lessthanjake wrote:Doctor MJ wrote:lessthanjake wrote:
These are good responses, and I’ll just respond to several things.
Regarding the above, I think 3% TS% is pretty significant. Garnett was taking about 24 true shot attempts a game in those playoffs. If he had a 3% higher TS%, he would’ve scored about 1.5 extra points a game. Which, roughly speaking, amounts to about 1.5 extra points of a impact per game (missed shots can be offensive rebounded and the team can score afterwards, which somewhat mitigates the effect of a lower TS%, but at the same time, making shots makes it easier on the next defensive possession, which exacerbates the effect of missing shots, so we can probably roughly say those things cancel out). From the perspective of player impact, I actually do think that that amounts to a tier of impact. And I think it certainly does if we add in the extra turnovers, as well as the lower per-possession output.
So, I certainly don't want to argue that an X% change in TS% over a volume Y doesn't translate to Z points on the scoreboard. That's certainly the arithmetic of it.
Consider though:
a) In addition to the presumably tougher playoff defense making it very possible that an X% change is what we should expect in the situation, there's also the natural variance of things. Over the splits of KG"s '03-04 RS, we see his TS% vary between 49.8 to 58.8. If a player's shooting efficiency varies + or - 4-5%, then why are we looking to draw a specific conclusion of something changing in the playoffs when he isn't even outside that RS range?
To put another way: If a difference in performance can be explained with normal variance patterns, while that doesn't mean that the bounce of the ball with good/bad luck isn't making a critical difference which can be linked to a players accomplishment, it does mean we should be cautious about insisting on a narrative explaining that normal variance as if we're in a fable requiring a moral takeaway.
I think this goes to a difference in how we’re thinking about this discussion conceptually. So let me just try to clarify a bit:
I’m not saying that Garnett having a lower TS% in the playoffs that year means he was systematically a worse playoff player than regular season player. In other words, I’m not arguing that if Garnett had played a large sample of playoff games that year that he’d have been worse in those games than he was in the regular season. I don’t know if that’s the case or not. What I *am* arguing is that, in the small sample of the actual playoff games he did play, he was not as good as he was overall in the regular season. I’m sure if we cut up portions of his regular season, we could find similar spans where he was no better (and perhaps actually worse) than he was in the playoffs. Which is to say that I’m not saying that I think Garnett performing less well in the playoffs couldn’t just be variance. But the thing is that “it’s just variance” only really is a counterpoint if the argument being responded to is taking a small sample of games to extrapolate how good a player was in general. That’s not what I’m doing though. I’m not extrapolating from the playoffs how good a basketball player Garnett was in general in 2004. What I’m doing is simply discussing how good he was in that small sample of games. Whether it’s a product of variance or not, how good he was in those games is how good he was in those games. And that has independent importance without any extrapolation, because the small sample of the playoffs are tremendously important.
The upshot is that I think there’s a lot of instances in which I could say I think Player A was a better player in general than Player B, but Player B would rank higher than Player A in an assessment of the “greatness” of their year, because variance (or potentially other factors) worked in Player B’s favor. I’m not opposed to the idea that that may be what’s going on with 2006 Wade and 2004 Garnett. Maybe if they played a large sample of playoff games, Garnett would’ve been better than Wade over the course of those games. It’s certainly plausible to me, given that I think Garnett was more impactful in the regular season. But they didn’t play a large sample of playoff games. They played a small sample, and Wade was significantly better in those games than Garnett was. And since those small samples of games are tremendously important, Wade being significantly better in the games that matter the most is enough for me to say his year was greater.
I appreciate you looking to figure out where we diverge. To me that's often the most valuable part of dialoguing through disagreements.
At the broadest level, what we're talking about is goodness in theory vs achievement in practice, and both of those things are worth discussing and ranking, so it's really a question then of what we're each using here, and if there's any concerns with our process given our criteria.
When I read your statement here, my main quibble is in tying the word "good" to extremely granular events.
It would appear to me that as you describe it, a player is playing "good" whenever his shot goes in, and not good when it doesn't, despite the fact that it's literally the same guy, oftentimes in back-to-back possessions, literally doing the same stuff with the same capabilities. I would argue this is a meaningless definition for players & coaches, as they would think about playing good in terms of process.
