OhayoKD wrote:There is no approach to RAPM to get Jokic to #1 beyond comparing the average of a much smaller sample to the average of a much larger one. If it wasn't to your chagrin you would be comparing over similar time frames, or checking frequency.
Comparing even samples is not desperate, it is an essential step and one you are constantly disregarding because you are either desperate for a certain conclusion or are just incompetent with numbers.
Liking Nikola Jokic is not an excuse to pretend averages over half the minutes of another player support him as being better over a similar-time frame(peak, prime, ect), and it's sad this has to be explained to you in the first place.
This is just silly. It’s a bizarrely elaborate way to say that you think looking at career RAPM is improper. After all, it’s career RAPM that Jokic is #1 in, so that’s what you’re saying is improper. Of course, I’m quite certain you’d think otherwise if LeBron were ahead of Jokic. But anyways, what do you want instead? To compare RAPM specifically in their peak years? That may be nice if we could do it, but there’s essentially no RAPM measures (with no box prior) that have multi-year data for both Jokic’s and LeBron’s peak years (and there’s serious potential different-scaling issues if we did). The one RAPM measure that I’m aware of that has multi-year data for Jokic’s peak years is NBAShotCharts, and it’s worth noting that Jokic’s luck-adjusted NBAShotCharts RAPM for 2020-2021 through 2022-2023 was the highest recorded by any player in any three-year span in their data set, including being higher than any three-year period from LeBron (the first three year data is 2013-2016, though, so it doesn’t have data from some of LeBron’s best years). And, since we don’t really know that things were scaled the same in different timeframes, I’ll note that Jokic absolutely cooked the next highest people in that timeframe, with *by far* the largest gap to #2 anyone had in any timeframe. So, if you don’t like career RAPM, the RAPM data we have from Jokic’s peak years specifically is also extremely good and is consistent with him being the #1 RAPM player overall. And that’s not even getting into the fact that most impact measures we talk about are basically just RAPM with some sort of prior, so peak Jokic doing incredibly well in those (as I showed in the post I previously linked to) is obviously indicative of peak Jokic having incredible RAPM.
Speaking of which
And yet you felt the need to
A. Compare uneven samples(5 years of Lebron: 4 for Jokic)
B. Not choose the best sample for Lebron (swap 2013 for 2008 and...he peaks higher than Jokic)
C. gigantically shrink the sample in a way that throws out Lebron's best data
And the two are still basically tied.
The fact Lebron has multiple 5-year stretches comparable(and one better) to the 4-year one you picked for Jokic(15-17 would also be up there) is a big hint that these are not comparable players. And of course, if you were really worried about "end of season games", you could have always went with what was a similar off-sample with a similar rotation concentrated in a single year(making it a much larger sample per-season):
You are wanting to cherry-pick LeBron’s best time periods in different stats and compare a nonexistent-in-reality combination of those against what Jokic has actually done in one particular time period.
Please stop flinging rhetorical terms you do not understand. Cherrypicking is when you take a smaller part of a whole to draw a conclusion about the whole.
By the measure you chose, Lebron still peaks higher than Jokic. Swapping years to get Lebron peaking lower is an example of cherrypicking. Only choosing 4 years of Jokic as opposed to 5 for Lebron is also cherrypicking. Removing the vast majority of the sample here is...cherrypicking..
The chosen approach and frame of comparison both cherrypick in a manner which negatively affects Lebron's numbers and... Lebron still peaks higher over more years.
At least with RAPM where outliers are artificially scaled down, if you made the comparison uneven enough to get an output that prefers Jokic. With WOWY, not even an uneven approach helps you, which is why you went with a prior instead of just comparing the best samples.
This is just complete projection. I am not the one “swapping years.” The analysis I provided you gave data for peak Jokic and peak LeBron across a ton of different measures. For both players, I used the same timeframe across all the different measures I looked at (i.e. the last 4 years for Jokic, and 2009-2013 for LeBron). That is the *opposite* of “swapping years.” I was completely consistent in what years I included and what years I didn’t. Meanwhile, you respond by saying that, for one of the measures, I should’ve measured using different years for LeBron, because those years would’ve been better for him in that measure. It is transparently you who is advocating for “swapping years.” And it’s quite ironic that your response to me saying that that’s cherry picking years was to say “[s]wapping years . . . is an example of cherrypicking.”
Again, pick a specific timeframe for LeBron that you want to compare peak Jokic to, so that we can compare the overall picture of all available data that these players had over those specific timeframes. Otherwise, you’re just picking and choosing what years to look at for different measures, creating an overall data picture for LeBron that never actually simultaneously existed, in order to say that this Frankenstein non-existent version of LeBron has better data than peak-Jokic. When I actually picked specific timeframes for both and compared them and used the consensus peak timeframe for LeBron (2009-2013) and Jokic (2021 onwards), Jokic looked good in the comparison, and your only response is to advocate for selective swapping of years in some measures if that’d help LeBron.
