Peak Thread #2 I found convincing.
...

Really the way I saw the data was that rankings aside, the actual values seem so close to eachother between some of those players that you honestly could mostly still just go either way. I'm not too big on single season hybrid metrics like I said for KG as a be-all-end-all, especially in the playoffs due to the smaller sample size.
...
The Kareem Vs Shaq stat-based argument is missing alot of the measures I think are more important(really no +/- or APM data) so i'm not too big on doing a comp between the 2 using mostly box score metrics.
70sfan did provide some compelling points(especially on defense which is why I removed 2001 Shaq from being at the top of my voting) but I also think one of the main differences in how he and I voted was that I don't believe Kareem's offense reached quite the same heights because I believe his creation was less valuable with the lack of 3 point shooting at his peak in the 70s - which he doesn't agree with doing I think.

As I read your argument, what I'm hearing is: 1) we don't have every available metric for Kareem which limits are statistical analysis, 2) The statistics for Shaq vs Kareem are super close overall, and 3) even if the playoff metrics favor Kareem, playoff sample size is an issue for single seasons. Thus, given how close it is, we can turn to other sources of analysis (e.g. Kareem's playmaking vs Shaq's, etc.), where Shaq ends up winning in your estimation (possibly due to his era, but you don't think we should do the time machine argument as much for Kareem's playmaking). Let me know if this summary missed anything!

Overall, that's fairly convincing! I see where you're coming from. I'll push back slightly on a few points (just because I'm enjoying the discussion! ), but overall I follow the reasoning and think it's a perfectly reasonable conclusion:
1a) For point 1, I absolutely agree that there's less data for Kareem. We're missing AuPM (regular season and postseason), full season RAPM/PIPM in Kareem's prime, and ESPN's RPM. We do still have Squared2020's Historical RAPM for 1985 Kareem. It's for a half season sample, and Kareem just pops off the page, which is particularly impressive given how many years after his peak he was. Still, it's a smaller sample size and different context than 1977 Kareem. Thinking Basketball also has 3 year Goldstein Playoff PIPM in his final video of the greatest peaks series (the top 10 summary). Not sure how he got his hands on it, but in it, 1977–1979 Playoff Kareem > 2000 –2002 Shaq.
1b) I think I'm not as low on box-score composites like Backpicks Box Plus Minus as you are. I think some of my confidence in it comes from this study (https://fansided.com/2019/01/08/nylon-calculus-best-advanced-stat/), which tests the predictive power of various box score stats and advanced stats. Backpicks BPM and Basketball Reference's BPM may not be as good as PIPM, but they still do a fairly good job (and a heck of a lot better than other stats like PPG and PER).
2) Agreed, the overall statistics are fairly close for both.
3) While it's true that one postseason sample is small for true plus/minus stats, the Box-score based estimations are much more stable (that's actually one of the reasons for using them)! Plus, Thinking Basketball somehow got his hands on 3-year playoff PIPM for Kareem, which is a larger sample. Kareem outperforms Shaq in both the more stable box-score estimations and the larger-sample 3 year playoff PIPM.
For those reasons (along with 70sFan's film analysis, which you mentioned), I think I've been convinced to go Kareem over Shaq. But still, I see where you're coming from.

Kareem to me is the guy from the bunch I'd be most considering here. I would consider Kareem's offense to be better than Olajuwon's, so it's then a question of how strong of an edge I'd give to Dream on defense where I think Kareem's prime was also excellent on defense.

Re: Kareem's '74 & '80s rTS%. It has to be noted that this wasn't his efficiency edge every post-season. While this is a peak project where in theory it makes sense to ignore weaker years, I think with Kareem we definitely saw him being more vulnerable to defensive match-ups than Olajuwon seemed to be, and while Kareem has a significant raw TS% edge in his career in the regular season, that basically disappears when we look over to their playoff careers where - let's note - Hakeem has the edge on PER and BPM (while Kareem maintains a slight lead by WS/48).

It's true that playoff Kareem is better in 77 than 74 and 79-80. The counter that 70sFan offered was that he was at his peak in 77 (and the surrounding years), but we lack any playoff sample in 75 or 76 -- if he had a larger sample in those years, he might show more playoff consistency closer to the heights he reached in 77 (though its hard to know for sure).

I also wonder whether the decline in other years is as big as you say. While his postseason Backpicks BPM peak was in 77 (7th best all time), he did get fairly close in 74 (would be 10th all time). Both playoff runs are over Shaq, the 77 playoff run is better than Hakeem's best by some separation, and the 74 playoff run is only worse than Hakeem's by 0.1. It's also worth mentioning his regular season BPM in 74/78/79/80 are higher than 77... this could make you lower on 77, or it could make you higher on the consistency of Kareem from 74-80, depending on your interpretation. What do you think?

Also... I don't see what you mean by his TS% advantage declining in the playoffs against Hakeem. He has a relative TS% of +13.7% in 1977, 12.5% in 79. 9.9% in 80, and 9% in 74. If we're only lookin at the years around their peak (Hakeem admittedly had better rTS% when he was younger), peak Hakeem's best playoff value is in 94 at +4.9% rTS. It looks like Kareem has the significant rTS% advantage in the playoffs too, at least by my interpretation. Do you disagree?

[ardee]

1. 2000 Shaquille O'Neal

This is difficult for me, not having Wilt in the top 3. But I've been doing some re-evaluating and I don't think I can deny Shaq's impact was higher at their respective bests.

Just stupidly dominant, and on both ends. This was the one year Shaq really tried on both ends all year and the results were spectacular. He won the scoring title, was second in rebounds and third in blocks. Almost a unanimous MVP and 2nd in DPoY. 19, 16 and 11 game winning streaks.

His effort in the Playoffs was mammoth, he was +22.9 On/Off and clearly did everything he could to get that team to a title, against a veritable murderer's row of opponents (played 4 of the other top 7 SRS teams somehow), and in the most important series he was often triple teamed by Scottie Pippen, Rasheed Wallace and the 7'3 300 lb Sabonis.

I would not put 2001 Shaq on the same level as this year over the next two guys because while the Playoffs were as dominant as 2000, he did not give full effort in the RS (defense was definitely worse) and at this level that does matter somewhat.

2. 1994 Hakeem Olajuwon

Second best defensive center (and frankly player) of all time IMO behind Russell. He came the closest of anyone ever after Bill to combine elite vertical defense (which you have guys like Ewing and Howard specialize at) with elite horizontal defense (KG, Draymond). The other one is David Robinson, but Hakeem's much more resilient offensive game in the Playoffs creates the separation here.

The '95 Playoff run and his demolition of David Robinson is what gets talked about a lot, and rightly so, but I think what Hakeem did to Ewing on both ends in the '94 Finals is one of the most underrated big stage performances ever. He averaged 27/9/4 with 4 bpg on 56% TS while holding Ewing to 19/12/2 on 39% TS. That's -16 from his RS TS%. That may even be more impressive than the '95 matchup with Robinson who at least managed to score against 'Keem at a decent rate.

To be a guy who can go off for 30/15 on any night against elite opposition while simultaneously being a 5x5 threat is just insane, a monster on both ends who would dominate in any era.

HM: 1993 Hakeem

3. 2003 Tim Duncan

Really not much separating him and Hakeem and I honestly could be persuaded to swap them.

Duncan had a stupendously dominant year on both ends. A 25/15/5 with 3 bpg average throughout the Playoffs as well as being the unquestioned best player in a season where KG, Kobe, McGrady, Shaq and Dirk were all close to peak level is no joke at all.

Others have mentioned his passing and I think it's symbolic of the gargantuan load he was forced to carry to lead that team to a title. From the closeout game 6 against the Lakers in the second round through the end of the Finals, he averaged 26.1/16.8/5.6 with 4.2 bpg on 57% TS over 13 games. His Playoff On/Off was +23.1 and the team was -14 without him on the court. I don't think anyone has ever dragged a worse cast to the title.

HM: 1995 Hakeem, 2001 Shaq
HM: 2002 Duncan

[Proxy]

1a) For point 1, I absolutely agree that there's less data for Kareem. We're missing AuPM (regular season and postseason), full season RAPM/PIPM in Kareem's prime, and ESPN's RPM. We do still have Squared2020's Historical RAPM for 1985 Kareem. It's for a half season sample, and Kareem just pops off the page, which is particularly impressive given how many years after his peak he was. Still, it's a smaller sample size and different context than 1977 Kareem.  Thinking Basketball also has 3 year Goldstein Playoff PIPM in his final video of the greatest peaks series (the top 10 summary). Not sure how he got his hands on it, but in it, 1977–1979 Playoff Kareem > 2000 –2002 ?

This PIPM is just using on/off estimates calculated using box score information. Goldstein used to use these estimates but i'm not sure if he does anymore. It's basically just a bloated BPM, which yeah it could be useful especially in smaller sample sizes, and is definitely not insignificant at all(maybe even better than BPM 2.0 or Backpicks BPM still). But, +/- data is still EXTREMELY important, and in the playoffs it is still very important even if it becomes less reliable due to sample size issues. For someone like myself that believes Shaq has more hidden value that is not captured by traditional box score measurements like i've argued before, this is just not nearly enough for me to think it's a fair comparison and alot of those metrics are telling me relatively similar things over and over again. I think your conclusion is still very reasonable though.

This is the link to his old on/off estimates spreadsheet:
https://docs.google.com/spreadsheets/d/1Rshz7YDmMjMQcLCCEhz5Disuy3WYKFoVKmL4buh1k0w/edit?usp=drivesdk

He deleted the original tweet linking the estimates, but these results were not very reliable(a r^2 of 0.49 is just not good at all) - I think he updated the PIPM formula since then but I have no idea what he changed it to

Read on Twitter

?t=56X3QynISNoCwPD3d9aPtg&s=19

[OhayoKD]

1a) For point 1, I absolutely agree that there's less data for Kareem. We're missing AuPM (regular season and postseason), full season RAPM/PIPM in Kareem's prime, and ESPN's RPM. We do still have Squared2020's Historical RAPM for 1985 Kareem. It's for a half season sample, and Kareem just pops off the page, which is particularly impressive given how many years after his peak he was. Still, it's a smaller sample size and different context than 1977 Kareem.  Thinking Basketball also has 3 year Goldstein Playoff PIPM in his final video of the greatest peaks series (the top 10 summary). Not sure how he got his hands on it, but in it, 1977–1979 Playoff Kareem > 2000 –2002 ?

This PIPM is just using on/off estimates calculated using box score information. Goldstein used to use these estimates but i'm not sure if he does anymore. It's basically just a bloated BPM, which yeah it could be useful especially in smaller sample sizes, and is definitely not insignificant at all(maybe even better than BPM 2.0 or Backpicks BPM still). But, +/- data is still EXTREMELY important, and in the playoffs it is still very important even if it becomes less reliable due to sample size issues. For someone like myself that believes Shaq has more hidden value that is not captured by traditional box score measurements like i've argued before, this is just not nearly enough for me to think it's a fair comparison and alot of those metrics are telling me relatively similar things over and over again. I think your conclusion is still very reasonable though.

