Top 25 peaks of the 2001-25: #1-#2 Spots

[Doctor MJ]

It seems like a weird term to use for a guy almost nobody thought was the best player in the league at that time, and has been retrospectively elevated by some due to advanced stats that are often wrong (and yes, I'm including plus minus, regardless of whether you want to call it advanced or not).

I just have to interject here with the historiography:

The consensus was literally that Garnett was the best player in the world during the '03-04 season and that's why he won the MVP in a blow out. That's not to say he'd ended the debate with Duncan generally, but rather that it was very much a Duncan vs Garnett debate.

It was the last years in Minnesota that re-classified Garnett as something a tier below Duncan, and why was that? Not because of a direct re-evaluation of Garnett based on his play, but based on the fact that his team wasn't good and everybody "knew" that that meant he couldn't be as good as Duncan whose teams were always good.

Regardless of whether that re-classification is something you judge to be right or wrong, that's literally what happened. Garnett went from a guy that people really weren't skeptical of to a guy that people were skeptical of from the span of '04 to '07, rather than an actual criticism of his play in '02-03 or '03-04.

To be clear, that doesn't mean people thought Garnett was as good as prime Shaq, because prime Shaq was considered in a tier separate from Duncan & Garnett, but just in terms of Garnett & Duncan, it really wasn't until '05-06 and especially '06-07 that they became perceived as being in different tiers with Garnett relegated to a tier below where Duncan by then was perceived.

And also to be clear: I was right there with everyone else in that time frame for the most part with the main exception being that I made clear that I had uncertainty about Garnett's placement because I wasn't sure how much of Minny falling off a superstar should have been able to prevent. When he got moved to Boston, I absolutely did not predict the success he would have there, but I did acknowledge that if he had major success that might re-frame my assessment of his Minnesota years... and of course that's eventually what happened.

No, KG won the MVP in 2004 because:

1) voter fatigue, after 2 consecutive MVP wins by Duncan

2) Duncan playing only 69 games that year, and working his way back from a foot injury, and

3) narrative reasons, e.g. “KG is due an MVP” and “Wolves finished higher”

I don’t even think casual fans thought KG was better than Shaq at this point, and many discerning fans still had him below Duncan.

The Spurs were 51-18 in games Duncan played, and only 6-7 in games he didn’t. If Duncan plays a full season like he did the previous 2 years, the Spurs win more games and he gets 3 straight MVPs.

I’m sure KG had many fans, I know he did because I followed the commentary at the time, but he was not the consensus best player in the world in 2004. No more than Dirk was when Dirk won an MVP in 2007. Voters said “well, nobody else has a bullet proof case, and the Mavs had a great record, and Dirk is kind of due, let’s give it to him”.

In short, I don’t agree with the narrative you have presented. Maybe for you that was the narrative, but it certainly wasn’t mine at the time, or many others. To describe it as “the consensus” strikes me as wrong. People did not expect the Wolves to win the title, the thought (correctly) that the Lakers were going to roll them. A big reason for that was KGs limitations. You can’t run the post-centric offense through him, and he can’t guard Shaq 1v1 like Duncan can, or act as the team’s defensive anchor. Not that he didn’t “anchor” the Wolves (and Celtics) D, he absolutely did, but he did it through his versatility, etc, and not rim protection which is more valuable.

While I appreciate you putting other reasons below, the fact you began with voter fatigue is just so problematic to start it speaks to taking shortcuts in process.

The term "voter fatigue" came into vogue to describe years in which voters still felt like the prior MVP had a strong case, but just enough of them went with someone new to swing the tally to the new guy.

So with Malone topping Jordan in 1997 or Embiid topping Jokic in 2023, Jordan & Jokic still got first place votes.

But Duncan in 2004 got no first place votes at all, and when you look at b-r's voting table and see Garnett & Duncan's stats, you see Garnett led Duncan in PPG, RPG, APG & SPG while not being far behind in BPG. He led by all the all-in-one metrics, and yeah, he led the better team.

Did we back then note that Duncan missed some time and wasn't quite at his best that season? Sure, as I said, the Garnett vs Duncan debate continued.

Did people generally feel like Duncan would have won the MVP if not for voter fatigue? Absolutely not, and when you assert otherwise, I have to question your memory.


Re: 2007 Dirk "voters said". This is a season where it is more appropriate to argue that voter fatigue was a factor, and you can tell because the guy who won the prior two MVPs got 44 first place votes rather than the zero that Duncan got in 2004. As noted, the rule is that people only consider it "voter fatigue" when there are voters going in the other direction, because it's largely the people who lobbied for the "fatigued" candidate who make the allegation.

I will say, as a general pro-Nash guy, I didn't see the 2007 vote as being actually determined by fatigue. I think Dirk wins that MVP regardless of who won the prior MVPs, and if I were to point to a single factor it would be the team record. Now, it's certainly a thing for people to question whether a player was winning the MVP simply because his team had the best record, and thus there is a path generally to say that the MVP wasn't seen as the best player of that season, but when a player has overwhelming box score advantage - as Garnett did over Duncan in '03-04 - this isn't a thing that people say.

The pro-Duncan side in 2004 wasn't about arguing that he should win MVP because he played better, or played better per game - else he'd certainly have gotten actual 1st place MVP votes - but that when he was 100% healthy and available he was the better player.

Last "people didn't expect Wolves to win the title, they thought the Lakers would". Well of course that's what they thought. The Lakers were expected to win in '02-03 as well and the general feeling was that the only reason anyone could beat the Lakers was because Shaq let himself get obese.

Such was the perceived unbeatability of the Shaq-Kobe Lakers that even after they got utterly annihilated by a Pistons' defense, people still thought that the Lakers should have won if they were healthy and playing their best. The fact that the Piston D had essentially rendered the Laker offense built around Shaq in the post obsolete wasn't understood by fans then... though to be fair, it's probably not well understood by them now either.