WThe correct and fair approach is to fix the time period for LeBron (which I did, using his pretty undisputed peak years) and compare data from just that time period to peak Jokic.
The correct and fair approach would have been to take the best 4 years of Jokic and compare it to the best 4 years for Lebron. You added an extra step(make it 5 not 4), still could not find the conclusion you wanted, and cooked the numbers accordingly.
Agenda-motivated stat-work at its finest.
This is absurd. You make this response while ignoring the fact that my post explicitly pointed out that I took a 5-year span for LeBron because, on balance across all data, that’s actually generally *better* for LeBron than taking any 4-year span would be (since 2009 and 2013 are great seasons and can’t both be in a 4-year span). Not only did I explicitly point that out in the post you responded to here (a part that you conveniently didn’t quote or respond to), but I’d already pointed out that exact same thing in the underlying post I made with the data we are talking about. So I have made this point multiple times now, and you just ignore it. And then it’s you that accuses me of “agenda-motivated stat-work.” Wild stuff. If you want me to compare Jokic’s last four years to LeBron’s peak four years, I’m happy to do so. Just tell me what four years you want to look at across all data. 2009 to 2012, or 2010 to 2013? Either way, on balance, the data picture is not going to look better overall for LeBron than the 2009-2013 time period did, so it’s really not going to be generally helpful to you.
And I have no idea what you’re talking about with “cook[ing] the numbers accordingly.” I didn’t cook any numbers. You just don’t like what the numbers say.
Put differently, your approach is to basically create a Frankenstein version of all of LeBron’s best data points that didn’t simultaneously occur, and then compare them to what Jokic has actually done in one specific timeframe.
I will repeat, Lebron peaks higher in the metric and (uneven) time-frame of your choosing, just like he scores a higher average over 13 years than Jokic does over 3. None of these are favorable approaches to Lebron, but Lebron still wins because he's better at basketball and more valuable to winning. As is also reflected in RAPM and, wait for it, on/off.
This seems to just be complete denial. I gave you a metric—WOWY SRS impact with end-of-season garbage games not included—and peak Jokic had a larger WOWY SRS impact than consensus-peak 2009-2013 LeBron. You then baselessly say that LeBron is higher in this metric in the timeframe I looked at. No, he’s not. WOWY is a very noisy stat, so there may be other non-peak timeframes where LeBron’s WOWY is higher, but LeBron’s output in other measures of impact in those timeframes would be far lower, so the overall data picture would still be worse. To the extent you want to actually argue a point about WOWY without undermining arguments regarding other data, your best bet is to point out that peak Jokic’s WOWY SRS impact was only slightly higher than peak LeBron’s WOWY SRS impact, but peak LeBron’s “on” SRS is a bit higher so it’s not really clear that peak Jokic’s WOWY SRS impact is actually superior. That’d be a fair point. What you’re saying here is not.
This is a consistent theme with you. When I’ve pointed out that someone like Steph has vastly superior impact data to LeBron in the years of Steph’s prime (i.e. 2013-2014 onwards), you have repeatedly responded that those aren’t LeBron’s best years and we should be comparing Steph’s data to LeBron’s data in his peak 2009-2013 years.
And this Jake, is an actual strawman. I have never specified 2009-2013 or even a specific time-frame.
Lol. Well, in discussions about Steph, you’ve stated to me that LeBron’s prime “barely” extended past 2012-2013. The obvious implication is that LeBron’s best years were from 2012-2013 and earlier. Meanwhile, you have also argued that years before 2008-2009 were not LeBron’s prime. So yeah, I’d say that the clear implication from what you’ve said is that LeBron’s best years were 2009-2013. Which would also be perfectly reasonable, since that’s the consensus peak period for LeBron. If I’m wrong about what you were saying LeBron’s peak timeframe was, then please enlighten me, so that I can do analysis across various measures for that specific timeframe.
What I have suggested was factoring in everything and comparing their best data to each other while taking into account the entirety of their careers, as opposed to just Miami, when deciding how to mentally curve/frame things. You don't quite like this because the whole of their careers suggests something very different than what one might derive if they only cared about Miami(and if we're being honest, specifically 2011).
This is you once again just wanting to create a Frankenstein data picture of LeBron that never actually simultaneously existed. The measures we look at in basketball are noisy. If you just take the “best data” for a guy with a long career, using different timeframes for different measures, and compare it to the data for a current player’s peak, you’re just creating a picture that never really existed and comparing it against someone who does exist. It’s not a fair comparison whatsoever, and is basically just leveraging as much positive noise as possible. No. Comparing all data in a specific timeframe for one player and all data in a specific timeframe for another player is the only reasonable way to compare peaks. You don’t want to do that because you don’t like the results (which I’ve shown in a lengthy post in another thread that I linked to).