This is the link to his old on/off estimates spreadsheet:
https://docs.google.com/spreadsheets/d/1Rshz7YDmMjMQcLCCEhz5Disuy3WYKFoVKmL4buh1k0w/edit?usp=drivesdk

He deleted the original tweet linking the estimates, but these results were not very reliable(a r^2 of 0.49 is just not good at all) - I think he updated the PIPM formula since then but I have no idea what he changed it to

Read on Twitter

?t=56X3QynISNoCwPD3d9aPtg&s=19

+/- wouldn't capture shaq fouling out frontlines since his teammates would benefit when he's off the court from things.

[DraymondGold]

More Shaq Discussion: does Shaq's +/- Data miss any of his value?

1a) For point 1, I absolutely agree that there's less data for Kareem. We're missing AuPM (regular season and postseason), full season RAPM/PIPM in Kareem's prime, and ESPN's RPM. We do still have Squared2020's Historical RAPM for 1985 Kareem. It's for a half season sample, and Kareem just pops off the page, which is particularly impressive given how many years after his peak he was. Still, it's a smaller sample size and different context than 1977 Kareem.  Thinking Basketball also has 3 year Goldstein Playoff PIPM in his final video of the greatest peaks series (the top 10 summary). Not sure how he got his hands on it, but in it, 1977–1979 Playoff Kareem > 2000 –2002 ?

This PIPM is just using on/off estimates calculated using box score information. Goldstein used to use these estimates but i'm not sure if he does anymore. It's basically just a bloated BPM, which yeah it could be useful especially in smaller sample sizes, and is definitely not insignificant at all(maybe even better than BPM 2.0 or Backpicks BPM still). But, +/- data is still EXTREMELY important, and in the playoffs it is still very important even if it becomes less reliable due to sample size issues. For someone like myself that believes Shaq has more hidden value that is not captured by traditional box score measurements like i've argued before, this is just not nearly enough for me to think it's a fair comparison and alot of those metrics are telling me relatively similar things over and over again. I think your conclusion is still very reasonable though.

This is the link to his old on/off estimates spreadsheet:
https://docs.google.com/spreadsheets/d/1Rshz7YDmMjMQcLCCEhz5Disuy3WYKFoVKmL4buh1k0w/edit?usp=drivesdk

He deleted the original tweet linking the estimates, but these results were not very reliable(a r^2 of 0.49 is just not good at all) - I think he updated the PIPM formula since then but I have no idea what he changed it to

Read on Twitter

?t=56X3QynISNoCwPD3d9aPtg&s=19

+/- wouldn't capture shaq fouling out frontlines since his teammates would benefit when he's off the court from things.

Proxy, thanks for explaining the source for the old Goldstein +/- data! Agreed, +/- data is extremely important, particularly for players like Shaq who gain offensive value from their gravity and ability to distort the defense.

OhayoKD, Great point! I always enjoy discussions about stuff that +/- data misses. It's certainly possible Shaq gets opponents in foul trouble more than his Greatest Peak competitors, which would help his teammates when Shaq's off the court, thus making his +/- data slightly underestimate his value (by misattributing value to his teammates who performed better because they got to play against easier opponents since Shaq's defender fouled out, when the real value came from Shaq for fouling out his defender).

But... are we sure this would be a massive amount of value? I'm not sure it would be much at all. I can't find data for how much Shaq fouled out opponents, but we do have data for how much opposing teams fouled Shaq's team (https://www.teamrankings.com/nba/stat/opponent-personal-fouls-per-game?date=2022-06-17).

Shaq's 2004 Lakers are first in fouls per game by 0.1 (+3 fouls over league average); they drop by 2.25 fouls/game when Shaq leaves in 2005.
Shaq's 2005 Miami are third in fouls per game (+1.25 fouls over league average); they improved by 1.55 fouls/game when Shaq joined in 2005.
So: Shaq's teams are toward the top of the league in fouls/game, but they are certainly not outliers. Comparing the Lakers/Miami when Shaq was on the team vs when he moved, Shaq brings about +1.9 fouls/game to his team, which is a ton of fouls for a single player to bring, but not so much that Shaq's teams become massive outliers in their total fouls drawn.

And remember: for the on/off data to be wrong, it would have to be misattributing Shaq's off value... which only apply in the minutes when Shaq's defender would have played but couldn't because they fouled out (likely small number of minutes) AND when Shaq was also not on the court (otherwise the value would get properly attributed to Shaq). Could this make a small difference between +/- data and true value? Absolutely! But is there evidence that Shaq's true value is that much more than his +/- data suggests? Not at all, at least as far as I can tell.

Since we're on the topic of value that plus/minus data misses, Duncan and Curry are also candidates to gain some value. For example, players who are good leaders and help set a positive culture / locker room would also have value that doesn't show up in plus minus data, since the good culture would still help their teammates while the superstar was on the bench. In theory, this would (slightly) boost Duncan and Curry's true value, just to name some famous locker room leaders in the same tier as peak Shaq. Plus/minus data would also miss value that superstars would have racked up against worse teams in blowouts, if they weren't sitting on the bench for some/all of the 4th quarter. Curry is the most famous example of this, though there may be others.

Steph Curry Discussion:

.

.
I think you both voted for Shaq first in this ballot. He definitely had a great peak! I also appreciate the reasons you gave, and I’ll try to give a more in depth response if I have time.

Do either of you have any counters for the statistical case I made on page 2? Looking across a variety of all-in-one metrics (which use actual plus minus data, plus minus data that adjusts for teammates, and box score estimatjons of plus minus data), Shaq looks like he’s below Curry, playoff Kareem, and arguably Duncan. If you’re less swayed by the data, 70sFan also did a film analysis in Peak Thread #2 I found convincing.

On another note, there’s already been some discussion of Hakeem, but I haven’t seen as much back and forth discussion about Wilt vs Kareem or Shaq. I think 67 Wilt clearly is the better defender, rebounder, and passer of the three. I imagine it’s the scoring gap in 67 which pulls Wilt down for people? Are there any other reasons (e.g are people less high on Wilt’s playoff performance)?

Really the way I saw the data was that rankings aside, the actual values seem so close to eachother between some of those players that you honestly could mostly still just go either way. I'm not too big on single season hybrid metrics like I said for KG as a be-all-end-all, especially in the playoffs due to the smaller sample size. But, while there was a chance Steph was still just better in his run compared to 2000 Shaq I would say he was probably in a more optimal situation, and think Shaq's game just had a longer sample size proving it's resilience both at an individual and team level at his peak without significant changes in his skillset, whether it was from Steph's durability or other factors - teams really did try so hard to wear him down and I do see slightly more dueability issues, so I just defaulted to him when making my list. I appreciate it though and think you brought up some interesting points.
...

Good points! It sounds like ike there were a few factors for why you have Shaq > Curry:
1) You're less a fan of hybrid metrics or plus minus data for a single postseason. This adds more uncertainty to the fact that 2017 Curry beats 2000 Shaq in 5/5 of the playoff-specific plus/minus stats I listed.
2) Curry's team fit was better.
3) Curry's less resilient than Shaq.
As always, let me know if I misunderstood or missed anything!

For 1), I mentioned some of my counters to this in my Kareem Discussion above, so I won't rehash it too much here -- the main points were that box plus/minus stats are stabler in small sample sizes like the playoffs, and the PIPM stat I listed was for 3 playoffs which should help with the sample size issue. Personally, I find the fact that Curry beats Shaq in all 5/5 playoff stats pretty compelling, though they're definitely close!
2) I tend to agree here. The Warriors were definitely great scenario for Curry, with great coaching, great defenders, and great offensive players alongside Curry. Still, Shaq also had some great coaching, a great running mate in young Kobe, and great supporting cast (especially in the 2001 playoffs, Shaq's supporting cast had a clear 3 point shooting advantage vs league average).

Personally, I find Curry's scalability/portability enough to make up for this. Historically, it's easier for players to have their best plus minus value with worse supporting casts (or at least with no co-stars competing for value). The better the supporting cast and the better the co-stars, the more the team improves but the harder it is for a single individual to dominate the team's value (since they're competing for value against better teammates). And yet... Curry dominates the Warriors in terms of value.

From 2017–2019 (larger sample to give more stable values), here's the net rating with each of the stars on or off:
-All 4 stars on: +17. (that's 20% better than the 1996 Chicago Bulls across 3 seasons!)
-Only Klay off: +15.64.
-Only KD off: +13.54 (still better than the 96 Chicago Bulls even with KD off)
-Only Draymond off: +12.77
-Only Steph on, all 3 other stars off: +10.81
-Only Steph off: +1.94
When all four stars are on the court, the 17-19 Warriors are significantly better than the 1996 Bulls. With all 3 other all stars off, and just Steph on, the 17-19 Warriors have a better net rating than the 16 Warriors, 13 Heat, 2000 Lakers, 91 Bulls, 87 Lakers, or 86 Celtics. With all 3 all stars on, and just Steph off, the 17-19 Warriors are worse than this season's 2022 Cavs.

If we include both the regular season and the playoffs, the difference decreases, but Curry still dominates (only KD off: +11.08. only Steph off: +3.66). Lots of people have said that Curry's had a better fit than other peaks, and that his team was stacked. This is true. But, as far as I can tell, they only dominated when Curry was on the court, and they completely fell apart (by their standards) without him. Source: https://www.pbpstats.com/wowy-combos/nba?TeamId=1610612744&Season=2018-19,2017-18,2016-17&SeasonType=Regular%2BSeason&PlayerIds=201142,201939,202691,203110

3) I agree here too, Curry is less resilient than Shaq. But I wonder how much of that decline comes from Curry's worse postseason health during his prime? Per Thinking Basketball's video, Curry on average declined by BPM from the regular season to the postseason (from 2013-2018). But if we only take healthy playoffs (2013-2018 except 16 and 18): Curry improves by +0.2 BPM. By AuPM, Curry improves by +0.57% from the regular season to the playoffs, even when including injured postseasons (from 2013-2021). With another healthy postseason in 2022, Curry improves by even more in both metrics.
Now it's true that Shaq improves by more according to AuPM: +0.67% (from 1994-2006, 2008, 2010). But personally, I don't find +0.57% vs +0.67% a big enough difference to sway me in favor of Shaq.

Like you hint at, the biggest cause of Curry's playoff decline is health concerns. And you're definitely right that prime Steph had greater health issues than prime Shaq. If you take different approach on health that I do (e.g. docking players in every season a little bit because they're injury risks), I could see that swaying you against 2017 Curry. And that would be a valid approach! From my point of view though, Curry was healthy in 2017, and he outperformed 2000 and 2001 Shaq in all 5/5 of my playoff-only plus minus metrics. So... he was the more valuable playoff performer then (at least to me ).

Could someone lay out the argument for 2017 as Curry's peak vs 2016? I just don't see him being better that year unless it's got to do with health.

I'm aware that he peaked in many APM stats in 2017 but personally I try not to rely on +/- stats so much because they're so heavily influenced by team context. Over large, multi-year sample sizes that encompass different team situations they can absolutely give you an idea of a player's intangible impact, but over a single season there's still so much room for variance.

I like to evaluate players from a skillset POV and watching Steph that year he didn't feel like a better player than he was in 2016 at all. He was maybe slightly better as a defender but as a point guard that isn't really gonna make a huge dent and his shot making was a far cry from what it was in 2016. In real time I actually thought it was a pretty weak RS from him until Golden State's 15 game run without KD at the end of the year.