By that same token, if you're watching a game or series and you can see one team winning based on the shots that actually go in, but the other one with the better process based on what will on average go in, the smart money is on the losing team catching up.
Now, sometimes the better process loses because of luck in the small sample, but that doesn't mean that that team should change their approach or that the winning team should think they won for reasons of skill, and so when I see a scaled-up version of reasoning that effectively ignores luck by series' end, I see something that isn't actually tied to what the players actually were, but rather a player assessment that happily includes luck.
To put another way: I see result-justification rather than process-analysis, which is another form of winning bias.
As always, I'm not alleging winning bias is something only other people experience, I see myself as prone to it as anyone else generally, and if I'm doing a better job of avoiding it than the average bear, it's only because I'm more pre-occupied with rooting it out of my process at this time due to al the times I've seen myself fall prey to it.
lessthanjake wrote:b) Let's say KG being worse was costing them 1.5 points in every game. How many games in the playoffs would 1.5 point swing from a loss to a win for the 2004 Timberwolves?
The answer is zero. The Wolves never lost a game by less than 6 in that run.
This then to say that while over a large enough sample a 1.5 point loss in impact will swing games, in the course of an NBA playoff run, it oftentimes wouldn't.
And if we're actually interested in how KG looked in the closest games of those playoffs:
Game 4, MIN at DEN, W 84-82: Garnett 27 pts on 57.2% TS
Game 3, MIN at SAC, W 114-113: Garnett 30 pts on 55.6% TS
Game 7, SAC at MIN, W 83-80: Garnett 32 pts on 57.5% TS
Game 5, LAL at MIN, W 98-96: Garnett 30 pts on 53.9% TS
We note that Garnett was always scoring above normal volume, and doing so with higher efficiency than the average playoff game.
Now I'm not looking to say that these 4 games prove anything about Garnett's peak - super-small sample of course - but just from the perspective of X% dip in efficiency must be treated as damage that held his team back, I really don't think the evidence actually bears that out.
Yes the team would be better if Garnett just made more of his shots, but that doesn't mean we should be quick to downgrade a player's playoff performance relative to his regular season estimation as if there's a clear meaningful failure at play throughout the post-season span.
So I’d say a couple things about this:
1. While I’m sure I’ve made similar points to this in the past so I get what you’re saying, I don’t think it really makes a lot of sense to basically say “This player’s impact went down by X points per game, and since every loss was by more than X points, it didn’t matter.” For one thing, there’s a real butterfly effect here, where we just don’t know what would happen if things were changed. Perhaps even more importantly, someone’s impact going up by X points per game doesn’t mean it goes up by X points every game. There’s a lot of variance to that. You note that his TS% was not bad in the closest games, but if he’d had a higher TS% in general, he may well have had an even better TS% in those specific games—perhaps even by a lot. For instance, maybe his higher TS% would’ve manifested itself by being similar in every game except for being a lot higher in a couple games, in which case it very well could’ve changed results.
2. Even if we posited that no results would’ve changed if he’d had a higher TS%, that doesn’t mean that Garnett wouldn’t have played better in that scenario than he did in reality. There’s a world where we could ramp up Garnett’s playoff performance enough that I’d put him above 2006 Wade, even if the Timberwolves still went out in the conference finals. But, for me at least, that’s not the world we have.
On 1: It does make sense to push back against the idea that we should ever dismiss statistical difference in a stat based on the team result being bigger than that effect. I bring it up to emphasize what a tiny sample of a tiny stat you're focusing on here. There are a million other things going on in the game that are going to swing impact this way or that, but rather than looking to deal with them all either a) holistically or b) with detailed basketball conversation, you're appearing to latch onto one stat and basically treating it like the player is playing that much better or worse based on what the stat says.
I know that you're a pretty knowledgeable dude and that you're not looking to simply use TS% as the only thing that matters, but because you have it as a clear stat to look at, you're giving it preeminence in the conversation above all the other stuff you can't quantify easily, and this is very much a known danger (tyranny of the quantifiable) wherein we over-index on whatever is easiest to measure.
Re: if he'd have had a higher TS% in general, he may well have had an even better TS% in those games. Let's take care to note the causality here:
Your talking about his average TS% as if it is causing his TS% in specific games, and so the fact that TS% wasn't a problem when it mattered is irrelevant is irrelevant, because if he'd been good-ing harder generally, then we'd expect his TS% in those games would be that much better.