[Owly]

I feel like the discussion on here must be tough for outsiders sometimes. There are these long posts replying to each other with 10 formulas, and replies with another 8 formulas and discussion of variance or standard deviations or whatever. I’m not saying stats don’t have their uses, and aren’t of interest, but I don’t think various advanced stats (or plus minus ones) should be a meaningful argument for why one player is better than another (and I say this as someone whose top 2 guys are god tier by most advanced stats).

There are so many reasons advanced stats can be wrong, and even the most ardent stats zealot would tell you as much. It can be due to superior/inferior back-ups, or match-ups, or the style/game-plan of teams, or it could just be random variance that happens in basketball games. It’s an interesting data point, but to my mind that’s all it should be; a data point, where we say “hmm, that’s interesting, it might hint at these broad conclusions”, but not “it means X”. too often advanced stats ARE the argument.

I’m more interested in a nuanced analysis of the context, and whether what happened really bears out the numbers. In the case of KG, the best example is the comparison between 2002 KG and Tim Duncan. KG and Duncan were both basically at their peak in 02. Most will say it was 03 for Duncan, and 04 for KG, because of team success, but really they weren’t meaningfully different in 02.

If you look at the Wolves, they had a clearly superior support cast that year. KG’s robin was Brandon at first, who was an all-star calibre point guard in the vein of Mike Conley. He didn’t put up big numbers, but he was a big-time player. Then when he went down with injury, Billups took over and proceeded to play like an all-star. Most didn’t realise how good Billups was yet, and obviously he got a little better in future years, but I feel comfortable to say he was playing at an all-star calibre level this year. Then KG had Wally, who actually made the all-star team. In addition, the Wolves had Joe Smith who was a fantastic starter, and rounded out the starting 5 with Rasho. Rasho was a decent 5 man. His help D was good, his man D wasn’t so good, and he had a decent midrange shot most times. Then off the bench KG had other decent role players like Peeler, Gary Trent, etc. It was a solid support cast, especially for that period.

If we look at Duncan, his support cast was nothing like that strong in 2002. He didn’t have anyone even remotely resembling an all-star calibre player. D.Rob was a shell of himself. Parker was an inconsistent rookie who a year later would lose his minutes in the finals to a journeyman point guard named Speedy Claxton. Steve Smith was washed. All he could do was hit open 3s and nothing else. Bruce Bowen was an elite perimeter defender (in the 59 games he played), but he could do nothing but shoot open corner 3s on offense. Given the era, he would have been unplayable if Duncan wasn’t opening up space for him with double teams. The bench has trash players like Malik Rose, Charles Smith, and Antonio Daniels.

Yet despite that clear difference in team quality, Duncan’s Spurs won 58 and had an SRS of 6.28, compared to the Wolves who won only 50 and had an SRS of 3.58. The Wolves were swept in the 1st round by the Mavs by a large margin, while the Spurs had a much closer than it looks 4-1 loss to the champs in round 2 despite D.Rob basically missing the series with injuries (and being almost useless when he did play). The Spurs lose game 1 by 6 points (led going into the 4th), game 3 by 10 points (down 3 going into the 4th), game 4 by 2 points (up 8 heading into the 4th), and game 5 by 6 points (down 1 heading into the 4th). Duncan did all he could that series, but in crunch time his complete lack of support was too much. Stat lines can be misleading, but I think the closing game one speaks loudly as to what I was seeing. Duncan has 34-25 in 45 minutes, as the Spurs ran almost every play through him while he spent much of the game (and series) matching up with Shaq. Meanwhile D.Rob had 0 points and 3 rebounds in 18 minutes, and was worse than even those numbers suggest.

KG could never carry a team in the way peak Duncan showed he could. He certainly wouldn’t be limiting Shaq the way Duncan did in the 02 series. KG was a great player, but his impact just wasn’t the same as Duncan. If it was we would have seen it in years when his team wasn’t good, the same as we often did for guys like Duncan, Lebron, etc. And before anyone tries to explain it away with the coaching, Flip was a great coach. Pop is obviously a historically better coach, but at the time Pop was just running every other play to Duncan in the post. It was a very simplistic offense compared to what Flip could do. It’s also a players league. Coaches just put players in a position to succeed, they don’t make bad players good. KG had all the coaching he needed to succeed, and he couldn’t to the level of the absolute top guys like Duncan and Lebron. His skillset wasn't suited to it.

I would agree that new metrics and the rate of change can be hard to follow.

I don’t think that would be a good argument not to use them. Perhaps to point out the workings/methodology where possible and thought useful.

“here are so many reasons advanced stats can be wrong, and even the most ardent stats zealot would tell you as much”
Well … I start to tend to disagree when you say “advanced stats can be wrong” … they are what they are and they are, for the most part, wonderfully consistent*. The inputs on a box-metric could be fudged by scorekeepers at the margins, or vary in their real value. The weights of statistical inputs can be poorly aligned with the real value of these things (in general, for specific players, due to changes in the game over time…). The conventional box-score has substantial gaps in covering important things happening (primarily on D) and hasn’t been available for all of men’s major league history. But for the most the inputs and process are fairly transparent (if the working a bit heavy, complex).

And the numbers could be misinterpreted but the main real way I’d say the number is wrong (rather than “a reader or writer has placed undue value on a number”) is if the weights are badly out of whack.

On the non-box side, raw stuff like plus-minus, on-off are unlikely to be “wrong” though input errors could happen and accurate possession count is needed for a precise, fair on-off at the margins (players could see more offensive or defensive possessions and thereby be unfairly rewarded or penalized at the margins). The more you tinker with it and what you do is a black box (even if the broad parameters are known) the more trust you’re putting in a creator (and/or those telling you it’s a good metric). With on-off available (and I assume PbP) you can have somewhat of check on the measures.

If one is saying in a small sample on-off is measuring the one player and not the 9 around him and then the 10 with their replacements ... something is wrong. I’d argue that’s the interpreter not the number. With the context of longer term trends … and as a context to and potential counterpoint to less nuanced use of team performance (perhaps otherwise used without context I’d say it’s a huge asset.