Lebron cooks Curry with the same years he cooks Jokic with and cooks both (again) if one bakes in playoff translation once he returns to Cleveland, and his natural position of small forward, since the cold regular-season data from 15-17 marks him as comparable to both. What is actually happening here is you want to curve down Lebron's best signals with the justification that Miami is his real peak without considering what "peak" implies. It is natural to curve Miami up and Cleveland down. It is natural to adjust the curving based on which is more unusual for the player(hint: it's Miami). It is natural to drop the curving and take the data at face value. What is not natural is throwing out the best signals because they are "no peak" years so you can present the player as worse statistically than he actually is.
Okay, so do you want to define LeBron’s peak as 2015-2017? Do you want me to compare 2015-2017 LeBron to peak Jokic across various metrics? Because I promise you that won’t look more favorable for LeBron overall than the analysis using 2009-2013 did.
In "massively reduced off sample" WOWY, Jokic's best 4 years lose to Lebron's best 5 years. That is the statistical lede but you wanted to bury it and we both know why.
Again, this is just you completely in denial. In the WOWY SRS impact data I provided from both of their peak years, peak Jokic had larger WOWY SRS impact than peak LeBron. And yet here you are just baselessly saying “Jokic’s best 4 years lose to LeBron’s best 5 years.” I gave you the data. And it actually took a while to calculate. The least you could do is not lie about it. And the irony of you then accusing me of “bury[ing]” the “statistical lede” is certainly amusing.
Come up with a LeBron time period that you want to compare peak Jokic to and stick with it across all types of data.
The key here is Lebron wins any frame(including the one you chose) not a particular one, something we can emphasize with:
Lebron 09-21
656-263 with lebron 0.714% win rate
37-73 without lebron 0.336% win rate
net rating with lebron +6.49 (59 win pace level)
net rating without lebron -5.50 (25 win pace level)
+8.6 ortg difference
-3.68 drtg difference
+12 total swing
jokic 2022-24
136-68 (66.7% win rate) with jokic
8-15 (34.8% win rate) without jokic
+4.1 net rating with jokic (53 win pace)
-4.6 net rating without jokic (28 win pace)
+6.5 ortg change
-2.2 drtg change
+8.7 overall change
Lebron over 13>Jokic over 2. Lebron over 5 with your approach > Jokic over 4. This is why presenting them as statistical peers is comedy.
Okay, so you’re not willing to provide any peak timeframe to compare all data across and are once again just going with lying about what the data I compiled and provided said. Lol.
None of which includes the ball-handling component I listed. Additionally, as various people have brought up and you have handwaved, RAPTOR's creator straight up said they think paint-protection and big defense is overrated, 2019 kawhi was the best defender in the nba, and they weighed their inputs accordingly.
This is a largely irrelevant rabbit hole, but since I’ve seen you make this claim many times but never actually seen any evidence of it, can you show me the basis for your claim here that “RAPTOR's creator straight up said they think paint-protection and big defense is overrated, 2019 kawhi was the best defender in the nba, and they weighed their inputs accordingly”?
Lastly, if you take out the box inputs...Lebron is the figure who stands as a gigantic outlier over the field.
You’re talking about RAPTOR here, and RAPTOR’s explanation specifically says that the non-box component is far less predictive of large-sample RAPM than the box-component. Specifically, they say “‘Box’ RAPTOR is quite a bit more predictive of out-of-sample performance than ‘On-Off’ RAPTOR.” You are just talking about using one component of a metric, and the less-predictive component at that! Talk about cherry-picking! Of course, this isn’t even an accurate claim anyways, since LeBron’s average in just the non-box component of RAPTOR is not as high as Steph’s. For instance, from 2014-2020, LeBron’s RS+Playoff average is +8.27, while Steph’s from 2014-2022 is 9.53. But yeah, I’ll give you that LeBron does a little better than Jokic in one specific, far less predictive *component* of a metric that Jokic dominates LeBron in overall. Hurray!
He is not even a match with box-inputs unless you cherrypick when the comparison starts for the bulk of available metrics and pretty much none of said metrics consider Lebron's gigantic advantage as a ball-handler and ball-progressor or what he offers as a defensive floor-general.
Lol, again, the accusation that I “cherrypick[ed] when the comparison starts for the bulk of available metrics” is just completely breathtaking, when I looked at available data for LeBron in one specific timeframe (2009-2013), while you are the one out here talking about cherry-picking LeBron’s best years for different measures.
Heej has already covered how silly "ball-handling is not a problem" is, so I'll let him cook.
Heej didn’t even really disagree with me much overall. In fact, Heej said “I agree with you that Jokic is one of the best ballhandlers ever for his size and that it is a plus.” So yeah, I don’t think that was a discussion where anyone really agreed with you. And I’ll note that Jokic just showed his fantastic ball-handling again in Game 5 against the Timberwolves, repeatedly driving to the hoop, working in the post without turning the ball over, bringing the ball up a fair bit, etc. It’s a big positive in his game!