My pithy language there might sound insulting and I apologize for that, but the idea that we must knock a player for a general TS% drop in the playoffs regardless of a) better opponents, b) noise, c) tiny effect relative to actual game scores, d) disappearance of the issues in the key games, as playing less good, just doesn't seem like a recipe for zeroing in on what they player was actually doing out there.
Like, if there were infinite universes, each differing based on effective luck, I'd expect that jake in those other universes would be arguing very different things if he used this reasoning, whereas what I'm trying to do is essentially find an approach that wouldn't lead to me disagreeing with other universe selves.
lessthanjake wrote:lessthanjake wrote:
I got Garnett’s playoff rTS% from the Thinking Basketball website. I’m virtually certain that they compare it to the opponent’s defense’s regular season TS%, rather than any league average.
Okay, that makes sense.
I would point out that generally for any given team they're going to be trying considerably harder on D in the PS than in the RS, and if you don't factor that in in 2004, you'd probably end up concluding that the PS offenses all forgot how to play at the same time, which I would not say is what was going on.
lessthanjake wrote:
The Timberwolves had a -1.2 rORTG in the playoffs, with the rORTG being calculated relative to the opponent’s regular season DRTG. Granted, I wouldn’t say playoff rORTG is a flawless measure (I talked about that some in this thread, with regards to Jokic), but I do think the Timberwolves being a subpar offense in the 2004 playoffs is consistent with what we might expect given Garnett’s offensive struggles.
I'll just reiterate that if you do this for everyone else in the 2004 playoffs, you're going to see similar stuff, and looking to specifically make "Garnett has playoff issues" but not applying that to the whole league doesn't make sense to me.
I’m not sure it’s true that we would expect lower TS% in general in the playoffs. Players do try harder on defense in the playoffs, but players also try harder on offense too. I recall looking at this in general in the past, and I found that regular season TS% and playoff TS% actually tend to be very similar. For instance, while it happens to have been a bit lower in the playoffs that year, it was also higher in the playoffs in the two surrounding years. And, considering that playoff defenses tend to be a lot better than average in the regular season, that means that rTS% tends to be higher in the playoffs. Of course, that isn’t super surprising, because the teams in the playoffs tend to be better offensively. But all of this is to say that I don’t think the intuition that offensive efficiency is lower in the playoffs is actually right, and I think it’s definitely not right when the number I’m keying in on is rTS%.
I don't think players actually coast on offense like they do on defense as a rule, because that's where the glory is.
I'd be interested to see a detailed rTS% perspective on stars to be clear, but over sufficient sample, I'd expect that NBA players underperform by rTS% in the playoffs.
lessthanjake wrote:lessthanjake wrote:
A couple responses to that:
1. I think there are real differences in how good players are in different seasons and different playoffs in different years, even within their primes. Some of that is just matchup related in the playoffs, and a lot of it is also health. For instance, just because a guy plays in two playoffs doesn’t mean his body was doing equally well both times. There’s also mental factors. People aren’t always as checked in at all times (which I think was probably the case with 2004 Kobe). That said, I do agree that sometimes things are just random.
2. Even if we say Kobe was in his prime and maybe just randomly didn’t have as good of results in the 2004 playoffs as he usually did, I think you’re comparing to Kobe in that particular year and then drawing a conclusion that compares to prime Kobe in general. If he played worse in the 2004 playoffs than he normally did, then there’s ample room for someone to have put up numbers as good as 2004 Kobe specifically while still being worse than prime Kobe in general. To take a fairly extreme example of this to illustrate my point, I wouldn’t say that since Dirk was better in the playoffs than 2011 LeBron, and since LeBron was in his prime in 2011 then we should conclude that 2011 Dirk was better than prime LeBron in general. In any event, I’m not sure it really matters much for purposes of this discussion, since Kobe’s rTS% in that series was better than Garnett’s.
So, where we appear to be diverging is in our perspective on season-to-season variance. Is a player having a relatively down season in the midst of his prime actually not playing like his prime self? or, is he basically the same guy but a confluence of factors is giving him a less impressive footprint?