*I’d caveat here on the consistency thing – and I’m more comfortable, acquainted with the Reference metrics …for BPM, having position specific weights makes this less the case. I do sometimes wonder which players would gain/lose most if notional position was altered.


Metrics then, certainly aren't perfect.
I’d say the big positive here, at least for simpler metrics and on-off and plus minus is the process is consistent, transparent and the aggregation of them can be done clearly and consistently. This isn’t to say non-quantative (or not presently quantified, or not presently public-domain quantified) stuff isn’t valuable.

That said
“KG and Duncan were both basically at their peak in 02” …
Were they though …
I can’t know how you came to this conclusion. It may be that you have better “nuanced analysis” of the players in question but it is opaque to everyone else.
Those box aggregates however would provide a simple test …
Duncan PER difference: 0.05 (very marginally up in 2004) 27.01 to 27.06
WS/48 difference: -0.0082 (marginally down in 2004) .2569 to .2487
BPM difference: 0.96 (a little up in 2004) 7.57 to 8.53
So I would say in aggregate the difference by these tools can perceive seems to be marginal (2004 marginally preferred). “Basically” the same … would seem an acceptable conclusion in terms of primarily box-oriented aggregates.
Kevin Garnett …
PER difference: 5.64 (substantially up in 2004) 23.80 to 29.44
WS/48 difference: 0.0788 (substantially up in 2004) .1935 to .2723
BPM difference: 3.96 (significantly up in 2004) 6.83 to 10.15
And the on-off (and so far as I’m aware other impact measures) … again 2004 looks significantly better.
Those numbers ... to my reading ... make "basically" the same ... a difficult reading to justify ... something that would require significant evidence. Maybe there is something beyond the numbers (box and otherwise) ... it isn't clear what that might be.

Maybe it genuinely is a close, detailed nuanced analysis comes to a conclusion that Rose, Smith and Daniels are “trash players”. That Joe Smith is a “fantastic starter”, that Nesterovic in ’02 was “a decent 5 man” and “Peeler, Gary Trent [Sr] etc [minutes gap here but next guys in minutes … Sam Mitchell? Felipe Lopez? Loren Woods?]” were “decent role players”. It maybe a close analysis that says Billups was “playing at an all-star caliber level” and that “Wally who actually did make the all-star team” is a fair summary of the standard of play of Szczerbiak that year.

But the working isn’t there. And Joe Smith as “a fantastic starter” whilst Rose and Daniels (both off the bench) are “trash players” … beyond the gulf in perceived quality (noting Joe Smith is not merely fantastic against league average, but amongst only starters) … just the vocab choices ... I’m not sure it comes off as nuanced, other readers’ mileage may vary.

[DraymondGold]

I've argued that the perimeter players shouldn't be forgotten in people's ballots, but I'm interested in hearing more on the bigs.

Have people done any film comparisons/analyses looking at any subset of 01 Shaq/03 Duncan/04 Garnett/23 Jokic?

(statistical and contextual comparisons are also welcome, although plenty have been discussed on those two topics already -- I just haven't seen much film discussion on the bigs yet in this thread). Even just links to past film analyses, or to accessible playoff games for some of those players would be interesting to see (preferably with a reminder for whether this is a good/bad/average game for the player if you're just linking a game).

[70sFan]

I've argued that the perimeter players shouldn't be forgotten in people's ballots, but I'm interested in hearing more on the bigs.

Have people done any film comparisons/analyses looking at any subset of 01 Shaq/03 Duncan/04 Garnett/23 Jokic?

(statistical and contextual comparisons are also welcome, although plenty have been discussed on those two topics already -- I just haven't seen much film discussion on the bigs yet in this thread). Even just links to past film analyses, or to accessible playoff games for some of those players would be interesting to see (preferably with a reminder for whether this is a good/bad/average game for the player if you're just linking a game).

I am finishing my work on Jokic scoring performance in 2023 and 2024 playoffs. I hope I will find time to share my work tomorrow.

[Cavsfansince84]

What doesn't sit that well with me regarding Steph is I feel like he gets credit for being his 2016 self in 2017 just so we can use his 2017 playoffs and the issue I have is that 2016 actually seems like a bit of an outlier season for him. More so, I think he actually was prob a better player in 2021 but just didn't stand out quite as much by then. Had his 2022 playoff run come after his 2021 rs I think it would be pretty clearly viewed as his peak now. Not that I'm low on Steph really because I think I'll have him at 6th for this time period so maybe low relative to some but I'm not really a fan of how 16&17 sort of get blended together quite often. I also am not quite as high on his 2017 playoff run as some seem to be. It seems more like a run to me where Steph just played like his usual self and his team crushes everyone because it's quite likely the goat team and even then got sort of lucky with the Kawhi injury.

[lessthanjake]

What doesn't sit that well with me regarding Steph is I feel like he gets credit for being his 2016 self in 2017 just so we can use his 2017 playoffs and the issue I have is that 2016 actually seems like a bit of an outlier season for him. More so, I think he actually was prob a better player in 2021 but just didn't stand out quite as much by then. Had his 2022 playoff run come after his 2021 rs I think it would be pretty clearly viewed as his peak now. Not that I'm low on Steph really because I think I'll have him at 6th for this time period so maybe low relative to some but I'm not really a fan of how 16&17 sort of get blended together quite often. I also am not quite as high on his 2017 playoff run as some seem to be. It seems more like a run to me where Steph just played like his usual self and his team crushes everyone because it's quite likely the goat team and even then got sort of lucky with the Kawhi injury.

Yeah, as I’ve said in my voting post, Steph didn’t actually combine in one year the 2016 regular season with the 2017 playoffs. In fact, I actually think it could even be argued that his best two regular season and best two playoffs were in four completely different years. It’d just require considering 2016 and either 2015 or 2021 to be his best regular seasons and 2017 and 2022 his best playoffs. This all makes Steph suffer a fair bit in a one-year peak ranking, because his best regular seasons and best playoffs don’t really line up much.