As for his postseason I view it as an outlier for him in the same way I don't think 09 LeBron had a switch where he could become a 45% from midrange, 37 PER player at will. For that same reason I'm also probably more forgiving of Steph's 2016 postseason than most. Even as unbelievable as he is, as a shooter, Steph is more prone to variance than other all time greats, which is why it's hard to take any 10-15 game sample as his "true" level of play, if that makes sense. His 2018, 2019, and 2022 runs are probably closer to what you can expect from him on average come playoff time.

Then you can get into how favorable his circumstances were. The talent gap between Golden State and everyone else was just comical, and the three teams he faced in the west all had their best defenders (Nurkic, Gobert, Kawhi) either missing or hobbled. In the finals the Cavaliers were the 27th ranked defense and JR and Kyrie, two notoriously boneheaded defenders, made miscommunication after miscommunication which torpedoed the Cavs. Not to say that it was completely unrelated to Steph's gravity, but I recently rewatched some games from that series they were making the same dumb mistakes over and over again defending Iguodala and KD in transition with Steph not even on the court. They defended him well the year before so more than anything it felt like they were overwhelmed by Golden State's firepower as a whole.

Was there a change to his approach on offense that I might've missed? Was his movement off the ball especially crisp that year or something? Even so, I don't think player's intangibles typically vary season to season as much as people pretend they do to justify what the +/- stats are telling them. People love to say the box score doesn't matter but that simply isn't true when it comes to comparing players to themselves playing the same role just a season ago.

FTR I've got no problem with Curry's peak being ranked highly (I wouldn't go as high as top 5 though) it's just that I view 2015, 2017, and 2018 all similarly and a clear tier down from 2016.

So, big thing:

I don't think Curry fits that well in a season-based peak conversation because different years have different arguments for and against him.

'14-15 wins the MVP and the championship.
'15-16 is his best regular season, and his worst (or near worst) post-season.
'16-17 is the smoothest season - MVP of the greatest team in the history of basketball - and his best statistical playoffs, but the degree of difficulty can be said to be all-time low.
'21-22 proved his capacity for latent impact like never before and seemed the most resilient in the playoffs ever, but had regular season cold streaks causing him to have the lowest TS Add (shooting volume * relative league efficiency) of his entire prime.

Which year to pick? I honestly don't feel super strongly about it other than I really don't think '14-15 should be in the conversation.

Great points letskissbro and Doctor MJ! To answer your question in short, here are the changes I see from 2016 to 2017:
1. Postseason health: you mention this of course. If you take a more probabilistic approach to injury (e.g. partially docking every season based on perceived chance of injury, rather than fully docking injured seasons and not docking healthy seasons), I could definitely see you taking 2016 > 2017. I think it's hard to estimate injury probabilities though, and since this is a 1 year peak project, I just go with the simple route: 2017 was healthy, so he doesn't get docked! If we're just looking at skill improvements, I think 2017 was slightly better at pacing himself through the regular season (a la 2013 LeBron vs 2009, 1977 Kareem vs 1972/71), which helped him improve his chances of being fresh and healthy in the postseason.
2. Improved resilience: increased weight and strength allows for better defense and better ability to take contact on and off ball. Slightly improved decision making allows for slightly fewer boneheaded turnovers and more consistent (but less flashy) passes / shot selection (Athletic Alchemy mentioned he noticed this in his film study at some point ~2017). Slightly tighter handle (which would continue to improve and show noticeable results by 2022).
3. Improved scalability: I have Curry as the GOAT portable/scalable player, and he's certainly still all-time scalable in 2016. However, the actual act of adjusting to Durant forced Curry to improve even more without the ball and be even more efficient with his on ball usage.

I think I also have fewer concerns with the 2017 regular season: I see 2017 regular season Curry as just as good as 2016 regular season Curry, though definitely less valuable. By good, his actual goodness as a player (irrespective of team context); value is more context-dependent.

So why did his performance and value decline? I think fitting next to Durant (and making the Scalability improvements I mentioned in #3 above) actually took some work in the first quarter/half of the season. This took him out of rhythm and slightly decreased his impact metrics. In interviews, Curry actually says just that! He says he was pulling back too much in order to make room for Durant in the early part of the season, but by the end, he had figured out how to go full force without taking away from Durant's offense.

Statistically, Curry's performance improved in the second half of the season with Durant, when Durant was out for those last few weeks, and when Durant was back from the playoffs. By my eye (and from what I can tell by the metrics), Curry was just as good as he was in 2016 regular season during that span, which gives me confidence that he was just as "good" of a player in the 2017 regular season (even if figuring out how to fit alongside Durant in the first half of his season lowered his regular season impact metric value).

Like Doctor MJ says, Curry continued to make improvements in other areas over the next few years. By 2022, he had noticeably better defense, decision making, passing, handle, and resilience. I still take 2017 (and healthy 2016) because of his athleticism (specifically his speed/quickness) and his overall impact metrics. Like other people have said though, if the 2022 regular season ended up just being a cold spell (not an actual decline in shooting due to age/motor/athleticism/increased effort elsewhere), then 2021/2022 might also have an argument at his peak. Let me know if you disagree or if you have any questions!

[falcolombardi]

This PIPM is just using on/off estimates calculated using box score information. Goldstein used to use these estimates but i'm not sure if he does anymore. It's basically just a bloated BPM, which yeah it could be useful especially in smaller sample sizes, and is definitely not insignificant at all(maybe even better than BPM 2.0 or Backpicks BPM still). But, +/- data is still EXTREMELY important, and in the playoffs it is still very important even if it becomes less reliable due to sample size issues. For someone like myself that believes Shaq has more hidden value that is not captured by traditional box score measurements like i've argued before, this is just not nearly enough for me to think it's a fair comparison and alot of those metrics are telling me relatively similar things over and over again. I think your conclusion is still very reasonable though.

This is the link to his old on/off estimates spreadsheet:
https://docs.google.com/spreadsheets/d/1Rshz7YDmMjMQcLCCEhz5Disuy3WYKFoVKmL4buh1k0w/edit?usp=drivesdk

He deleted the original tweet linking the estimates, but these results were not very reliable(a r^2 of 0.49 is just not good at all) - I think he updated the PIPM formula since then but I have no idea what he changed it to

Read on Twitter

?t=56X3QynISNoCwPD3d9aPtg&s=19

+/- wouldn't capture shaq fouling out frontlines since his teammates would benefit when he's off the court from things.

Proxy, thanks for explaining the source for the old Goldstein +/- data! Agreed, +/- data is extremely important, particularly for players like Shaq who gain offensive value from their gravity and ability to distort the defense.

OhayoKD, Great point! I always enjoy discussions about stuff that +/- data misses. It's certainly possible Shaq gets opponents in foul trouble more than his Greatest Peak competitors, which would help his teammates when Shaq's off the court, thus making his +/- data slightly underestimate his value (by misattributing value to his teammates who performed better because they got to play against easier opponents since Shaq's defender fouled out, when the real value came from Shaq for fouling out his defender).

But... are we sure this would be a massive amount of value? I'm not sure it would be much at all. I can't find data for how much Shaq fouled out opponents, but we do have data for how much opposing teams fouled Shaq's team (https://www.teamrankings.com/nba/stat/opponent-personal-fouls-per-game?date=2022-06-17).

Shaq's 2004 Lakers are first in fouls per game by 0.1 (+3 fouls over league average); they drop by 2.25 fouls/game when Shaq leaves in 2005.
Shaq's 2005 Miami are third in fouls per game (+1.25 fouls over league average); they improved by 1.55 fouls/game when Shaq joined in 2005.
So: Shaq's teams are toward the top of the league in fouls/game, but they are certainly not outliers. Comparing the Lakers/Miami when Shaq was on the team vs when he moved, Shaq brings about +1.9 fouls/game to his team, which is a ton of fouls for a single player to bring, but not so much that Shaq's teams become massive outliers in their total fouls drawn.

And remember: for the on/off data to be wrong, it would have to be misattributing Shaq's off value... which only apply in the minutes when Shaq's defender would have played but couldn't because they fouled out (likely small number of minutes) AND when Shaq was also not on the court (otherwise the value would get properly attributed to Shaq). Could this make a small difference between +/- data and true value? Absolutely! But is there evidence that Shaq's true value is that much more than his +/- data suggests? Not at all, at least as far as I can tell.

Since we're on the topic of value that plus/minus data misses, Duncan and Curry are also candidates to gain some value. For example, players who are good leaders and help set a positive culture / locker room would also have value that doesn't show up in plus minus data, since the good culture would still help their teammates while the superstar was on the bench. In theory, this would (slightly) boost Duncan and Curry's true value, just to name some famous locker room leaders in the same tier as peak Shaq. Plus/minus data would also miss value that superstars would have racked up against worse teams in blowouts, if they weren't sitting on the bench for some/all of the 4th quarter. Curry is the most famous example of this, though there may be others.

I dont think is as much about fouling out as it is about getting rival teams quickly in the bonus where defense gets harder to do

[OhayoKD]

+/- wouldn't capture shaq fouling out frontlines since his teammates would benefit when he's off the court from things.

Proxy, thanks for explaining the source for the old Goldstein +/- data! Agreed, +/- data is extremely important, particularly for players like Shaq who gain offensive value from their gravity and ability to distort the defense.

OhayoKD, Great point! I always enjoy discussions about stuff that +/- data misses. It's certainly possible Shaq gets opponents in foul trouble more than his Greatest Peak competitors, which would help his teammates when Shaq's off the court, thus making his +/- data slightly underestimate his value (by misattributing value to his teammates who performed better because they got to play against easier opponents since Shaq's defender fouled out, when the real value came from Shaq for fouling out his defender).

But... are we sure this would be a massive amount of value? I'm not sure it would be much at all. I can't find data for how much Shaq fouled out opponents, but we do have data for how much opposing teams fouled Shaq's team (https://www.teamrankings.com/nba/stat/opponent-personal-fouls-per-game?date=2022-06-17).

Shaq's 2004 Lakers are first in fouls per game by 0.1 (+3 fouls over league average); they drop by 2.25 fouls/game when Shaq leaves in 2005.
Shaq's 2005 Miami are third in fouls per game (+1.25 fouls over league average); they improved by 1.55 fouls/game when Shaq joined in 2005.
So: Shaq's teams are toward the top of the league in fouls/game, but they are certainly not outliers. Comparing the Lakers/Miami when Shaq was on the team vs when he moved, Shaq brings about +1.9 fouls/game to his team, which is a ton of fouls for a single player to bring, but not so much that Shaq's teams become massive outliers in their total fouls drawn.

And remember: for the on/off data to be wrong, it would have to be misattributing Shaq's off value... which only apply in the minutes when Shaq's defender would have played but couldn't because they fouled out (likely small number of minutes) AND when Shaq was also not on the court (otherwise the value would get properly attributed to Shaq). Could this make a small difference between +/- data and true value? Absolutely! But is there evidence that Shaq's true value is that much more than his +/- data suggests? Not at all, at least as far as I can tell.

Since we're on the topic of value that plus/minus data misses, Duncan and Curry are also candidates to gain some value. For example, players who are good leaders and help set a positive culture / locker room would also have value that doesn't show up in plus minus data, since the good culture would still help their teammates while the superstar was on the bench. In theory, this would (slightly) boost Duncan and Curry's true value, just to name some famous locker room leaders in the same tier as peak Shaq. Plus/minus data would also miss value that superstars would have racked up against worse teams in blowouts, if they weren't sitting on the bench for some/all of the 4th quarter. Curry is the most famous example of this, though there may be others.

