The-Power wrote:Doctor MJ wrote:In all seriousness, I'm trying to think of ways to apparent Green's lift edge as something other than 1) noise, or 2) Green actually lifting team performance more, and I still have nothing.
Obviously I have noticed Green's edge in +/- as well and pondered on the reasons for it. Let me start off by proclaiming two fundamental things:
1. I do believe Green has superstar-level impact in his own right.
2. I do not believe that the data we have is just noise, it's real because we can see a pretty clear pattern.
So, what does this leave us with and how do I answer the question about impact as someone who is still in Camp-Curry? Well, there is one assumption to be made which we can't verify nor falsify but it's an assumption we need to make in order to have a viable discussion in the first place. The assumption would be: Curry's impact on his team's point differential is greater than Green's when both are on the court. Since they spend most of the time together on the court, this would make up for the majority of their raw +/- numbers and also greatly influence their RAPM. If the assumption is valid, this would mean that Green is being overrated to some extend by the +/- numbers due to playing next to Curry while Curry is being underrated relative to Green overall.
In this case, what we have to explain is Green's superioty when one of them is on the bench. This means mostly guesswork for us and while this isn't particularly satisfying I see no way around it.
1. Line-ups. I don't believe they explain everything, especially given the relatively small sample size everybody is aware of, but they mean at least something. The most eye-popping number with probably the most explanatory power is playing time with Klay Thompson. Green on/Curry off is a sample of 346 minutes, and Klay was on the floor for 297 minutes or 86% of the time. Curry on/Green off is a sample of 240 minutes, and Klay was on the floor for 106 minutes or 44% of the time. The difference is obviously huge and important as long as we assume Klay to have significant impact (which I do, although not at superstar-level). We also see some random outliers like Barbosa who - while only taking 30 TSA with Curry on the floor and Green on the bench, but given the sample size it factors in - shoots is incredibly ineffecient in this scenario (36% TS) and everybody who watches Barbosa play should know his style and it is 'hit or miss' most of the time, regardless of the teammates around him. Curry also played 24.5% of the time with either Varejao, Jason Thompson, McAdoo or Looney - all poor big men, especially on defense - while Green played only 8.6% (!) of the time with them. In other words: Green's replacements are worse overall than Curry's and this matters. We could go on but the first part if definitely more important than anything else we could potentially find.
2. Intensity. The key with Green on and Curry off is not only to maintain decent offensive efficiency but rather to keep up the defensive intensity. When Curry sits on the bench, the team knows that they must be focused defensively and take good shots on offense. In order to maintain offensive efficiency, the pace drops noticeably and the % of FGM assisted reaches an incredibly high level (76%). Having more playing time with Igoudala and Livingston, adding up their minutes played, certainly helps in this regard. Defensively, they know they can overcome the loss of Curry only by being active on defense while not having to play with Varejao/Thompson/McAdoo as much helps as well. The ORTG/DRTG is roughly 110/102 (+8 NetRtg). Curry has to deal with Barbosa and V/T/M more often which drags down defense and offense alike. The ORTG/DRTG is rougly 113/110.5 (+2.5 NetRtg). The defensive intensity somewhat goes with Green but while I do believe Green has clear DPoY impact, I don't believe it's 8.5 points (more like 5-6 compared to the average player) and maybe defensive intensity compared to a lime-up without Curry also goes because the teammates know they should still have the offensive advantage. Therefore we must consider line-ups as well as a different mindset to be important, too.
3. Game planning. We all know that teams prepare against Curry in particular, this is no secret. When Curry is on the floor and Green on the bench, Warriors' most important counter to teams overplaying Curry is basically gone. Therefore the offense becomes worse, although it's still more than respectable (113 ORTG). When Green sits, the Warriors don't have a reliable PnR-playmaker anymore and most of the minutes from Green go to Barnes (he plays 180 of the 240 minutes with Curry in the discussed constellation) who plays a totally different style. In other words: Golden State isn't well-prepared to play without Green on offense or on defense. This speaks volume about his importance but maybe not so much about on-court impact (although it's still high) or individual goodness (although he's extremely good). So what happens with Green on the court and Curry on the bench with the Warriors' offense? Well, here's my theory. As I mentioned above, the team looks to get more good looks and consequently lowers the pace (to 98.9). Not only does Klay help to overcome the absence of Curry to some extent, the Warriors are a well-coached team and they run more sets with Curry on the bench which allows them to maintain good offensive efficiency. But we have to keep in mind: this doesn't work for larger minutes or entire games, basically whenever the teams can focus on stopping the Warriors from executing their sets rather than stopping Curry from doing his thing. Unfortunately for me as a statistician, but fortunately for me as a fan, we only have three games this season in which Curry missed the game entirely and I believe only two in which the other team could scratch Curry out of the game-plan. With Green on the court, the Warriors performed -4.5 (vs. Dallas), +1.1 (vs. Houston) and +0.6 (vs. Atlanta) relative to the expectation (DRTG of the team they faced). Overall, this means their offense was worse than the average offense without Curry and the average ORTG is well below the 110 we can see in the WOWY data. An even smaller sample size, sure, but I didn't want to hide it.