All that said, I actually think Steph’s 2015 is pretty close to 2017, and could arguably be considered Steph’s peak instead. I’d probably actually consider 2015 his second-best regular season, behind 2016 but ahead of 2021 and the Durant years. It gets a bit lost because of the 73-win season that came afterwards and the GOAT 2017 team, but Steph led the 2015 Warriors to a 67-win, 10.01 SRS season. This was one of the best regular seasons a team has ever had. And they didn’t have Durant for this, nor was Draymond as good as he’d be the next year. It was almost certainly Klay’s best year, so that definitely helped, but leading that team to 67 wins and 10+ SRS was a really historic achievement IMO. Meanwhile, Steph was really good in those playoffs. Probably not as good as he was in 2017, but still really good. People kind of ding his playoff performance for the fact that Iguodala got Finals MVP, but that was basically just nonsense, and Steph was enormously impactful in the Finals (and throughout the playoffs). In fact, taking the playoffs as a whole, I’m not sure he was better in the 2022 playoffs than the 2015 playoffs. So yeah, I’m kind of ambivalent between 2015 and 2017 as Steph’s peak, to be honest. I went with 2017 largely because that’s the general consensus, but I’m not certain that that consensus is correct.

[f4p]

What doesn't sit that well with me regarding Steph is I feel like he gets credit for being his 2016 self in 2017 just so we can use his 2017 playoffs and the issue I have is that 2016 actually seems like a bit of an outlier season for him. More so, I think he actually was prob a better player in 2021 but just didn't stand out quite as much by then. Had his 2022 playoff run come after his 2021 rs I think it would be pretty clearly viewed as his peak now. Not that I'm low on Steph really because I think I'll have him at 6th for this time period so maybe low relative to some but I'm not really a fan of how 16&17 sort of get blended together quite often. I also am not quite as high on his 2017 playoff run as some seem to be. It seems more like a run to me where Steph just played like his usual self and his team crushes everyone because it's quite likely the goat team and even then got sort of lucky with the Kawhi injury.

Yeah a lot of people like Steph and want to rank him higher so they just kind of blend things together or handwave other things to get him up higher.

[Doctor MJ]

What doesn't sit that well with me regarding Steph is I feel like he gets credit for being his 2016 self in 2017 just so we can use his 2017 playoffs and the issue I have is that 2016 actually seems like a bit of an outlier season for him. More so, I think he actually was prob a better player in 2021 but just didn't stand out quite as much by then. Had his 2022 playoff run come after his 2021 rs I think it would be pretty clearly viewed as his peak now. Not that I'm low on Steph really because I think I'll have him at 6th for this time period so maybe low relative to some but I'm not really a fan of how 16&17 sort of get blended together quite often. I also am not quite as high on his 2017 playoff run as some seem to be. It seems more like a run to me where Steph just played like his usual self and his team crushes everyone because it's quite likely the goat team and even then got sort of lucky with the Kawhi injury.

Understandable you feel this way.

From my perspective, the focus on single-year peaks is a problematic approach specifically because it drives us to seek out "perfect seasons without blemishes" rather than talking about the basketball specifics. I get why we do it, but when we zoom out to multiple years the little blips that tend to come from getting banged up tend to even out.

[One_and_Done]

Yeh, I disagree that the absence of 1st placed MVP votes in a given year is indicative of that player no longer being considered the best player. Steph was unanimous MVP in 2016, it didn't mean 'the consensus' was that he was better than LeBron. Nor was he. It just meant the voters agreed his narrative was right that year.

I just heard 'advanced stats are very consistent'. Yeh, except when they're not. Absolutely nobody argues we should rate the top 10 guys each year based on who was top 10 in RAPM or whatever. People invoke advanced stats selectively with context, yet that exact context shows why you can't take them too seriously as a way of distinguishing one great player from another.

[Cavsfansince84]

Understandable you feel this way.

From my perspective, the focus on single-year peaks is a problematic approach specifically because it drives us to seek out "perfect seasons without blemishes" rather than talking about the basketball specifics. I get why we do it, but when we zoom out to multiple years the little blips that tend to come from getting banged up tend to even out.

I get this approach as well and to some degree I will use it. It's just when people point to Steph's 2017 postseason run as like an atg run I'm not sure I agree with that either. For me I will probably have 2021 as his peak because I think physically he had become more resilient and also figured out how better to pick his spots as a scorer. Personally if I'm blending two seasons together it would prob be 21&22. I also think a hidden benefit of having KD for Steph was getting to do less in the rs which helped them to get to 5 finals in a row.

[DraymondGold]

Just commenting on these points on 2019
...
Nobody picked a sweep so you can say they outperformed in that sense. I think it's a bit overzealous to go any further than that.

Fair point, I stand corrected.

Against the Rockets (statistically one of the best playoff defenses of the century), a team with an innovative coach again focused entirely on stopping Curry, facing some of the most defensive attention a star has ever faced (literally facing a box and one!)... Curry looked again like one of the GOAT offensive players, again under historic amounts of pressure.

Watch on YouTube

"years from now, or maybe days, it will be cited as a playoff failure by critics, but rarely will you ever see a player influence an offense more than Curry did in game 2 of the finals". Just because they lost doesn't mean Curry wasn't dominant under pressure.

(for the sake of not confusing anybody who's reading this, the quoted should say Raptors and not Rockets in case it's not obvious)

Anyway, I think this was a series where Steph gets most of his criticism for the back end of the series.

He averaged 26.3 PPG in Games 4-6 which is quite fine, but on 25-62 FG (0.403) and 10-34 3P (0.294).

Game 4: GSW leads by 4 at half. Lead crumbled away in a 3rd quarter where the Warriors were outscored by 16 and Steph played all 12 minutes.

Game 5: GSW leads by 6 at half. Steph shoots 3-12 FG and 2-9 3P in the second half - but they do escape with a win thanks to clutch shots from Klay and a timely game tying 3 from Steph.