With Kobe in '03-04 what we can definitely say is that he reduced his shooting volume as part of the team's incorporation of Malone & Payton, and given that a) Kobe's best years are always associated with high volume scoring, and b) the team didn't win the chip and frankly got embarrassed in the finals, it makes sense why none of us are going to talk about that season as if it was his best.
But Kobe was absolutely prime Kobe that year and it doesn't make sense to look at that entire season as if Kobe was, say, injured. What you got from him that year is what is basically what you should expect from any prime version of Kobe placed on a team with Shaq/Malone/Payton and Jackson as coach.
Now to be clear, if I was looking to favor someone else by scoring volume as if that represented a ceiling of what they were capable of, it would be wrong to judge Kobe based on '03-04, but his efficiency was perfectly normal that year by his own standards, and so treating it like "not really prime Kobe" just doesn't make sense to me.
Also to be clear, '03-04 was an anomalous year for Kobe also because he was constantly flying back & forth to Colorado for trial, and you might argue that this made him play worse on the court... but this was the opposite of the narrative. The narrative was about how incredible it was that he could handle that disruption without it seeming to affect his play at all.
k, just summing up here with a broad point:
On the PC board we all know there are various types of threads. One type of thread that I mostly avoid is of the time "How many seasons was X better than Y in the year 2001?". Why? Because I think threads like this effectively encourage people to magnify narrative nitpicks. You'll get stuff like:
1. X in '03 - perfect
2. Y in '01
4. X in '02 - he made that one mistake and his team got eliminated
5. X in '04 - he made that other mistake and his teammates saved him
X was the basically same guy for 3 years, but now we're not just ranking those years, but finding a place for a wedge between seasons based on another player based on little bits of randomness like this, implying that one bad day out of 365 is the difference between better or worse than Y. And while this is literally possible from a value-add perspective, it's quite unlikely that some small moment is determining who contributed more throughout the year.
Okay, so I think my view is that you make a valid point but that you may be taking it a bit further towards the land of homogenization than I think is warranted.
Basically, I think you’re right that we sometimes distinguish between how good a player was in very similar years, on the basis of very small moments that can be pretty random. Such moments may make a player’s year “greater” IMO, because those moments can make a year more memorable or significant. But they don’t necessarily mean the player was better in general. I think you’re right about that, and it’s a good point to make, particularly in a peaks project.
However, I think it’s a bit overly simplistic to essentially define what we think were a player’s prime years and then say differences in performances between those years must have just been randomness. People do actually systematically perform better at things in different timespans than others. There’s just a lot of factors that go into how well people perform at things, including a whole bunch of outside-the-court stuff that affects their focus and whatnot. I’ve certainly not been a professional sports player, but I have done things competitively over the span of years and I definitely think there’s been real ebbs and flows in just how good I’ve been at those things. I assume the same is true of NBA players.
Which is to say that not all prime years are equal, but I agree we should be careful to reflexively conclude that a player was better in one year than another just because of a random moment or two. On the latter point, I get that there’s some tension between that and me basically saying I care about what actually happened in the playoffs regardless of whether it’s potentially a product of variance or not. The distinction I’d make here is (1) I’m not basing my views here on very specific moments, but rather at least entire playoff performances—the playoffs as a whole is subject to variance, but certainly not as much as specific moments are; and (2) as mentioned above, I do think that things that are potentially a product of variance can absolutely still validly affect the “greatness” of a player’s year, even if they shouldn’t necessarily have much of an effect on exactly how good we think the player was. To take a very recent example to illustrate these points, I think the fact that 2025 Tyrese Haliburton made a bunch of game-winning shots in the playoffs would increase my ranking of the “greatness” of his year, even if the fact of making those shots doesn’t really move me much in terms of assessing how good I think he is in general.
Regarding 2004 Kobe, I just want to take a step back for a moment and talk about how this came up, so that I can be sure to respond based on where it matters for these purposes. We were talking about Garnett’s 2004 playoff TS%, and you brought up Kobe to point out that his TS% in the Timberwolves series was similar to Garnett’s and Kobe is considered an amazing scorer. Let’s leave aside for a moment the fact of Kobe’s higher volume, and the fact that Kobe’s rTS% in that series was notably higher than Garnett’s. My point was that the playoff efficiency numbers Kobe put up in 2004 were definitely on the lower end for his prime (regardless of whether that’s just randomness or him being systematically worse than normal that year). If those were Kobe’s playoff efficiency numbers throughout his career rather than being just one year on the low end for his prime, then he wouldn’t be seen as quite as good a scorer as he’s seen in reality. Which I think breaks the logical chain of your point a bit. Kobe’s TS% in the Timberwolves series was similar to Garnett’s, but Kobe is basically considered an amazing scorer in spite of his efficiency in the 2004 playoffs, rather than because of it. Which I think really blunts the point you made.