I dont think is as much about fouling out as it is about getting rival teams quickly in the bonus where defense gets harder to do

You'd still expect higher foul rates though for the team right?

[falcolombardi]

Proxy, thanks for explaining the source for the old Goldstein +/- data! Agreed, +/- data is extremely important, particularly for players like Shaq who gain offensive value from their gravity and ability to distort the defense.

OhayoKD, Great point! I always enjoy discussions about stuff that +/- data misses. It's certainly possible Shaq gets opponents in foul trouble more than his Greatest Peak competitors, which would help his teammates when Shaq's off the court, thus making his +/- data slightly underestimate his value (by misattributing value to his teammates who performed better because they got to play against easier opponents since Shaq's defender fouled out, when the real value came from Shaq for fouling out his defender).

But... are we sure this would be a massive amount of value? I'm not sure it would be much at all. I can't find data for how much Shaq fouled out opponents, but we do have data for how much opposing teams fouled Shaq's team (https://www.teamrankings.com/nba/stat/opponent-personal-fouls-per-game?date=2022-06-17).

Shaq's 2004 Lakers are first in fouls per game by 0.1 (+3 fouls over league average); they drop by 2.25 fouls/game when Shaq leaves in 2005.
Shaq's 2005 Miami are third in fouls per game (+1.25 fouls over league average); they improved by 1.55 fouls/game when Shaq joined in 2005.
So: Shaq's teams are toward the top of the league in fouls/game, but they are certainly not outliers. Comparing the Lakers/Miami when Shaq was on the team vs when he moved, Shaq brings about +1.9 fouls/game to his team, which is a ton of fouls for a single player to bring, but not so much that Shaq's teams become massive outliers in their total fouls drawn.

And remember: for the on/off data to be wrong, it would have to be misattributing Shaq's off value... which only apply in the minutes when Shaq's defender would have played but couldn't because they fouled out (likely small number of minutes) AND when Shaq was also not on the court (otherwise the value would get properly attributed to Shaq). Could this make a small difference between +/- data and true value? Absolutely! But is there evidence that Shaq's true value is that much more than his +/- data suggests? Not at all, at least as far as I can tell.

Since we're on the topic of value that plus/minus data misses, Duncan and Curry are also candidates to gain some value. For example, players who are good leaders and help set a positive culture / locker room would also have value that doesn't show up in plus minus data, since the good culture would still help their teammates while the superstar was on the bench. In theory, this would (slightly) boost Duncan and Curry's true value, just to name some famous locker room leaders in the same tier as peak Shaq. Plus/minus data would also miss value that superstars would have racked up against worse teams in blowouts, if they weren't sitting on the bench for some/all of the 4th quarter. Curry is the most famous example of this, though there may be others.

I dont think is as much about fouling out as it is about getting rival teams quickly in the bonus where defense gets harder to do

You'd still expect higher foul rates though for the team right?

Not necesarrily, teams in the bonus defend more lightly to avoid giving free throws so the end result is still bad for the defense even if they dont pick up more fouls than usual

[Lou Fan]

1. 2000 Shaquille O'Neal (b - 2001, c - 2002)
2. 1967 Wilt Chamberlain (b - 1964, c - 1968)
3. 1977 Kareem Abdul-Jabbar (b - 1974, c - 1980)

Since it's hard to pick from 3 near perfect center seasons Shaq gets the edge for doing in more recently. I don't see the difference between 67 and 77 to be that relevant since you can argue the competition in the 60s is better.

I would consider Curry but I think his best regular seasons he has some issues in the playoffs, whereas his recent strong finals performances came with more ok regular seasons for him.

How do you feel about 19 Curry? I would say he was the second best player in the regular season (behind Giannis) and at worst second best in the playoffs (possibly behind Kawhi). Same question @DoctorMJ? I saw you left it off Curry's best seasons. Why? I am firmly in the 17 Curry is his most complete season camp but 15/19/22 all seem like reasonable choices to me. 21 Curry too but I understand why that can't be in consideration given he didn't play in the playoffs.

[DraymondGold]

I dont think is as much about fouling out as it is about getting rival teams quickly in the bonus where defense gets harder to do

You'd still expect higher foul rates though for the team right?

Not necesarrily, teams in the bonus defend more lightly to avoid giving free throws so the end result is still bad for the defense even if they dont pick up more fouls than usual

That's true falcolombardi! I didn't think of that There might be a slight increase in offensive effectiveness due to defenses being in the bonus when Shaq draws fouls. And this might not

But... Shaq's teams aren't drawing fouls significantly more than other teams (they aren't even 1st in fouls drawn per game if we take a two year average in 04/05; see above). And for plus minus data not to capture this foul-drawing advantage, his teammates would have to performing better while opponents were in the bonus due to Shaq AND while Shaq was not on the court.... to me, this seems like this would be a pretty small sample of minutes and thus would be a pretty small adjustment from his measured plus minus value. Do either of you disagree?

1. 2000 Shaquille O'Neal (b - 2001, c - 2002)
2. 1967 Wilt Chamberlain (b - 1964, c - 1968)
3. 1977 Kareem Abdul-Jabbar (b - 1974, c - 1980)

Since it's hard to pick from 3 near perfect center seasons Shaq gets the edge for doing in more recently. I don't see the difference between 67 and 77 to be that relevant since you can argue the competition in the 60s is better.

I would consider Curry but I think his best regular seasons he has some issues in the playoffs, whereas his recent strong finals performances came with more ok regular seasons for him.

How do you feel about 19 Curry? I would say he was the second best player in the regular season (behind Giannis) and at worst second best in the playoffs (possibly behind Kawhi). Same question @DoctorMJ? I saw you left it off Curry's best seasons. Why? I am firmly in the 17 Curry is his most complete season camp but 15/19/22 all seem like reasonable choices to me. 21 Curry too but I understand why that can't be in consideration given he didn't play in the playoffs.

Hi Lou! I edited my previous response above to also address this. As I see it, 2017's impact metrics and athleticism advantage are enough to take 2017 over the later years (at least for now), though there are improvements in defense/passing/decision making/resilience after 2017.

If his 2022 regular season shooting ends up being just noise (not a true decline in value), I could see arguments for 2021/2022 then (probably over 2019). I think it would end up being like arguing 2016/17/18 LeBron (which parallels 2021/22 Steph) over 2013/2012 LeBron (which parallels 2017 Steph) over 2009 LeBron (which... sort of parallels 2016 Steph, though I think this is a stretch lol ). In short, the argument would go: I know there's a decline in impact metrics as he got older, but I think he was more valuable "when it mattered" and the regular season decline isn't big enough for me to care otherwise. I don't think it'll be consensus for sure, but perhaps in a few years, it'll be a possible argument!

[falcolombardi]

You'd still expect higher foul rates though for the team right?

Not necesarrily, teams in the bonus defend more lightly to avoid giving free throws so the end result is still bad for the defense even if they dont pick up more fouls than usual

That's true falcolombardi! I didn't think of that There might be a slight increase in offensive effectiveness due to defenses being in the bonus when Shaq draws fouls. And this might not

But... Shaq's teams aren't drawing fouls significantly more than other teams (they aren't even 1st in fouls drawn per game if we take a two year average in 04/05; see above). And for plus minus data not to capture this foul-drawing advantage, his teammates would have to performing better while opponents were in the bonus due to Shaq AND while Shaq was not on the court.... to me, this seems like this would be a pretty small sample of minutes and thus would be a pretty small adjustment from his measured plus minus value. Do either of you disagree?

1. 2000 Shaquille O'Neal (b - 2001, c - 2002)
2. 1967 Wilt Chamberlain (b - 1964, c - 1968)
3. 1977 Kareem Abdul-Jabbar (b - 1974, c - 1980)

Since it's hard to pick from 3 near perfect center seasons Shaq gets the edge for doing in more recently. I don't see the difference between 67 and 77 to be that relevant since you can argue the competition in the 60s is better.

I would consider Curry but I think his best regular seasons he has some issues in the playoffs, whereas his recent strong finals performances came with more ok regular seasons for him.

How do you feel about 19 Curry? I would say he was the second best player in the regular season (behind Giannis) and at worst second best in the playoffs (possibly behind Kawhi). Same question @DoctorMJ? I saw you left it off Curry's best seasons. Why? I am firmly in the 17 Curry is his most complete season camp but 15/19/22 all seem like reasonable choices to me. 21 Curry too but I understand why that can't be in consideration given he didn't play in the playoffs.

Hi Lou! I edited my previous response above to also address this. As I see it, 2017's impact metrics and athleticism advantage are enough to take 2017 over the later years (at least for now), though there are improvements in defense/passing/decision making/resilience after 2017.

If his 2022 regular season shooting ends up being just noise (not a true decline in value), I could see arguments for 2021/2022 then (probably over 2019). I think it would end up being like arguing 2016/17/18 LeBron (which parallels 2021/22 Steph) over 2013/2012 LeBron (which parallels 2017 Steph) over 2009 LeBron (which... sort of parallels 2016 Steph, though I think this is a stretch lol ). In short, the argument would go: I know there's a decline in impact metrics as he got older, but I think he was more valuable "when it mattered" and the regular season decline isn't big enough for me to care otherwise. I don't think it'll be consensus for sure, but perhaps in a few years, it'll be a possible argument!

I think the effect is probably marginal to be honest

Just wanted to point out forcing a lot of fouls has more effect than just fouling out players or the free throws themselves

Your curry-lebron comparision is interesting, 2009 lebron = 2016 curry is not a bad comparision for regular season (less so for playoffs)

2010 lebron may work out better tho (was not actually much worse than 2009 in reg season, had messy playoffs)

[Dr Positivity]

1. 2000 Shaquille O'Neal (b - 2001, c - 2002)
2. 1967 Wilt Chamberlain (b - 1964, c - 1968)
3. 1977 Kareem Abdul-Jabbar (b - 1974, c - 1980)

Since it's hard to pick from 3 near perfect center seasons Shaq gets the edge for doing in more recently. I don't see the difference between 67 and 77 to be that relevant since you can argue the competition in the 60s is better.

I would consider Curry but I think his best regular seasons he has some issues in the playoffs, whereas his recent strong finals performances came with more ok regular seasons for him.

How do you feel about 19 Curry? I would say he was the second best player in the regular season (behind Giannis) and at worst second best in the playoffs (possibly behind Kawhi). Same question @DoctorMJ? I saw you left it off Curry's best seasons. Why? I am firmly in the 17 Curry is his most complete season camp but 15/19/22 all seem like reasonable choices to me. 21 Curry too but I understand why that can't be in consideration given he didn't play in the playoffs.