Game 6: GSW leads by 2 going into the 4th quarter. Steph shoots 1-6 FG and 0-4 3P in the fourth quarter.

I don't think this is some godawful series he needs to get held against him or anything. Like you said, he was facing a historic defense and certainly faced a lot of pressure. But to my eye and I think many eyes, he did not close out the series (G4-6) looking like an offensive GOAT nor was he particularly dominant. These are sort of the situations where to some, it feels like the standard for Steph is remarkably low relative to his billing as a player.

Well, I think it's easy to over index on the results, and forget the relative difference in the teams' talent and depth, and just how much defensive attention Curry was garnering. The Raptors were seriously more talented than the Warriors, once you factor in Health. They were a historic defense, again giving perhaps the most defensive attention a perimeter player has ever received late in the playoffs.

They played a box and one (and a triangle and two) against someone -- it's hard to emphasize just how crazy that is now that people are more familiar with weird zones (e.g. those Heat zones that Spoelstra pulled out in the playoffs recently), but one should expect the box performance of a star player to go down when facing a box and one. This opens up opportunities for teammates, but requires teammates to be even playoff starter caliber... which many of Curry's teammates were not by the 2019 finals (more on that in a sec).

Take this film analysis, for example, of the end of the 4th quarter of the 2019 finals:

Watch on YouTube

With 3:30 left in the 4th and behind by 4, the Warriors run a series of offensive players. They all center around Curry, several work, and the Warriors end up winning. Explicitly:
(1) good off-ball movement by Curry to generate an open shot,
(2) Curry double team creating an open 3 for Klay (which he makes),
(3) more good off-ball movement by Curry to generate an open shot (which he makes).
The video also shows how these plays which centered around Curry worked at other points in game 5 and in other games in the series. This doesn't seem like much of a decline as the series went on and the pressure ramped up.

People have the impression that the Warriors were a deep dynasty with lots of firepower behind Curry (true most of the time), but by the time we get to the 2019 finals it's pretty false. The Warriors had no business being in that series given their health -- it's genuinely great floor raising from Curry to bring the series to 6.

Among their top players:
+ Steph Curry: played and was reasonably healthy (Broken / dislocated finger, but not a major injury and played through it)
+ Draymond Green: played and was reasonably healthy (although by this point he had lost his shot entirely, shooting 23% from 3 in the playoffs
x Klay Thompson: Hamstring strain, missed the end of game 2, game 3, and the end of Game 6. The Warriors were 2–2 in games Klay played the whole time with 3 away games. Klay then played Game 4 and Game 5 and the rest of Game 6 injured.
x Kevin Durant: played a total of 12 minutes in game 5 of the finals, and missed Games 1-4/6 entirely
x Andre Iguodala: Injured his Hamstring in the Western Conference finals, then reinjured his leg and missed the end of Game 2. Games 3–6 were then played injured
x Kevon Looney: Sprained and fractured Collarbone. He was injured in game 2, missed Game 3 along with Klay, then played injury in Games 4–6. At times he wasn't able to lift his arms over his shoulders.
x DeMarcus Cousins: Torn Quad injury. He missed most of the playoffs, then came back early to try to play through injury, but was clearly not fully healthy, and was a walking liability defensively. He didn't play a single game the following year due to injury.

So... 5/7 of their top players were injured, with a combined 7 games being fully missed by their top players, a combined 12 games being missed at least partially, and more games being played through injury.
Compare this to the Raptors top 7, who didn't miss a single game in the playoffs. Kawhi was getting banged up like Curry, but was still able to play, and after that all of the Raptors were clearly significantly healthier and less worn down than their counterparts on the Warriors.

The Warriors' depth beyond that was
-a 33 year old Shaun Livingston (who never played another NBA game after this),
-Quinn Cook (-2.49 RAPM in 2019; and was out of the league in 2 years),
-Alfonzo McKinnie (-0.79 RAPM in 2019, per NBA Shot Charts, and was out of the league in 3 years)
-34 year old Andrew Bogut (who never played another NBA game after this),
-Jonas Jerebko (who never played another NBA game after this)

... which is to say some of the worst depth of a finals team of the decade, accounting for health. Compare that to the Raptors who had very positive depth (17th in Sansterre's Top 100 teams ever in his Depth rating).

The quality of his teammates becomes particularly important when facing historic defenses and innovative defensive schemes designed explicitly to stop one player. When Curry is doubled (e.g. in the video above) or facing a box and one, this naturally creates openings for teammates. When teammates are either out of the game entirely, hobbled enough to have lost their bounce, or players who aren't healthy enough or good enough to play in the league next year... that's not exactly a great group of players to make an open shot in the NBA finals, nor are they good enough to disincentivize the Ratpors from throwing a full array of zones and player-specific schemes explicitly to stop Curry.

This was pretty clearly the worse depth of any Warriors dynasty year. I'd think playing the NBA champions with one of the best playoff defenses of the century to 6 games would be considered a good achievement. 2015 LeBron playing well with an injured team against great defense isn't a compelling reason for why he shouldn't be in contention for an all-time peak two years earlier in 2013... so I don't think 2019 Curry playing well with an injured team against a great defense is a compelling reason for why he shouldn't be in contention for an all-time peak two years earlier in 2017.

What doesn't sit that well with me regarding Steph is I feel like he gets credit for being his 2016 self in 2017 just so we can use his 2017 playoffs and the issue I have is that 2016 actually seems like a bit of an outlier season for him. More so, I think he actually was prob a better player in 2021 but just didn't stand out quite as much by then. Had his 2022 playoff run come after his 2021 rs I think it would be pretty clearly viewed as his peak now. Not that I'm low on Steph really because I think I'll have him at 6th for this time period so maybe low relative to some but I'm not really a fan of how 16&17 sort of get blended together quite often. I also am not quite as high on his 2017 playoff run as some seem to be. It seems more like a run to me where Steph just played like his usual self and his team crushes everyone because it's quite likely the goat team and even then got sort of lucky with the Kawhi injury.