Re: not going by a moment or two but by entire playoffs. I know you believe that, but I would argue that focusing in on TS% changes as statements of goodness is you essentially saying players are taking "good shots" and "bad shots" based on whether the shot goes in regardless of process is essentially going by moment. The fact that you're using aggregated data over the entire playoffs in your assessment just means you're applying moment-orientation coarsely.
Now, I'm obviously focusing a lot on your choice of words here with "good shot" being such a granular, results-oriented thing, and I think it probably feels like I'm straw-manning you when I do this, so feel free to come back and steel your reasoning so for example when you're saying variance can affect greatness even if it doesn't affect goodness, you can expand on that, but just remember that your previous wording didn't make that distinction. You connect small changes to TS% over small samples to changes in how good a player was playing, and that's why I pushed back as I did.
lessthanjake wrote:Re: Dirk & LeBron in 2011. Just to address quickly:
LeBron was the better player, but Dirk was the more valuable player.
From a POY perspective, I see Dirk as the clear #1 for the year, but LeBron's value was being held back by the fact that he was playing with 2 stars who fit poorly next to him and then a bunch of scrap and hence, from a goodness perspective, no reason to argue LeBron was a worse player in 2011 than he'd been in 2009 or 2010.
I like everyone else will tend to favor the years in which a guy was able to max out value in this project over the years in which he didn't, but I see it largely as a kind of tiebreak. The fact that 2011 wasn't the most glorious for LeBron doesn't mean we can't do talk about his play that year as prime LeBron.
Yeah, I think I agree with this. However, if we had to rank 2011 Dirk and 2011 LeBron for purposes of a “greatest peaks” project, I’d very easily rank 2011 Dirk higher, because he very easily had a “greater” year, even if LeBron might have still been a better basketball player that year in a vacuum.
That makes sense and continues the distinction between greatness as connected to impact while further removed from goodness.
As I've said though, where this tends to bleed back into the analysis is when we start focusing on granular details to nitpick our lists.
Choosing peak Wade over peak KG is one thing, choosing it with a major focus on KG seeing his TS drop 2-3 percentage points across small sample is another. Are those percentage points the primary reason why KG should drop X number of spots in a greatest peak conversation? If so, then that's letting the tail wag the dog.
lessthanjake wrote:lessthanjake wrote:
To be clear, I’m not really thinking that Garnett was systematically worse in the playoffs. I just think that in the small but very important sample of the playoffs, his performance wasn’t quite up to the level he was at overall in the regular season. If those playoffs were played over again, maybe that wouldn’t have been the case. But what happened in the playoffs is what happened, and I do see a significant gap between 2006 Wade’s playoff performance and 2004 Garnett’s playoff performance, even though I don’t actually think Garnett was bad at all in the 2004 playoffs. He just was clearly not as good as 2006 Wade, which there’s no shame in.
And what I'd emphasize is that if your judgment over who the better player is ends up focusing on a player's TS% going through pretty typical variance in small sample size, then you might be over-indexing on things a scout would believe it wise to not conclude much from.
On this, I’d emphasize what I said earlier in this post, which is that I’m not using playoff data to make a judgment over who the better player is in general. I’m simply using playoff data to answer the question of who played better in the playoffs. And being better in the playoffs has a lot of independent importance in a discussion about “greatest peaks” even if the playoffs is a small sample that we can’t necessarily extrapolate to answer who the better player was in general. I actually think 2004 Garnett was probably generally a better/more-impactful player than 2006 Wade. But I don’t think he had a greater year, because in the most important part of the year, I think Wade played way above his normal level (dragging his team to a title in doing so), while Garnett played decent bit below his normal level, with the result being that Wade was significantly better in the most important part of the year.
Hmm, here you're back to using misses & makes as a proxy for goodness of play as if goodness of play is a different concept for goodness in general. This is what I am rejecting.
