No-more-rings wrote:For me it’s still KG pretty comfortably. The project got it right even though i’d have him more than just one spot ahead. KG is the better defender by a decent amount and I don’t think I’ve seen the case that Giannis stacks up in impact data.
Impact data depends on the person running it for rapm I think, but the shotcharts one that goes to 2010 iirc I think in terms of std dev in the year and stuff giannis’s 2020 season is about where brons 2010 one is, I’d have to check again tho (and I think the shotcharts data for 2010 is pretty similar to the data for 2010 from JE Dropbox)
Edit: yeah act it was a good deal better in Std devs, but it might scale differently for later years tho? 2010 bron is higher than 04 garnett in that regard so transitive property
Another way to look at it is seeing what data JE did release (he went up to 2017 as far as I know) and trying to see the shotcharts data under that light.
It’s a bit convoluted, but unless JE changed his scale for his 2017 data vs his other years of data, I think that’s the GOAT seasons by far in terms of standard deviations of impact data by RAPM (for reference, 2010 bron is +3.78 standard deviations away, 04 garnett is +3.4, both are I think, the best Shaq year in the set is at +3.5, Curry and bron are both at +4.2 ish, in standard deviations, so it’s an outlier among outliers basically)
Giannis’s data in that separate dataset is like a tad bit worse than the 2017 data but not close to that extent z
Iirc He also has the best LEBRON ever, which goes from 2010-2022 I think, LEBRON isn’t perfect or anything but that’s mostly from it overrating shot blocking bigs at times, which he isnt, but in a general basis for all in one stuff for single year I know for a fact it craps on anything other than EPM and I think it’s the most predictive of all the impact data stuff
The main thing is giannis and the bucks as a whole haven’t really tried as hard on defense win the RS the past few years, which I think is hard to deny unless you view them as a league average ish defensive team, so his impact data is a bit down. His 2020 DPOY year iirc is up there all time in terms of his defensive impact data (I think it’s better than pre Celtics Garnett’s actually
Obv I don’t have him as high as what that would suggest but I think giannis has totally shown that if he goes tryhard he’ll have historic impact data, his 2020 season goes up with luck adjusted metrics too which generally isn’t the case for top years.
By standard deviations it actually beats out 2017 currys in the shotcharts data set with luck adjusted RAPM (I think that’s the only place that has that), which is entirely due to 2020 giannis’s rapm increasing rather than the other way around, so at the very least that would imply it’s not a situation where the noisy aspect of impact data helped out giannis vs it hurting an ATG impact season (to be clear he was third on my ballot that year by a distance lol)
So like, if we’re talking about cross year RAPM comparisons, it’s kind of difficult to compare separate years in general but it’s not exactly lagging behind, it’s a bit convoluted but basically it looks like by shot charts stuff (which is the only one that has multiple years of data that include a year higher than Garnett’s + Giannis’s peak rapm year) but giannis’s season comes out as the 2nd best year RAPM ever in its raw form and the best ever with luck adjusted
If shot charts changed their scale for later years, then 2020 giannis still can be inferred to be the second best ever based on JE’s 2017 rapm vs his 2001-2015 rapm data set as well, because of how close it was to 2017 in general in that separate data set
Also LEBRON has his 2020 year as the best season in its dataset which goes 2010-2022, very slightly ahead of 2010 lebron (both kind of form a clear tier ahead of every other year)
Giannis argument revolves around his playoff D being similar to his DPOY D in 2019 and 2020, I think the impact stuff and lineup data does point to that.
As far as I know in impact data (2020)
The only things I know that you could directly compare them would be
PIPM (I think giannis had a higher one? But this is just a discount LEBRON basically)
RPM (which sucks to be clear, but giannis wins this one)
Aupm (I think garnett wins)
RAPM by std dev (unless JE changed his scale here, giannis wins pretty heftily)
I can’t think of other stuff where there’s data in them both, but if we’re going by impact stuff I feel aupm is arguably the weakest one here in terms of how it’s calculated (rpm is fully a private one tho Tbf lol but it doesn’t perform super well in testing)
Now of course the argument is whether 2021 playoff giannis = 2020 RS giannis, which I feel you could make a good case for