Yeah, as I’ve said in my voting post, Steph didn’t actually combine in one year the 2016 regular season with the 2017 playoffs. In fact, I actually think it could even be argued that his best two regular season and best two playoffs were in four completely different years. It’d just require considering 2016 and either 2015 or 2021 to be his best regular seasons and 2017 and 2022 his best playoffs. This all makes Steph suffer a fair bit in a one-year peak ranking, because his best regular seasons and best playoffs don’t really line up much.

All that said, I actually think Steph’s 2015 is pretty close to 2017, and could arguably be considered Steph’s peak instead. I’d probably actually consider 2015 his second-best regular season, behind 2016 but ahead of 2021 and the Durant years. It gets a bit lost because of the 73-win season that came afterwards and the GOAT 2017 team, but Steph led the 2015 Warriors to a 67-win, 10.01 SRS season. This was one of the best regular seasons a team has ever had. And they didn’t have Durant for this, nor was Draymond as good as he’d be the next year. It was almost certainly Klay’s best year, so that definitely helped, but leading that team to 67 wins and 10+ SRS was a really historic achievement IMO. Meanwhile, Steph was really good in those playoffs. Probably not as good as he was in 2017, but still really good. People kind of ding his playoff performance for the fact that Iguodala got Finals MVP, but that was basically just nonsense, and Steph was enormously impactful in the Finals (and throughout the playoffs). In fact, taking the playoffs as a whole, I’m not sure he was better in the 2022 playoffs than the 2015 playoffs. So yeah, I’m kind of ambivalent between 2015 and 2017 as Steph’s peak, to be honest. I went with 2017 largely because that’s the general consensus, but I’m not certain that that consensus is correct.

Agreed, 2017 was a slight down from 2016 in terms of production/value. But I think it's important to remember:
-that we'd expect a player to decrease their production when joined by an MVP high-production offensive star
-that we'd expect a player to have an adjustment period when joined by an MVP high-production offensive star
-that we'd expect a player to not go full motor every night of the regular season, the year after their team lost in the finals due to the injuries/wear of going all-out every night
-that 2016 is in conversation for the GOAT regular season run... so Curry has a long way to fall before being worse than some of the competition here
-that 2017 Curry was still *significantly* more impactful in the regular season than some of the other bigs here (Duncan and Shaq specifically don't seem as impactful in the regular seasons in this era)
-that the 2017 regular season Warriors were still *significantly* more impactful in the regular season than the competition and was still comparable to 2016 (they were even better by SRS, although they didn't win the same number of games)

Comparing a player to other versions of himself is definitely helpful for seeing their own evolution, but just because a player had a slight regular season decline compared to their younger selves, that doesn't mean they're necessarily worse than a different player entirely. In most of the stats we have, 2017 RS Curry really doesn't look like too much of a drop off from 2016 RS Curry, and does still pretty consistently look better than some of the other competing bigs (e.g. Duncan, Shaq) in the regular season
(Usual qualifiers: no stat is perfect, any impact stat has an uncertainty range, and all these stats are noisy enough that we wouldn't expect a single peak to dominate even if it was best, etc.)

Regular Season Plus minus (per 48):
- 2017 Curry 18.46 > 2016 Curry 18.17 > 2015 Curry 16.9 >> 23 Jokic 13.22 > 04 Garnett 9.11 > 03 Duncan 8.37 > 01 Shaq 7.75

Regular Season On-off (per 48):
- 23 Jokic 24.04 > 2016 Curry 23.44 > 2017 Curry 20.68 > 2015 Curry 20.21 > 04 Garnett 20.1 > 01 Shaq 5.8 > 03 Duncan 5.9

Augmented Plus Minus
- 2016 Curry 7.4 > 23 Jokic 7.3 > 04 Garnett 6.9 > 2017 Curry 6.6 > 2015 Curry 6.5 > 03 Duncan 5.9 > 01 Shaq 5.8

Adjusted Plus Minus (un-regularized APM, and not one of the many versions of RAPM)
- 2017 Curry 8.5 > 2015 Curry 7.0 > 03 Duncan 6.8 > 2016 Curry 6.6 > 04 Garnett 6.4 > 23 Jokic 6.2 > 01 Shaq 5.6

EPM (best available all-in-one hybrid stat)
2016 Curry 10.5 > 2015 Curry 8.9 > 2017 Curry 8.2 > 23 Jokic 7.9 > 04 Garnett 6.4 > 03 Duncan > 03 Shaq (no data before 2002)

Taking a wider lens, looking at multi-year runs to get stability in these stats, we still see Curry looking similarly dominant over the competition. Curry doesn't need some hybrid 2016 RS + 2017 PS to look more impactful than post-2000 Shaq... he already straight up does look more impactful in 1/2/3/4/5 year runs.

[Owly]

I just heard 'advanced stats are very consistent'. Yeh, except when they're not. Absolutely nobody argues we should rate the top 10 guys each year based on who was top 10 in RAPM or whatever. People invoke advanced stats selectively with context, yet that exact context shows why you can't take them too seriously as a way of distinguishing one great player from another.

So I would guess this is a reference to my post without a quote.

At the margins I will grant that my phrasing was somewhat unclear ... that part is (as is developed as the topic within the paragraph) with regard to box-metrics and refers to methodology, to input-process-output. The same stats in the same league context will get the same math done and the same output. If the output changed it's because the input changed. I don't know that that can be applied to +/- as that's just scorekeeping ... but that's same process same for everyone and ditto on-off. As noted more sophisticated impact models get more opaque and it would be tougher to claim.

It's not output consistency. And in impact stuff this is noted with regard to the need for larger samples (paragraph 6).

And it should be clear it's not about player output numbers necessarily stay the same over the years ... because it explicitly highlights an instance where this is not the case primarily looking on the box-side (for Garnett '02 compared to '04) ... and that's the more stable side.