A guy doing his normal muscle memory motion that he makes 47% of the time isn't "doing it right" 47% of the time and messing it up 53% of time, he's just doing his thing 100% of the time, and that's how the ball bounces.
lessthanjake wrote:Re: Garnett vs Wade. So, I've largely avoided the actual comparison part of this and just focused on the drop in TS% you alluded to, but I'd note that if you'd think the same about Wade > Garnett even if he had the same TS% as in the RS, I don't think that's absurd. I can quibble about other things of course, but the thing that concerns me isn't Garnett losing the comparison, but the idea of putting a lot of focus on changes to TS% that to me seem largely shaped by variance that is actually the norm in how basketball stats look.
I suspect my general reaction to this is probably obvious from my above responses. In terms of discussing 2004 Garnett and 2006 Wade in the playoffs, I’m concerned with what happened in the playoffs for its own sake rather than as a way to extrapolate a view on the two players’ general goodness, so I’m not really all that concerned with whether the playoff performances were shaped by variance or not.
But yeah, I imagine I would think 2006 Wade > 2004 Garnett even if we upped Garnett’s playoff TS% a bit. If we also increase Garnett’s per-possession volume to be in line with the regular season and decrease his turnovers to be in line with the regular season, then suddenly it might become a tougher question. Not because I think that’d make 2004 Garnett’s playoff performance as good as 2006 Wade’s. I don’t. But I think 2004 Garnett was better in the regular season than Wade was so if the playoff gap got small enough, then I could put 2004 Garnett ahead.
Yup, I'll skip this and head to the finale rather than repeating what I've said above, and just emphasize again that I'm sorry for anything that felt insulting coming from me.
lessthanjake wrote:Re: Duncan better suited to produce 3's for his team because of gravity. So I get the concept, but I don't think the WOWY stats back it up. I'll try to find data later today if someone else doesn't do it first.
FWIW, according to PBPstats, the 2002-2005 Timberwolves had a 14.02% 3PAr with Garnett on the court, and a 16.52% 3PAr with Garnett off the court. So they did attempt fewer threes with Garnett on the court. To be fair, the Spurs in those same years had a bit higher of a 3PAr with Duncan off than on too (19.20% with Duncan on and 20.69% with Duncan off). All of this suggests that there’s more to the 3PAr of these teams than just Garnett and Duncan, but also that Garnett probably had a bigger negative effect on his team’s 3PAr than Duncan did.
Okay good, PBPstats is a great thing to look at here.
I would start my analysis though looking to see the WOWY effect on each of the teammates rather than the team as a whole because when a star is 2-point-oriented and shoots at volume, the team will probably shoot less 3's when he's out there. And of course, back then, all these stars were 2-point oriented.
So in '03-04, here were the main 3-point shooters by volume in the regular season on the Wolves:
1. Sprewell
2. Cassell
3. Hoiberg
4. Hudson
5. Wally
Now their personal 3PAr with and without Garnett:
Sprewell: 25.3 & 14.1
Cassell 14.7 & 14.1
Hoiberg 46.7 & 39.6
Hudson 43.5 & 26.3
Wally 24.4 & 12.9
So we can see that all of these guys are more 3-point oriented with Garnett than without him, and some of them by quite a lot.
What about their 3P%?
Sprewell: 34.8 & 15.4
Cassell 40.4 & 36.0
Hoiberg 38.1 & 65.8
Hudson 40.4 & 40.0
Wally 45.5 & 38.5
Here the results are in some ways less clear cut. A naive interpretation of Hoiberg would give the impression he's the GOAT 3-point shooter whenever Garnett's not playing, but that's obviously just sample size.
On the other hand, Sprewell shot by far the most 3's on the team, and he was way higher by both 3PAr & 3P% with Garnett on the floor, so while the explanation for what happened may be more complicated, we can say pretty clearly that Garnett wasn't holding back the most significant 3 point shooter on the team (Spre).
