04-27-2021, 03:55 PM
(04-27-2021, 09:41 AM)mvossman Wrote: I did not see your prior post, but the main point of my argument is that because defense is so hard to capture in a stat, you need a very large sample size. I would not expect a defensive stat to do a very good job in the relatively small sample that we have this year (that's why I went over 3 years).
I mean lets look at some of the context of this year. Powell started the first 9 games and was terrible, both offensively and defensively. He clearly was not ready to play anywhere near his prior level. Yet the defensive rating of the team looked good. Why is that? Likely because opponent 3 percentage was abnormally low (another stat that needs a big sample to be of any value). Then he played sparingly vs bench units as he tried to get back to his old self, only recently starting to get decent minutes. He is still averaging less than 15 a game.
Maxi, has not looked the same since Covid. Not sure if this is going to be a problem for the rest of the season. He seemed to recover after the all star break but has really struggled the last several games he has played in (don't know if that is injury related).
The point is this is not really a good season to evaluate these guys as they have both been dealing with unique circumstances. The question is who is the better defender in general, and for that I think it would be more useful to look at a multi year sample.
Also, not all stats are the same. Some are more useful than others. I'm a particular fan of RPM because it takes about as much context in account as you can. There are others that I found useful, not as an overall stat, but something to contribute (like Raptor or the stat someone posted earlier showing the quality of offensive opponent). I would be curious to see some of those stats over the last couple of years instead of just this year.
I think you make some good contextual points. Teams weren’t shooting as well overall in the early season. But, Dallas was ranked 4th defensively in that environment and Powell, Maxi and WCS were all posting very similar D-Ratings (until KP came back). You can’t write off good individual numbers to league-wide poor shooting when the team was beating 83% of its peers who were benefiting from the same issue.
Fast forward to March and April. Yeah, Maxi is struggling. But Powell is again thriving and outperforming the team as the team’s numbers improve. Is it small sample size noise, or a player who if finally healthy hustling his butt off. Or, is it something else. Whatever the reason, the data is the data. My assertion wasn’t that Powell is always the better defender. It was Powell is the better defender this year. Your use of 3 year data to prove DRPM’s superiority doesn’t mean much because 3 years ago Maxi was clearly the better defender...and it wasn’t close. This is confirmation bias. You seem to be a good stats person. I imagine you know the danger of trying to create a narrative to make the data conform to your view as much as anyone. It feels to me that you are doing just that to show your one data point is somehow better than the 8 or 9 that say something different (this year). BTW, half of those data points are just variants of RPM designed to fix issues other stat geeks have found (believe exist) with RPM.
Interesting question regarding D-Raptor. In 18/19 DP-MK were 43rd/93rd percentile. In 19/20 it was 68th/74th. This season it is 93rd/74th. BBall Index has a player comparison functionality that is kind of cool, but I think you have to subscribe to at least some of their stuff.