[One_and_Done]

I just heard 'advanced stats are very consistent'. Yeh, except when they're not. Absolutely nobody argues we should rate the top 10 guys each year based on who was top 10 in RAPM or whatever. People invoke advanced stats selectively with context, yet that exact context shows why you can't take them too seriously as a way of distinguishing one great player from another.

So I would guess this is a reference to my post without a quote.

At the margins I will grant that my phrasing was somewhat unclear ... that part is (as is developed as the topic within the paragraph) with regard to box-metrics and refers to methodology, to input-process-output. The same stats in the same league context will get the same math done and the same output. If the output changed it's because the input changed. I don't know that that can be applied to +/- as that's just scorekeeping ... but that's same process same for everyone and ditto on-off. As noted more sophisticated impact models get more opaque and it would be tougher to claim.

It's not output consistency. And in impact stuff this is noted with regard to the need for larger samples (paragraph 6).

And it should be clear it's not about player output numbers necessarily stay the same over the years ... because it explicitly highlights an instance where this is not the case primarily looking on the box-side (for Garnett '02 compared to '04) ... and that's the more stable side.

You can also quantify the number of times you scored while the temperature was 30 degrees outside, or while the moon was in the transit of Venus, I'm just not sure it's telling us anything reliable.

[lessthanjake]

I just heard 'advanced stats are very consistent'. Yeh, except when they're not. Absolutely nobody argues we should rate the top 10 guys each year based on who was top 10 in RAPM or whatever. People invoke advanced stats selectively with context, yet that exact context shows why you can't take them too seriously as a way of distinguishing one great player from another.

So I would guess this is a reference to my post without a quote.

At the margins I will grant that my phrasing was somewhat unclear ... that part is (as is developed as the topic within the paragraph) with regard to box-metrics and refers to methodology, to input-process-output. The same stats in the same league context will get the same math done and the same output. If the output changed it's because the input changed. I don't know that that can be applied to +/- as that's just scorekeeping ... but that's same process same for everyone and ditto on-off. As noted more sophisticated impact models get more opaque and it would be tougher to claim.

It's not output consistency. And in impact stuff this is noted with regard to the need for larger samples (paragraph 6).

And it should be clear it's not about player output numbers necessarily stay the same over the years ... because it explicitly highlights an instance where this is not the case primarily looking on the box-side (for Garnett '02 compared to '04) ... and that's the more stable side.

You can also quantify the number of times you scored while the temperature was 30 degrees outside, or while the moon was in the transit of Venus, I'm just not sure it's telling us anything reliable.

Does the number of times someone scored while the moon was in the transit of Venus correlate quite well with RAPM? If not, then it’s certainly not meaningfully comparable to something like BPM.

I know you’ll say that you don’t like RAPM either. At a certain point, though, “I don’t trust advanced box data or RAPM-based data” basically just ends up amounting to “My approach is to decide who I think was the best player in a given time period, and then I will cobble together whatever qualitative narrative I can to support that player in any argument regarding that time period.” Which I guess is fine and not super different from what the general layperson’s approach is to these discussions, but it does basically amount to starting at a conclusion and working your way there while explicitly refusing to consider hard data that might disagree.

[One_and_Done]

So I would guess this is a reference to my post without a quote.

At the margins I will grant that my phrasing was somewhat unclear ... that part is (as is developed as the topic within the paragraph) with regard to box-metrics and refers to methodology, to input-process-output. The same stats in the same league context will get the same math done and the same output. If the output changed it's because the input changed. I don't know that that can be applied to +/- as that's just scorekeeping ... but that's same process same for everyone and ditto on-off. As noted more sophisticated impact models get more opaque and it would be tougher to claim.

It's not output consistency. And in impact stuff this is noted with regard to the need for larger samples (paragraph 6).

And it should be clear it's not about player output numbers necessarily stay the same over the years ... because it explicitly highlights an instance where this is not the case primarily looking on the box-side (for Garnett '02 compared to '04) ... and that's the more stable side.

You can also quantify the number of times you scored while the temperature was 30 degrees outside, or while the moon was in the transit of Venus, I'm just not sure it's telling us anything reliable.

Does the number of times someone scored while the moon was in the transit of Venus correlate quite well with RAPM?

Obviously I’m not making a literal comparison, but more extreme examples are often useful for illustrating a point by showing the logical flaw in what is being advanced. Nobody thinks we should rank the top 10 players based on who had the top 10 in RAPM, or VORP, or whatever. Everyone knows these stats are often wrong/misleading. Yet many of the arguments being advanced are just walls of advanced stats. I would suggest that the actual logical argument should be 90-95% of a post, with references to advanced stats that might indicate the arguments are true being a footnote at the bottom, rather than presented as the arguments themselves.

Explaining anything requires a lot of context, and no context is not just more advanced stats. Let’s take the example of a player who “hypothetically” carried a team of scrubs. Just by virtue of the fact that 4 other guys are always sharing the court with a guy who is able to carry them and make the team look good, is going to warp the on-off numbers completely. Then imagine the guy is playing over 40 minutes a game, and the minutes where he’s off are often minutes where the coach had thrown the towel in. Add random variance and small sample size theatre, and the plus-minus numbers for that team won’t mean much. Similarly, you’re going to get more win-shares on a team that wins more. I could keep going.

Advanced stats might be one data point that hints at something, but they also might not be. I think they are getting much more emphasis than they should, because many people find them fun and prefer something “quantifiable” to stuff that can’t be reduced to a formula. Unfortunately there is no formula that can accurately quantify how good a player is. You need lots of context. I say that as someone who hates arguments based on intangible stuff like “killer instinct” or “leadership”, or that says you “just needed to watch the games to know”.

Nobody is saying you should start with your conclusion and work backwards, obviously we don't want that. Opinions should be contestable and reconsidered when new facts emerge. That said, extraordinary claims should also require extraordinary proof. When one guy is rated as the consensus best player at a certain point in time, and there's a number of things that seem to indicate that was logically sound, there needs to be more than a wall of math that is not a reliable indicator of value.