It's an option but I still feel like Curry's most valuable seasons are the ones without Durant

Since he's unlikely to get in for a few threads I'll weigh it over though

[DraymondGold]

1. 1967 Wilt Chamberlain
2. 2017 Steph Curry
(2b 2016 Steph Curry)
3. 1977 Kareem Abdul-Jabbar
(3b. 1974 Kareem. )

1. Reasoning for Wilt. Here are my metrics for Wilt vs Curry:

Spoiler:

5. Curry vs Wilt: Plus-minus based stats:
Ai. AuPM (no data available for Wilt. 2016 Curry 2nd all time).
Aii. Postseason AuPM (no data available for Wilt. 2017 Curry 2nd all time).
Bi. Goldstein RAPM / Historical Square2020 RAPM: (miniscule sample available for Wilt, far below Curry’s 7th all time).
Bii. Goldstein Playoff PIPM (3 years for sample size): (no data available for Wilt. Curry’s 8th all time)
Additional plus minus stats: C. on/off: (no data available for Hakeem. Curry’s 1st all time)
Additional plus minus stats: D. WOWY: 65-70 Wilt >= 2016-2017 Curry (not sure about full prime WOWY. I brought in 16 Curry because Ben hasn't finished publishing Curry's mid/post-2017 WOWY numbers yet).
Additional plus minus stats: E. ESPN’s RPM: (no data for Wilt. 16 Curry 2nd all time)
Additional plus minus stats: F. Backpicks’ CORP evaluation: 1967 Wilt > 2017 Curry (healthy 2016 Steph Curry 4th all time > Wilt.)

Box score-based data
Gi. Backpicks BPM: 2017 Curry > 1967 Wilt (and healthy 2016 Curry is 2nd all time)
Gii. Postseason Backpicks BPM: 2017 Curry (4th all time) > 1967 Wilt
Hi. BR’s BPM: (no data for Wilt. healthy 2016 Curry’s 4th all time)
Hii. BR’s Postseason BPM: (no Data for Wilt).
Additional box score stats: Ii. WS/48: 1967 Wilt > 2017 Curry (but healthy 2016 Curry 3rd all time > Wilt)
Additional box score stats: Iii. Postseason WS/48: 2017 Curry > 1967 Wilt

We barely have any metrics to compare vs Wilt. Of the limited sample we do have, Wilt and Curry are tied 3 to 3. Curry leads in the postseason-only stats. If we add a neighboring year (2016 for Curry and either 1966 or 1968 for Wilt), Curry comes out on top in 5/6 stats. Still, we’re missing the majority of my preferred stats, so there's more uncertainty here. There's also enough contextual factors in Wilt's favor that sway me. Here's my 7 contextual factors:
1. Scalability (here Curry > Wilt). 2. Resilience (both were seen by media as playoff strugglers. Wilt suffered from GOAT-level defensive opponents, while Curry suffered from health. With favorable context, I see them equally resilient). 3. Health risks (which favor Wilt > Curry). 4. Defense, which plus/minus data can sometimes struggle measuring (obviously Wilt > Curry). 5. Fit (both had favorable fit in their peak year). 6. Time machine argument (the 60s certainly favor Wilt over Curry. In future eras, Wilt gains value with data / coaching / strategy / medicine, while losing value with the 3 point line and increased perimeter focus).

I want to focus on point 2: Resilience. ty 4191's comment made me realize just how much harder Wilt's opposing playoff defense was vs other all time great peaks.1967 Wilt faced Bill Russell and Nate Thurmond in 11/15 playoff games.... By my account, that means 1967 Wilt played 73% of his playoff games against the GOAT defender or the GOAT Big Man man-defender. Did Wilt decline in the playoffs? Yes by some metrics (ws/48), no by other metrics (Backpicks BPM). But this level of defensive opposition is absolutely ludicrous! The metrics where Wilt declined certainly do not account for opponent difficulty. Despite this, Wilt outperforms some competitors (Hakeem) in the metrics we have. While Curry, Shaq, and Kareem arguably have better metrics, they certainly did not face this level of defense. I'm willing to curve Wilt's performance here.

Counter to Wilt: Wilt wasn't able to "put it all together" (i.e. balance and combine his scoring value, playmaking value, or defensive value) all at once. 1967 might be his best balance of the three, but the fact that he doesn't replicate this balance in other years should limit our interpretation of 1967.
I think this has some merits. If I were to dock Wilt: 1) I would say his intentional decision to focus on one of those three areas (e.g. scoring over passing in 1962, passing over scoring in 1968) did slightly limit his overall career impact. 2) I would say he was worse at quick decision making, and would sometimes get stuck in "scoring mode" or "playmaking mode." But! I think some of this comes from his era. The prevailing belief at the time actively encouraged him to focus on scoring over playmaking. Coaches even encouraged him to focus on this unbalanced approach. In a modern era, with better metrics and coaching, I think Wilt would improve in this area. Longer discussion of this point here:

Spoiler:

I haven’t seen as much back and forth discussion about Wilt vs Kareem or Shaq. I think 67 Wilt clearly is the better defender, rebounder, and passer of the three. I imagine it’s the scoring gap in 67 which pulls Wilt down for people? Are there any other reasons (e.g are people less high on Wilt’s playoff performance)?

Wilt is criminally underrated here at Real GM. That's why.

I'd like to once again, post this, just to see if anyone is listening/paying attention.

Note: You have to click the Tweet to see the entire thing.

Read on Twitter

Thanks for the response! It's definitely possible Wilt's underrated. In this board's defense, I do think there's more uncertainty with Wilt given we have less film and fewer statistics, at least compared to the more recent players. I see him as clearly in this Top Tier of peaks and clearly a top 10 peak of all time. If I were lower on his ability to "put it all together" (see below), he might end up on the lower end of this tier, while if I thought he successfully put it all together, he'd end up on the higher end of this tier.

One thing that stuck out to me after reviewing the tweet though: WOW, that's a difficult playoff run! I know you've been doing research on playoff opponent difficulty, where Wilt always is near the top of playoff opponent defense (viewtopic.php?f=64&t=2185164). Thanks for your great research with this! I haven't gone through the most recent update (yet!), but it's been a joy to read so far. But seriously... facing Bill Russell and Nate Thurmond in 11/15 playoff games.... By my account, that means 1967 Wilt played 73% of his playoff games against the GOAT defender or the GOAT Big Man man-defender.

I wish we had better metrics for the 60s! Maybe one day we could get Squared2020 enough games (and time!) to go through and supply RAPM. Pasting my Wilt statistics here (using Curry as the barometer, as I've been doing throughout this project):

5. Curry vs Wilt: Plus-minus based stats:
Ai. AuPM (no data available for Wilt. 2016 Curry 2nd all time).
Aii. Postseason AuPM (no data available for Wilt. 2017 Curry 2nd all time).
Bi. Goldstein RAPM / Historical Square2020 RAPM: (miniscule sample available for Wilt, far below Curry’s 7th all time).
Bii. Goldstein Playoff PIPM (3 years for sample size): (no data available for Wilt. Curry’s 8th all time)
Additional plus minus stats: C. on/off: (no data available for Hakeem. Curry’s 1st all time)
Additional plus minus stats: D. WOWY: 65-70 Wilt >= 2016-2017 Curry (not sure about full prime WOWY. I brought in 16 Curry because Ben hasn't finished publishing Curry's mid/post-2017 WOWY numbers yet).
Additional plus minus stats: E. ESPN’s RPM: (no data for Wilt. 16 Curry 2nd all time)
Additional plus minus stats: F. Backpicks’ CORP evaluation: 1967 Wilt > 2017 Curry (healthy 2016 Steph Curry 4th all time > Wilt.)

Box score-based data
Gi. Backpicks BPM: 2017 Curry > 1967 Wilt (and healthy 2016 Curry is 2nd all time)
Gii. Postseason Backpicks BPM: 2017 Curry (4th all time) > 1967 Wilt
Hi. BR’s BPM: (no data for Wilt. healthy 2016 Curry’s 4th all time)
Hii. BR’s Postseason BPM: (no Data for Wilt).
Additional box score stats: Ii. WS/48: 1967 Wilt > 2017 Curry (but healthy 2016 Curry 3rd all time > Wilt)
Additional box score stats: Iii. Postseason WS/48: 2017 Curry > 1967 Wilt
We barely have any metrics to compare vs Wilt. Of the limited sample we do have, Wilt and Curry are tied 3 to 3. Curry leads in the postseason-only stats. If we add a neighboring year (2016 for Curry and either 1966 or 1968 for Wilt), Curry comes out on top in 5/6 stats. Still, we’re missing the majority of my preferred stats, so there's more uncertainty here.

[EDIT: may post Wilt vs Kareem and Duncan statistically if I have time here].
Wilt does show some postseason decline in WS/48, though he retains just as much value in Backpicks BPM. A few of the years he's competing against are behind in the regular season and creep ahead in the playoffs... but after reviewing the competition in your tweet, I wonder how well the postseason metrics we do have for Wilt capture the opponent difficulty. There's no way any of Kareem, Shaq, Duncan, Hakeem, Curry faced playoff defenses as good as Bill Russell and Nate Thurmond in 73% of your playoff games!

Wilt: Inconsistency year to year gives me a bit less confidence in him, in 1967 it seems like he put things together, and that 76ers team was for sure one of the more dominant teams of all time, as well as him putting one of the more dominant playoff runs ever, but how sustainable is his value? Was that team just a perfect fit and I should have less confidence when picking him to lead my team in a vacuum? It wasn’t too long after his 64 and 67 seasons where he just had a flat-out questionable impact from the WOWY data we had(1965 and 1969)

I'm not the biggest fan of Wilt as a scorer in 1967 and don't think he blended his scoring attack with creating tor teammates at the same time quite as well as those two so it made me hold back on his offense a bit, there were also the weird WOWY indicators across his prime(where some just look straight up unimpressive) where i'm not sure if it was just a best fit situation or not and if I should have the same confidence in him as my other picks who had consistent very high impact indcators throughout their entire prime in different situations/team constructions. I would still give Wilt all 3 of those advantages and personally I could've seen myself voting him as high as like 3 if I viewed him a bit more optimistically, hopefully we can get more discussion going on him in later threads.

Thanks for responding to my question! I see where you're coming from. Wilt changed his play style more than many of the all-time greats over his career. With this stylistic change, the source of his value also changed. Earlier on, his value of course came more from his scoring volume. Later on, it came more from his defense and passing.

This is what I meant before by Wilt "putting it all together." In theory, if you think Wilt could put all these sources of value together at once (his scoring, efficiency, defense, passing, rebounding, screening, etc.), he'd have a tremendous argument for the greatest peak in this tier. A lower interpretation would posit the he could only choose a few areas at once, and thus be lower in this tier. It sounds like we're in agreement on this!

The question then becomes: did 1967 Wilt put it all together at once (i.e. did he successfully balance and combine scoring value, the playmaking value, and the defensive value)? You say you're less convinced on his ability to blend scoring value with playmaking, which I think is understandable -- like you say, if you were higher on his scoring value at his peak, you'd probably bump him in your ranking. To me, I think he still maintained enough scoring value in 67 to be in the conversation, which probably explains why I have him higher. I think he clearly balanced defense and playmaking in 1967. The question is scoring. Per Backpicks' ScoreVal, 67 Wilt is higher than peak Hakeem in the regular season by a lot and in the postseason. He's just behind Shaq and a bit more behind Kareem, with the gap larger in the playoffs. But as I mentioned earlier, I wonder how much the lost value is underrating his opponents defense (he faced Bill Russell and Nate Thurmond 73% of the time).