If I do the same thing for the '02-03 Spurs with Duncan:
Main 3 point shooters:
1. Jackson
2. Parker
3. Bowen
4. Ginobili
5. Kerr
3PAr with or without Duncan:
Jackson 37.0 & 33.1
Parker 23.5 & 22.1
Bowen 48.6 & 40.8
Ginobili 38.4 & 34.5
Kerr 53.3 & 39.3
and 3P%:
Jackson 30.2 & 40.8
Parker 32.7 & 41.4
Bowen 44.5 & 40.0
Ginobili 29.6 & 47.5
Kerr 40.4 & 37.1
I don't know about you, but to me if anything, Garnett seems to be having more of an effect on shooting accuracy.
Zooming out qualitatively to the theory behind Duncan being the more beneficial player for his 3-point shooters:
If I'm understanding your perspective, I'd expect the reasoning is that low post should have more gravitational capacity than high post due to being further from the 3 point line.
While that makes sense to me, something I'd note is that when we talk about great big man passers, we're generally talking about guys playing high post (they may play low post oriented too, but not in their big passing years). Prior to the 3 the reason for that was pretty clear: a) Easier to get the ball to the high post, b) Less clogging of space by basket, and c) the high post is the epicenter of the half court which maximizes passing angles.
While the new focus on the 3 gives us the possibility of that changing things, I'm not sure I see any evidence it has. While Jokic does post up gradually getting closer to the basket, and when he gets close he can throw great passes out to the perimeter, he's primarily a high post guy, and his forays to the low post are more about taking advantage of what the defense gives him rather than being dependent on being down there so he can make shots.
Additionally, while low post gravity is certainly a thing and has the unique benefit spatially of potentially giving all the perimeter players an open shot, the man who is synonymous with gravitational impact is Curry, and that's now how his gravity works. Curry's gravity is mostly about pulling the defense to one side - making the strong-side defense have to fortify itself to be super-strong, and there by leaving open space on the other side of the half court which gives both open 3's and lay-up opportunities.
I'll head over to nbarapm's 6 factors now for a bit, but will note that I'm using their 4Y data as the primary. Not trying to cherry pick - it's just that while I think the WOWY style of play factors can be seen quite well with a single year, anything with regression would be better with more years.
So with the 6 factors, I want to emphasize the Off TS Val & Off TOV Val impact. The first is the more important factor generally, and the more tied to that player's passing, but the latter is important, and is particularly noteworthy when talking about players who typically look to plant themselves in the paint and rely on teammates to give them the ball.
So for KG, if I use the 2006, which would mean '02-03 to '05-06, Garnett has values of +4.3 & +1.3 respectively for the stats, and adding them together yields a +5.6.
For TD, it's more complicated because his TS peak is in the 2003 study but his TOV peak is a tie between 2008 & 2013. But I can say that the 2003 study represents his overall peak.
So for TD, if I use the 2003, which would mean '99-00 to '02-03, Duncan has values of +2.8 & +0.4 respectively for the stats, and adding them together yields a +3.2.
So +5.6 is a lot bigger than +3.2, and just from a RS perspective, I don't see these guys in the same tier at all in these categories. It's possible the PS is drastically different, but I don't think that's likely.
And yeah, all this relates to why I furrow my eyebrows when people imagine the Spurs playing Duncan like Jokic in today's game. You have a guy who is not your best ballhandler, shooter, or passer, and playing through him means risky passes that turn you over, but you want him run the offensive scheme through him anyway. Why would you do that? Just because he won chips decades ago playing volume post scorer on a team that won with defense? Not a line of reasoning that resonates with me.
Okay, let me hit one other set of stats I got from TB - just KG related:
From 2002-04 in the RS for KG:
Team TS% with KG on: 53.4%
Team TS On-Off wrt KG: +4.5%
Mates TS On-Off wrt KG: +4.1%
So, better with KG out there, mostly because his teammates shoot better with him out there.
Same time frame but PS:
Team TS% with KG on: 53.6%
Team TS On-Off wrt KG: +8.1%
Mates TS On-Off wrt KG: +8.9%
So, slightly better TS% compared to in the regular season in KG"s minutes, but much bigger WOWY indicators for KG.
This painting more of the same picture that there's really not a reason to demand an explanation for underachievement with KG by any of this data. HIs teams did well with him in both RS & PS, it's just that in the PS they eventually got eliminated, and since that happened every year rather than just most years (like it is for other superstars), we feel a need to explain why something NEVER happened by saying it COULDN'T happen and end up moving KG down a tier based on tiny pieces of data that we really wouldn't want to stand behind when push comes to shove.