To take just one example I referred to above. Joe Smith did not have big stats. Joe Smith was viewed as a failure because he was a #1 pick who disappointed, and is remembered more for a contract scandal than anything else. But Joe Smith was also a very good role player. Alot of the indicators of this are not math based. Different coaches on good teams with different coaching philosophies kept giving Smith minutes. Different teams, many who were competent, kept trying to sign him for good money. Joe Smith had a good skillset. Also I watched Joe Smith, and came to the same conclusion. I have no idea what his VORP was, but if it said he was bad I wouldn't care in the slightest.

[Cavsfansince84]

My question with Steph is also to what degree can we separate his impact from Draymond's? I know this is something that's been discussed many times before but I would find it helpful if someone could say to what degree we actually can with ramp and +/- because not that it's a huge knock on Steph but it's something that should be addressed imo.

[DraymondGold]

I suppose to help prompt more big man discussion like I asked above, how do people feel about Duncan vs KG vs Jokic from a team-building perspective?

Duncan and KG are obviously both some of the best defensive players ever, so presumably easy to build around defensively.

With Duncan, he’s of course able to hold up an offense with his scoring, but the offensive and overall team results aren’t that dominant in an all-time sense (the 03 Spurs were good but not all time good). And notably he played a style that would start to become outdated in just 2 years. I definitely give a lot of intangibles credit to Duncan for being a good teammate and being willing to take a back seat in the offense as his teammates improved, and as they shifted to more Manu ball in 05+. But at the same time, even if he showed an unselfishness that was better than continuing to play selfishly, the actual impact Duncan had offensively and overall clearly took a dip as his role in the offense shifted in a smarter and more effective offensive scheme. This makes it a little unclear to me how it would look if his peak age 26 (2003) season had come just two seasons later, or on a different and more offensively slanted team… would he still have such lift in the different role?

With KG, he seems like a dream to build around as an option 1 on defense, option 2/1b offense (the kind of role Duncan played towards the mid/late 2000s, and closer to what he played in Boston). At the same time, he lacked just enough scoring chops to be the kind of floor raiser Duncan was in 03, and people do have the playoff resilience concerns… albeit we really don’t have as much playoff data as we’d like in the middle of his prime.

With Jokic, his Offense is obviously pretty ideal from a team building perspective. Fantastic floor raiser, but also highly scalable - able to play 2 or 3 man game with his passing and scoring, able to play off ball if a teammate needs the ball more as a screener or pick and pop player of offensive rebounder, etc. He’s probably the most versatile of the offensive GOATs (which isn’t to say there aren’t arguments for other offensive players in this thread… I tend to think Jokic doesn’t have quite as game-breaking of a superpower as Curry’s on/off-ball shooting and LeBron’s driving game, and the lack of handle / occasional discomfort volume shooting late in the playoffs has come to bite him a little bit).

But with Jokic, I have concerns for the defense, as I alluded to in the other post. That’s not to say I see him as a negative defender. No, he has great rebounding, great hands, doesn’t foul, has shown he’s able to hedge and recover and avoid mismatch hunting fairly well these recent years (after some defensive improvement and surrounded by better defensive teammates). But I do wonder if he puts a bit of a cap on how good your defense can be. He occupies the most important defensive position, and he doesn’t provide much vertical room protection, or horizontal defense, or even as much back communication as some of the most active defenders. I wonder how good a defense could be, as it would be pretty hard to put another big out there with Jokić — viable twin-towers lineups are hard to come by these days, given the opponent shooting and mismatch hunting.

Which people feel more comfortable building a championship level roster? Dynasty level roster? Who’s more scalable? Of course, these questions require a bit of projection, and it can be hard to not be biased by the actual events (they really did build a Dynasty around Duncan, even if they were at their most dominant team-wise in 99/14 when they were playing the least amount of Duncan-ball)… but I’d love to hear people’s thoughts!

[DraymondGold]

My question with Steph is also to what degree can we separate his impact from Draymond's? I know this is something that's been discussed many times before but I would find it helpful if someone could say to what degree we actually can with ramp and +/- because not that it's a huge knock on Steph but it's something that should be addressed imo.

Just to address this quickly (though I’m a bit tired of talking just Curry), they definitely both benefit from each other. But they also spend a lot of time together, so pure impact metrics might have trouble separating them in single-season sample sizes. So to split them up, you really need longer samples (think 3-5+ years) and regular season data to get enough diversity of lineups and opponents and larger sample sizes (full season data is fine too).

You probably saw this post — https://forums.realgm.com/boards/viewtopic.php?p=119539800#p119539800 — which splits out the Warriors’ ratings with different stars, where Curry looks all-time/GOAT tier in his impact even without Draymond.

In 2015-2019, Warriors had a +12.84 Net Rating with Curry but without Draymond, while they had a -4.46 Net rating without either, for a +16.3 Net rating swing. This doesn’t correct for the other lineups, but it’s still pretty fantastic.

Curry usually looks the best in multi-year WOWY measures (Curry’s GOAT-tier in WOWY, Draymond’s not) and in multi-year ~several year RAPM, though it depends on the version. E.g. in Vanilla RAPM, 15-19 Curry’s +5.89 (1st in league, 43% better than Dray who’s +4.11).

Of course, I’ve argued that the ability to be synergistic and fit well with your best teammates is a skill, and a really valuable one. Curry’s chemistry with Draymond is one of the things that made the Warriors so dominant, just like Duncan with Manu, or Shaq with Kobe, or Jokic with Murray. It is important to isolate the superstar’s impact, but it’s also important to see how well they can build chemistry with their best costars, as that chemistry is important if you want to win championships!

[One_and_Done]

Actually the 99 Spurs were much more Duncan centric compared to 05 or 07, as I've discussed before.
viewtopic.php?t=2367353

[tsherkin]

Hmm.

My early thoughts for top-4 would be 09 Lebron, 00 Shaq, 23 Jokic and some version of Duncan, 02 or 03. I need a minute to think and I'm mid-set, so I don't have time for a thoughtful post, but I'll drop something over the weekend.