Why did Wilt put it altogether in 1967, but not in other years (or at least not as effectively)? You mention his inability to balance Scoring, Playmaking, and Defense in other years lowers your confidence in 1967. To me, this comes down to coaching and mindset.
-In the early 1960s (peaking in 1962), there are tons of stories of young Wilt intentionally focusing on scoring, both because of his own preference (he though focusing on scoring would produce the most value) and due to coaching.
-In the mid 1960s (peaking in 1967), Wilt and his coach saying they intentionally wanted to balance scoring, playmaking, and defense. In 1968, Wilt said he intentionally wanted to focus on playmaking over scoring.
-In the early 1970s, Wilt said he further de-emphasized scoring with the approval of his coaches.
-The Argument that Wilt DIDN"T "put it all together": 1) If you're lower on Wilt, I think you could argue his intentional decision to hyper-focus on scoring in the early 60 and playmaking in 68 limited his overall value. Whereas Russell saw the benefit of a balanced offense, Wilt's inability to see the benefits lowers his in-game value. 2) While Wilt was known as a smart player, his IQ advantage tended to be more slow and methodical. You could argue his in-game quick decision making wasn't quite as high. For example (like Hakeem, Duncan, and Durant), Wilt could sometimes get into the mindset of "Okay, now I'm in scoring mode. Okay, now I'm in playmaking mode," without being able to fluidly switch between scoring and playmaking (and even leverage them against each other) using quick decision making, like the best offensive players do. I think there's some merit to both of these counters.
-The Argument that wilt DID "put it all together:" Still, it gives me pause that many of these decisions were approved and even encouraged by the coaching. It's much harder for Wilt to balance scoring and playmaking in 1962 if his coach is actively encouraging him to try to average 50 ppg. I wonder how much worse analytics and strategy in the 60s enabled this behavior. This is partially why I'm actually high on the time-machine argument for Wilt. What if, rather than having a coach encourage only scoring, Wilt was taken to a time where coaches valued more offensive balance (and had the perimeter spacing to better enable that balance)? Wilt was clearly very statistics driven... what if, rather than playing in a time where the best stat to maximize were points per game, he was taken to a time when the best stat to maximize was plus minus per game? Wilt might have had quite a few more years where he "put it all together."

2. Reasoning for Curry:
In short, I think by the data, Curry slightly outperforms Kareem / Duncan, and clearly outperforms Shaq / Hakeem.

1a. Curry vs Shaq:

Spoiler:

Plus minus data
Ai. AuPM: 2017 Curry > 2000 Shaq (and healthy 2016 Curry (2nd all time) >> 2000 Shaq )
Aii. Postseason AuPM: 2017 Curry (2nd all time) >> 2000 Shaq (4th all time)
Bi. Goldstein RAPM: 2000 Shaq (5th all time) > 2017 Curry
Bii. Goldstein Playoff PIPM (3 years for sample size): 2017 Curry (8th all time) > 2000 Shaq
Additional plus minus stats: C. on/off: 2017 Curry > 2000 Shaq
Additional plus minus stats: D. WOWY: 2000 Shaq > 2017 Curry
Additional plus minus stats: E. ESPN’s RPM: 2017 Curry > 2000 Shaq
Additional plus minus stats: F. Backpicks’ CORP evaluation: 2000 Shaq (2nd all time) > 2017 Curry (though healthy 2016 Steph Curry is 4th all time)

Box score-based data
Gi. Backpicks BPM: 2000 Shaq > 2017 Curry (but healthy 2016 Curry (2nd all time) > 2000 Shaq)
Gii. Postseason Backpicks BPM: 2017 Curry (4th all time) >> 2000 Shaq
Additional box score stats: Hi. BR’s BPM: 2000 Shaq > 2017 Curry (but healthy 2016 Curry (4th all time) > 2000 Shaq)
Additional box score stats: Hii. BR’s Postseason BPM: 2017 Curve > 2000 Shaq
Additional box score stats: Ii. WS/48: 2000 Shaq > 2017 Curry (but healthy 2016 Curry (5th all time) > 2000 Shaq)
Additional box score stats: Iii. Postseason WS/48: 2017 Curry > Shaq

In short, 2017 Curry beats 2000 Shaq in 8/14 of these total stats and in 5/5 of the playoff-specific stats. If we add healthy 2016 Curry to the mix, Curry beats Shaq in 11/14 stats. Adding 2001 Shaq to the mix does not help Shaq. The only 3 stats where Shaq beats Curry are Goldstein's regular season RAPM (but not PIPM), WOWY (which is a particularly noisy stat in smaller samples), CORP (which is Ben Taylor's personal evaluation). There are some added contextual factors:

Spoiler:

possible contextual factors worth considering:
1. Scalability. If you value ceiling raising over floor raising, this supports 2017 Curry > 2000 Shaq.
2. Resilience. Shaq may have the advantage here over the course of his prime, but the data universally shows 2017 Curry as the better playoff performer over 2000 Shaq. Further studies have shown Curry does not play statistically worse in the playoffs when he’s healthy (which he is in 2017). Again, according to the data: Playoff 2017 Curry > Playoff 2000 Shaq. (though for overall primes, Shaq > Curry in resilience largely due to health). Speaking of health...
3. Health. Both Shaq and Curry are injury risks. Although Shaq is healthier in his prime, injuries are not a factor in their peak years.
4. Defense. Shaq is the better defender in a vacuum, but I’m not sure he’s that much better relative to position, and the data suggests the defensive advantage is not enough to put peak Shaq over peak Curry.
5. Fit. Both had favorable team circumstances, though in my estimation 2017 Curry had a more favorable fit than 2000 Shaq. Here's a point for Shaq!
6. Time machine. Hard to know for sure. People say Curry would suffer if forced to shoot less 3s in the past; on the other hand, if he found a coach that did let him shoot 3s in volume, his relative offensive advantage would be even greater. People say Shaq would be more valuable offensively playing bully-ball against smaller lineups today; on the other hand, higher pace (with Shaq’s poorer conditioning), stricter big man offensive fouling calls, and a massive increase in the importance of perimeter and pick and roll defense would decrease Shaq’s value.

But put simply, Curry clearly beats Shaq according to the plus minus metrics, and the contextual factors aren't enough to sway me otherwise.

1d. Curry vs Hakeem:

Spoiler:

Plus-minus based stats:
Ai. AuPM: 2017 Curry > 1994 Hakeem
Aii. Postseason AuPM: (no data for peak Hakeem. 2017 Curry 2nd all time)
Bi. Goldstein RAPM / Historical Square2020 RAPM: (no data for peak Hakeem. Partial data in 91/96 and 97 Hakeem are far below Curry, who’s 7th all time).
Bii. Goldstein Playoff PIPM (3 years for sample size): 2017 Curry (8th all time) > 1994 Hakeem
Additional plus minus stats: C. on/off: (no data available for Hakeem. Curry 1st all time)
Additional plus minus stats: D. WOWY: 1993-1995 Hakeem > 2016-2017 Curry (not sure about full prime WOWY. I brought in 16 Curry because Ben hasn't finished publishing Curry's mid/post-2017 WOWY numbers yet).
Additional plus minus stats: E. ESPN’s RPM: (no data for Hakeem. 16 Curry 2nd all time)
Additional plus minus stats: F. Backpicks’ CORP evaluation: 2017 Curry > 1994 Hakeem (healthy 2016 Steph Curry and 1993 Hakeem tied 4th all time)

Box score-based data
Gi. Backpicks BPM: 2017 Curry > 1994 Hakeem (and healthy 2016 Curry is 2nd all time)
Gii. Postseason Backpicks BPM: 2017 Curry (4th all time) > 1994 Hakeem
Additional box score stats: Hi. BR’s BPM: 2017 Curry > 1994 Hakeem (but healthy 2016 Curry (4th all time) > 2003 Duncan)
Additional box score stats: Hii. BR’s Postseason BPM: 2017 Curry > 1994 Hakeem
Additional box score stats: Ii. WS/48: 2017 Curry > 1994 Hakeem (but healthy 2016 Curry (3rd all time) > 2003 Duncan)
Additional box score stats: Iii. Postseason WS/48: 2017 Curry > 1994 Hakeem

2017 Curry beats 1994 Hakeem 4/4 of my more trusted stats and by 9/10 stats total. If we add 2016 Curry and either 1993 or 1994 Hakeem (whichever helps Hakeem more), Curry beats Hakeem in 8/10 stats with 1 tie. The only stats Hakeem ties or beats Curry in are WOWY (which is famously noisy) and CORP (which is Ben Taylor's personal evaluation). In the four of the stats that aren’t old enough for Hakeem, Curry is at least 2nd all time in three of them. In short: I don't think there's any statistical argument for Hakeem > Curry.

1c. Curry vs Kareem:

Spoiler:

Plus-minus based stats:
Ai. AuPM (no data available for Kareem. 2016 Curry 2nd all time).
Aii. Postseason AuPM (no data available for Kareem. 2017 Curry 2nd all time).
Bi. Goldstein RAPM / Historical Square2020 RAPM: 1985 Kareem (6th all time) > 2017 Steph Curry (7th all time) (But only a 41 game sample for Kareem. Earlier Historical RAPM years are lower but with smaller sample)
Bii. Goldstein Playoff PIPM (3 years for sample size): 2017 Curry (8th all time) > 1977 Kareem (10th all time)
Additional plus minus stats: C. on/off: (no data available for Kareem. Curry 1st all time)
Additional plus minus stats: D. WOWY: Kareem > 2016/2017 Curry (not sure about full prime WOWY. I brought in 16 Curry because Ben hasn't finished publishing Curry's mid/post-2017 WOWY numbers yet).
Additional plus minus stats: E. ESPN’s RPM: (no data available for magic. 2016 Curry 2nd all time)
Additional plus minus stats: F. Backpicks’ CORP evaluation: 1977 Kareem > 2017 Curry (but healthy 2016 Steph Curry is 4th all time over Kareem)

Box score-based data
Gi. Backpicks BPM: 2017 Curry > 1977 Kareem (and healthy 2016 Curry (2nd all time) > 1977 Kareem)
Gii. Postseason Backpicks BPM: 2017 Curry (4th all time) > 1977 Kareem (7th all time)
Additional box score stats: Hi. BR’s BPM: 1977 Kareem > 2017 Curry (but healthy 2016 Curry (4th all time) > 1977 Kareem)
Additional box score stats: Hii. BR’s Postseason BPM: 1977 Kareem > 2017 Curry
Additional box score stats: Ii. WS/48: 2017 Curry > 1977 Kareem
Additional box score stats: Iii. Postseason WS/48: 1977 Kareem > 2017 Curry

2017 Curry beats 1977 Kareem by 3/4 of my more trusted metrics, though 1977 Kareem beats 2017 Curry by 6 metrics to 4 total. Their playoff-only stats are tied 2-2. In all four of the stats that aren’t old enough for Kareem, Curry is either 1st or 2nd all time. If we add 2016 Curry and either 1976 or 1978 Kareem (whichever helps Kareem more), Curry beats Kareem by 6 to 4 stats. It's close (closer than Shaq), but I think the added context favors Curry.

1d Curry vs Duncan:

Spoiler:

Plus-minus based stats:
Ai. AuPM: 2017 Curry > 2003 Duncan
Aii. Postseason AuPM: 2017 Curry > 2003 Duncan
Bi. Goldstein RAPM / Historical Square2020 RAPM: 2003 Duncan > 2017 Curry
Bii. Goldstein Playoff PIPM (3 years for sample size): 2003 Duncan (1st all time) > 2017 Curry (8th all time)
Additional plus minus stats: C. on/off: 2017 Curry (1st all time) > 2003 Duncan
Additional plus minus stats: D. WOWY: 2003 Duncan > 2016/2017 Curry (not sure about full prime WOWY. I brought in 16 Curry because Ben hasn't finished publishing Curry's mid/post-2017 WOWY numbers yet).
Additional plus minus stats: E. ESPN’s RPM: 2017 Curry > 2003 Duncan (2016 Curry 2nd all time)
Additional plus minus stats: F. Backpicks’ CORP evaluation: 2017 Curry > 2003 Duncan (and healthy 2016 Steph Curry is 4th all time)

Box score-based data
Gi. Backpicks BPM: 2017 Curry > 2003 Duncan (and healthy 2016 Curry is 2nd all time)
Gii. Postseason Backpicks BPM: 2017 Curry (4th all time) = 2003 Duncan (tied 4th all time)
Additional box score stats:Hi. BR’s BPM: 2003 Duncan > 2017 Curry (but healthy 2016 Curry (4th all time) > 2003 Duncan)
Additional box score stats:Hii. BR’s Postseason BPM: 2003 Duncan > 2017 Curry
Additional box score stats: Ii. WS/48: 2003 Duncan > 2017 Curry (but healthy 2016 Curry (3rd all time) > 2003 Duncan)
Additional box score stats: Iii. Postseason WS/48: 2003 Duncan > 2017 Curry

Curry beats Duncan in 3-2 by my more trusted metrics, but Duncan beats Curry 7 stats to 6, including 3 to 2 postseason stats and 1 postseason tie. If we add 2016 Curry and either 2003 or 2004 Duncan (whichever helps Duncan more), Curry beats Duncan 8 to 5.


Duncan and Kareem have the closest arguments, but I side with Curry. I think Curry gains clear separation over Shaq and especially Hakeem.

Counter to Curry 1: Fit.
The team around Steph was optimal fit (arguably better fit than these opponents), and the team was dominant. But the data seems to suggest the team's dominance was primarily driven by Curry. (Chuck might say Curry was the bus driver ).

From 2017–2019 (larger sample to give more stable values), here's the net rating with each of the stars on or off:
-All 4 stars on: +17. (that's 20% better than the 1996 Chicago Bulls across 3 seasons!)
-Only Klay off: +15.64.
-Only KD off: +13.54 (still better than the 96 Chicago Bulls even with KD off)
-Only Draymond off: +12.77
-Only Steph on, all 3 other stars off: +10.81
-Only Steph off: +1.94
With all 3 other all stars off, and just Steph on, the 17-19 Warriors have a better net rating than the 16 Warriors, 13 Heat, 2000 Lakers, 91 Bulls, 87 Lakers, or 86 Celtics. With all 3 all stars on, and just Steph off, the 17-19 Warriors are worse than this season's 2022 Cavs. This pattern remains in the playoffs (more info below).

Counter to Curry 2: Resilience.
I would agree that the other all time peaks might be more resilient than Curry. But studies have shown this decline almost entirely correlates with postseason health. Per Per BPM and AUPM, Curry actually improves in the playoffs when he's healthy. Though others improve more in the playoffs, the difference isn't significant enough to sway me (e.g. Shaq's career +0.67% improvement vs Curry's career +0.57% improvement), particularly when 2017 Curry outperforms his opponents per the above statistics.

More in depth discussion of Curry's context and counters here:

Spoiler:

Steph Curry Discussion:

.
.
I think you both voted for Shaq first in this ballot. He definitely had a great peak! I also appreciate the reasons you gave, and I’ll try to give a more in depth response if I have time.

Do either of you have any counters for the statistical case I made on page 2? Looking across a variety of all-in-one metrics (which use actual plus minus data, plus minus data that adjusts for teammates, and box score estimatjons of plus minus data), Shaq looks like he’s below Curry, playoff Kareem, and arguably Duncan. If you’re less swayed by the data, 70sFan also did a film analysis in Peak Thread #2 I found convincing.

On another note, there’s already been some discussion of Hakeem, but I haven’t seen as much back and forth discussion about Wilt vs Kareem or Shaq. I think 67 Wilt clearly is the better defender, rebounder, and passer of the three. I imagine it’s the scoring gap in 67 which pulls Wilt down for people? Are there any other reasons (e.g are people less high on Wilt’s playoff performance)?

Really the way I saw the data was that rankings aside, the actual values seem so close to eachother between some of those players that you honestly could mostly still just go either way. I'm not too big on single season hybrid metrics like I said for KG as a be-all-end-all, especially in the playoffs due to the smaller sample size. But, while there was a chance Steph was still just better in his run compared to 2000 Shaq I would say he was probably in a more optimal situation, and think Shaq's game just had a longer sample size proving it's resilience both at an individual and team level at his peak without significant changes in his skillset, whether it was from Steph's durability or other factors - teams really did try so hard to wear him down and I do see slightly more dueability issues, so I just defaulted to him when making my list. I appreciate it though and think you brought up some interesting points.
...

Good points! It sounds like ike there were a few factors for why you have Shaq > Curry:
1) You're less a fan of hybrid metrics or plus minus data for a single postseason. This adds more uncertainty to the fact that 2017 Curry beats 2000 Shaq in 5/5 of the playoff-specific plus/minus stats I listed.
2) Curry's team fit was better.
3) Curry's less resilient than Shaq.
As always, let me know if I misunderstood or missed anything!

For 1), I mentioned some of my counters to this in my Kareem Discussion above, so I won't rehash it too much here -- the main points were that box plus/minus stats are stabler in small sample sizes like the playoffs, and the PIPM stat I listed was for 3 playoffs which should help with the sample size issue. Personally, I find the fact that Curry beats Shaq in all 5/5 playoff stats pretty compelling, though they're definitely close!
2) I tend to agree here. The Warriors were definitely great scenario for Curry, with great coaching, great defenders, and great offensive players alongside Curry. Still, Shaq also had some great coaching, a great running mate in young Kobe, and great supporting cast (especially in the 2001 playoffs, Shaq's supporting cast had a clear 3 point shooting advantage vs league average).

Personally, I find Curry's scalability/portability enough to make up for this. Historically, it's easier for players to have their best plus minus value with worse supporting casts (or at least with no co-stars competing for value). The better the supporting cast and the better the co-stars, the more the team improves but the harder it is for a single individual to dominate the team's value (since they're competing for value against better teammates). And yet... Curry dominates the Warriors in terms of value.

From 2017–2019 (larger sample to give more stable values), here's the net rating with each of the stars on or off:
-All 4 stars on: +17. (that's 20% better than the 1996 Chicago Bulls across 3 seasons!)
-Only Klay off: +15.64.
-Only KD off: +13.54 (still better than the 96 Chicago Bulls even with KD off)
-Only Draymond off: +12.77
-Only Steph on, all 3 other stars off: +10.81
-Only Steph off: +1.94
When all four stars are on the court, the 17-19 Warriors are significantly better than the 1996 Bulls. With all 3 other all stars off, and just Steph on, the 17-19 Warriors have a better net rating than the 16 Warriors, 13 Heat, 2000 Lakers, 91 Bulls, 87 Lakers, or 86 Celtics. With all 3 all stars on, and just Steph off, the 17-19 Warriors are worse than this season's 2022 Cavs.

If we include both the regular season and the playoffs, the difference decreases, but Curry still dominates (only KD off: +11.08. only Steph off: +3.66). Lots of people have said that Curry's had a better fit than other peaks, and that his team was stacked. This is true. But, as far as I can tell, they only dominated when Curry was on the court, and they completely fell apart (by their standards) without him. Source: https://www.pbpstats.com/wowy-combos/nba?TeamId=1610612744&Season=2018-19,2017-18,2016-17&SeasonType=Regular%2BSeason&PlayerIds=201142,201939,202691,203110

3) I agree here too, Curry is less resilient than Shaq. But I wonder how much of that decline comes from Curry's worse postseason health during his prime? Per Thinking Basketball's video, Curry on average declined by BPM from the regular season to the postseason (from 2013-2018). But if we only take healthy playoffs (2013-2018 except 16 and 18): Curry improves by +0.2 BPM. By AuPM, Curry improves by +0.57% from the regular season to the playoffs, even when including injured postseasons (from 2013-2021). With another healthy postseason in 2022, Curry improves by even more in both metrics.
Now it's true that Shaq improves by more according to AuPM: +0.67% (from 1994-2006, 2008, 2010). But personally, I don't find +0.57% vs +0.67% a big enough difference to sway me in favor of Shaq.

Like you hint at, the biggest cause of Curry's playoff decline is health concerns. And you're definitely right that prime Steph had greater health issues than prime Shaq. If you take different approach on health that I do (e.g. docking players in every season a little bit because they're injury risks), I could see that swaying you against 2017 Curry. And that would be a valid approach! From my point of view though, Curry was healthy in 2017, and he outperformed 2000 and 2001 Shaq in all 5/5 of my playoff-only plus minus metrics. So... he was the more valuable playoff performer then (at least to me ).

Could someone lay out the argument for 2017 as Curry's peak vs 2016? I just don't see him being better that year unless it's got to do with health.

I'm aware that he peaked in many APM stats in 2017 but personally I try not to rely on +/- stats so much because they're so heavily influenced by team context. Over large, multi-year sample sizes that encompass different team situations they can absolutely give you an idea of a player's intangible impact, but over a single season there's still so much room for variance.

I like to evaluate players from a skillset POV and watching Steph that year he didn't feel like a better player than he was in 2016 at all. He was maybe slightly better as a defender but as a point guard that isn't really gonna make a huge dent and his shot making was a far cry from what it was in 2016. In real time I actually thought it was a pretty weak RS from him until Golden State's 15 game run without KD at the end of the year.

As for his postseason I view it as an outlier for him in the same way I don't think 09 LeBron had a switch where he could become a 45% from midrange, 37 PER player at will. For that same reason I'm also probably more forgiving of Steph's 2016 postseason than most. Even as unbelievable as he is, as a shooter, Steph is more prone to variance than other all time greats, which is why it's hard to take any 10-15 game sample as his "true" level of play, if that makes sense. His 2018, 2019, and 2022 runs are probably closer to what you can expect from him on average come playoff time.

Then you can get into how favorable his circumstances were. The talent gap between Golden State and everyone else was just comical, and the three teams he faced in the west all had their best defenders (Nurkic, Gobert, Kawhi) either missing or hobbled. In the finals the Cavaliers were the 27th ranked defense and JR and Kyrie, two notoriously boneheaded defenders, made miscommunication after miscommunication which torpedoed the Cavs. Not to say that it was completely unrelated to Steph's gravity, but I recently rewatched some games from that series they were making the same dumb mistakes over and over again defending Iguodala and KD in transition with Steph not even on the court. They defended him well the year before so more than anything it felt like they were overwhelmed by Golden State's firepower as a whole.

Was there a change to his approach on offense that I might've missed? Was his movement off the ball especially crisp that year or something? Even so, I don't think player's intangibles typically vary season to season as much as people pretend they do to justify what the +/- stats are telling them. People love to say the box score doesn't matter but that simply isn't true when it comes to comparing players to themselves playing the same role just a season ago.

FTR I've got no problem with Curry's peak being ranked highly (I wouldn't go as high as top 5 though) it's just that I view 2015, 2017, and 2018 all similarly and a clear tier down from 2016.

So, big thing:

I don't think Curry fits that well in a season-based peak conversation because different years have different arguments for and against him.

'14-15 wins the MVP and the championship.
'15-16 is his best regular season, and his worst (or near worst) post-season.
'16-17 is the smoothest season - MVP of the greatest team in the history of basketball - and his best statistical playoffs, but the degree of difficulty can be said to be all-time low.
'21-22 proved his capacity for latent impact like never before and seemed the most resilient in the playoffs ever, but had regular season cold streaks causing him to have the lowest TS Add (shooting volume * relative league efficiency) of his entire prime.

Which year to pick? I honestly don't feel super strongly about it other than I really don't think '14-15 should be in the conversation.

Great points letskissbro and Doctor MJ! To answer your question in short, here are the changes I see from 2016 to 2017:
1. Postseason health: you mention this of course. If you take a more probabilistic approach to injury (e.g. partially docking every season based on perceived chance of injury, rather than fully docking injured seasons and not docking healthy seasons), I could definitely see you taking 2016 > 2017. I think it's hard to estimate injury probabilities though, and since this is a 1 year peak project, I just go with the simple route: 2017 was healthy, so he doesn't get docked! If we're just looking at skill improvements, I think 2017 was slightly better at pacing himself through the regular season (a la 2013 LeBron vs 2009, 1977 Kareem vs 1972/71), which helped him improve his chances of being fresh and healthy in the postseason.
2. Improved resilience: increased weight and strength allows for better defense and better ability to take contact on and off ball. Slightly improved decision making allows for slightly fewer boneheaded turnovers and more consistent (but less flashy) passes / shot selection (Athletic Alchemy mentioned he noticed this in his film study at some point ~2017). Slightly tighter handle (which would continue to improve and show noticeable results by 2022).
3. Improved scalability: I have Curry as the GOAT portable/scalable player, and he's certainly still all-time scalable in 2016. However, the actual act of adjusting to Durant forced Curry to improve even more without the ball and be even more efficient with his on ball usage.

I think I also have fewer concerns with the 2017 regular season: I see 2017 regular season Curry as just as good as 2016 regular season Curry, though definitely less valuable. By good, his actual goodness as a player (irrespective of team context); value is more context-dependent.

So why did his performance and value decline? I think fitting next to Durant (and making the Scalability improvements I mentioned in #3 above) actually took some work in the first quarter/half of the season. This took him out of rhythm and slightly decreased his impact metrics. In interviews, Curry actually says just that! He says he was pulling back too much in order to make room for Durant in the early part of the season, but by the end, he had figured out how to go full force without taking away from Durant's offense.

Statistically, Curry's performance improved in the second half of the season with Durant, when Durant was out for those last few weeks, and when Durant was back from the playoffs. By my eye (and from what I can tell by the metrics), Curry was just as good as he was in 2016 regular season during that span, which gives me confidence that he was just as "good" of a player in the 2017 regular season (even if figuring out how to fit alongside Durant in the first half of his season lowered his regular season impact metric value).

Like Doctor MJ says, Curry continued to make improvements in other areas over the next few years. By 2022, he had noticeably better defense, decision making, passing, handle, and resilience. I still take 2017 (and healthy 2016) because of his athleticism (specifically his speed/quickness) and his overall impact metrics. Like other people have said though, if the 2022 regular season ended up just being a cold spell (not an actual decline in shooting due to age/motor/athleticism/increased effort elsewhere), then 2021/2022 might also have an argument at his peak. Let me know if you disagree or if you have any questions!

3. Reasoning for Kareem: There's also an argument for Kareem > Shaq.

Spoiler:

Plus-minus based stats:
A. AuPM (no data available for Kareem. Shaq 6th)
Bi. Goldstein RAPM / Historical Square2020 RAPM: 2000 Shaq(5th all time) > 1985 Kareem (6th all time) (But only a 41 game sample for Kareem and non peak).
Bii. Goldstein Playoff PIPM (3 years for sample size): 1977 Kareem (10th) > 2000 Shaq
Additional plus minus stats: C. on/off: (no data available for Kareem)
Additional plus minus stats: D. WOWY: Shaq > Kareem
Additional plus minus stats: E. ESPN’s RPM: (no data available for Kareem)
Additional plus minus stats: F. Backpicks’ CORP evaluation: Shaq > Kareem

Box score-based data
Gi. Backpicks BPM: 2000 Shaq > 1977 Kareem (though 71/72 seasons compare)
Gii. Postseason Backpicks BPM: (2017 Curry 4th all time) > Kareem (7th all time) > 2000 Shaq
Additional box score stats: Hi. BR’s BPM: Kareem > Shaq (just barely)
Additional box score stats: Hii. BR’s Postseason BPM: Kareem > Shaq
Additional box score stats: Ii. WS/48: Shaq > Kareem (just barely)
Additional box score stats: Iii. Postseason WS/48: Kareem > Shaq

Shaq wins 5/10, Kareem wins 5/10. Kareem wins 4/4 playoff-only numbers. Adding 2001 Shaq and 78 Kareem doesn’t make a difference. If you value the playoffs, Kareem has a surprisingly compelling case over peak Shaq. Check out 70sFan's previous comment for some fascinating film analysis vs Shaq here: https://forums.realgm.com/boards/viewtopic.php?f=64&t=2204332 (Greatest Peak Thread 2, Comment 2).

Counters against Kareem: 1. lack of data (which might allow e.g. Shaq > Kareem), and 2. smaller playoff sample size, and possibly 3. some amount of coasting (i.e. worse regular season plus/minus metrics) in the regular season compared to younger Kareem or possibly other Greatest Peak competitors. Earlier Discussion of the concerns can be found in Thread 2 (https://forums.realgm.com/boards/viewtopic.php?f=64&t=2204332&start=29) ~page 2 between me, 70sFan, and falcolombardi. More recent discussion of concerns here:

Spoiler:

...

Really the way I saw the data was that rankings aside, the actual values seem so close to eachother between some of those players that you honestly could mostly still just go either way. I'm not too big on single season hybrid metrics like I said for KG as a be-all-end-all, especially in the playoffs due to the smaller sample size.
...
The Kareem Vs Shaq stat-based argument is missing alot of the measures I think are more important(really no +/- or APM data) so i'm not too big on doing a comp between the 2 using mostly box score metrics.
70sfan did provide some compelling points(especially on defense which is why I removed 2001 Shaq from being at the top of my voting) but I also think one of the main differences in how he and I voted was that I don't believe Kareem's offense reached quite the same heights because I believe his creation was less valuable with the lack of 3 point shooting at his peak in the 70s - which he doesn't agree with doing I think.

As I read your argument, what I'm hearing is: 1) we don't have every available metric for Kareem which limits are statistical analysis, 2) The statistics for Shaq vs Kareem are super close overall, and 3) even if the playoff metrics favor Kareem, playoff sample size is an issue for single seasons. Thus, given how close it is, we can turn to other sources of analysis (e.g. Kareem's playmaking vs Shaq's, etc.), where Shaq ends up winning in your estimation (possibly due to his era, but you don't think we should do the time machine argument as much for Kareem's playmaking). Let me know if this summary missed anything!

Overall, that's fairly convincing! I see where you're coming from. I'll push back slightly on a few points (just because I'm enjoying the discussion! ), but overall I follow the reasoning and think it's a perfectly reasonable conclusion:
1a) For point 1, I absolutely agree that there's less data for Kareem. We're missing AuPM (regular season and postseason), full season RAPM/PIPM in Kareem's prime, and ESPN's RPM. We do still have Squared2020's Historical RAPM for 1985 Kareem. It's for a half season sample, and Kareem just pops off the page, which is particularly impressive given how many years after his peak he was. Still, it's a smaller sample size and different context than 1977 Kareem. Thinking Basketball also has 3 year Goldstein Playoff PIPM in his final video of the greatest peaks series (the top 10 summary). Not sure how he got his hands on it, but in it, 1977–1979 Playoff Kareem > 2000 –2002 Shaq.
1b) I think I'm not as low on box-score composites like Backpicks Box Plus Minus as you are. I think some of my confidence in it comes from this study (https://fansided.com/2019/01/08/nylon-calculus-best-advanced-stat/), which tests the predictive power of various box score stats and advanced stats. Backpicks BPM and Basketball Reference's BPM may not be as good as PIPM, but they still do a fairly good job (and a heck of a lot better than other stats like PPG and PER).
2) Agreed, the overall statistics are fairly close for both.
3) While it's true that one postseason sample is small for true plus/minus stats, the Box-score based estimations are much more stable (that's actually one of the reasons for using them)! Plus, Thinking Basketball somehow got his hands on 3-year playoff PIPM for Kareem, which is a larger sample. Kareem outperforms Shaq in both the more stable box-score estimations and the larger-sample 3 year playoff PIPM.
For those reasons (along with 70sFan's film analysis, which you mentioned), I think I've been convinced to go Kareem over Shaq. But still, I see where you're coming from.

Kareem to me is the guy from the bunch I'd be most considering here. I would consider Kareem's offense to be better than Olajuwon's, so it's then a question of how strong of an edge I'd give to Dream on defense where I think Kareem's prime was also excellent on defense.

Re: Kareem's '74 & '80s rTS%. It has to be noted that this wasn't his efficiency edge every post-season. While this is a peak project where in theory it makes sense to ignore weaker years, I think with Kareem we definitely saw him being more vulnerable to defensive match-ups than Olajuwon seemed to be, and while Kareem has a significant raw TS% edge in his career in the regular season, that basically disappears when we look over to their playoff careers where - let's note - Hakeem has the edge on PER and BPM (while Kareem maintains a slight lead by WS/48).

It's true that playoff Kareem is better in 77 than 74 and 79-80. The counter that 70sFan offered was that he was at his peak in 77 (and the surrounding years), but we lack any playoff sample in 75 or 76 -- if he had a larger sample in those years, he might show more playoff consistency closer to the heights he reached in 77 (though its hard to know for sure).

I also wonder whether the decline in other years is as big as you say. While his postseason Backpicks BPM peak was in 77 (7th best all time), he did get fairly close in 74 (would be 10th all time). Both playoff runs are over Shaq, the 77 playoff run is better than Hakeem's best by some separation, and the 74 playoff run is only worse than Hakeem's by 0.1. It's also worth mentioning his regular season BPM in 74/78/79/80 are higher than 77... this could make you lower on 77, or it could make you higher on the consistency of Kareem from 74-80, depending on your interpretation. What do you think?

Also... I don't see what you mean by his TS% advantage declining in the playoffs against Hakeem. He has a relative TS% of +13.7% in 1977, 12.5% in 79. 9.9% in 80, and 9% in 74. If we're only lookin at the years around their peak (Hakeem admittedly had better rTS% when he was younger), peak Hakeem's best playoff value is in 94 at +4.9% rTS. It looks like Kareem has the significant rTS% advantage in the playoffs too, at least by my interpretation. Do you disagree?