These are the three most important aspects to keep in mind when interpreting Green's and Curry's +/- when one of them is on the bench. If we take all this into account, the ORTG/DRTG would look different and ultimately - this is my and probably most people's belief - favor Curry compared to Green.
One thing I think is that the Warriors performances are "more dependent" on curry, if that makes sense? So like, if curry has a bad game, it will effect them more than green having a bad game, and if curry has a good game, the opposite is true. Now, one could say that this shows green is more consistent, and yes, the way he does impact the game is more consistent, but on this we are focusing on offense, because we know what level his defense is.
And like, in terms of rapm, I feel like a good curry performance will pull up draymonds on-off if he has a bad game. I can't say the opposite is true.
I mean, looking at a few games a 5 game sample size
Febuary 2. Green posts a 1/5, 3 rebound 6 assist performance +- of 16
, curry has a typical 35 point game on 24 shots and 4 free throws +- of 19
Orlando
Curry has 41 points on 24 shots and 6 free throws, along with 14 rebounds, plus minus of 21
Green goes 0/3, 10 assists, 9 rebounds (7 turnovers). + of 19
Detroit. Curry goes for 38 on 26 shots. 5 assists, 7 rebounds + of 2
Green goes 1/7, 9 assists and 5 rebounds. + of 4
Charlotte.
Green goes 2/9, 9 assists 11 rebounds. plus of 19
Curry breaks 100% TS. +- of 19
Febuary 22
Green goes 2/10. 9 assists, 14 rebounds, +15
Curry goes for 36 on 22 shots and 3 free throws, 8 assists and 6 rebounds +13
Okc game, green has a good game outside of shooting, 14 assists, 14 rebounds. 0/8 shooting +10
Curry breaks the 3 point record.+6
Curry is +80
Green is +82
Each game was a win aside from the pistons game, where aside from curry the Warriors shot 30.8%, 3/10 on threes.
Yes, greens impact comes off the box score, but like, no, his non box score impact doesent make up for things like this, where curry has an above average game and green has a bad one. (Missing 4 shots you normally make is like 8-10 off the box score already)
On that last point, in terms of sample size, it's small. And no, I'd didn't cherry pick results, I just took the bottom 6 shooting games green had, sure it's not the best way to look at it but I do t have the time to do it for all games.
And yes, I get its because curry is missing more shots because he takes more shots, but im willing to say a good day for curry> a great day for green, basically I'm trying to say they live and die by curry, and rely more on him.
Doing the same 6 games for curry makes him +17
Green in those 6 is +53
And now you might say "oh but that shows they can rely on green more"
The Warriors are 3-3 in those games, 1 was a one point win against Memphis, the other a 3 point win against the nuggets.
In the win against the Knicks, draymond and klay played amazing,
Aside from curry, the Warriors were 37/56 overall, 13/18 from 3
Now, one might say it's better if we do turnovers for green, and in games where green is
6 or more turnovers, they are 3-4, but this doesent fix the plus minus trend where curry still has about the same if not lower, and I don't think the difference between 5 and 6 turnovers is that huge (it would be a misleading cut off, they won all 7 games with 5 turnovers)
And that could be countered by games where he had 5 or less assists, another cut off that is kinda weird, they are 24-2
Now, looking at games where curry did not play, we have a small sample size. But in that sample size, klay becomes the clear go to scorer.
Against Dallas, both klay and draymond attempted 15 shots. They lost pretty badly, a net rating of -11 for both.
Against the rockets, green has a great game. Thompson though, scores 38 on 27 shots. While he filled the curry role, they didn't have the same lineups, and his plus minus was 10 while draymond was at -4
Against Atlanta, klay isn't very good offensively. 8/27 (6/16 on threes) to get 36 points.
But they won right? But here's my thing with this game,
Outside of klay, the Warriors still **** badly as a unit. 43% Fg. They kept the ball well, green has 9 assists and 5 turnovers but I don't have stats on time of possession so I'll assume he took good care of the ball, though it might have been more everyone else. The plus minus seemed okay here, klay had +2 and green was +9
Well, they had 23 more shots than Atlanta. 18 offensive rebounds, 8 more than their average and 5 more than the NBA leading team. But this was because of Barnes, who had 6 by himself. 2nd most of their season, and the Hawks had 17 turnovers, and green was responsible for 4 of these.
Green was 6/15 overall.
Small sample size but it's notable. Under regular circumstances all of these games would have been losses Imo. The rockets game had klay on curry mode. The Atlanta game had them take 103 shots despite playing at a slightly sub 90 possession pace.
Just to add on something, I think the reason klay and green are nearly always together when curry is off the floor is because, imo, he needs someone that can score, whether from off ball movement like klay or from half courts from curry. Looking at the rockets game, the majority of klays points were off assists, yes, but they weren't "oh you made that" assists, it was a set play, then a simple pass to a somewhat open klay for a jumper. Draymond didn't have much to do with most of them iirc. So this ties in with the coaching theory the power had.
Since klay scores 27.3 per 36 minutes with them on and curry off I think that looking at that rockets play is important, and iirc that's what I saw. If that's true that would explain some of it.
Beyond that we have sample size. Bogut is scoring 73% on 47 points for example, which is not sustainable period. Meanwhile, sleights is above 50% outside of 16 feetto 3 point on 19 shots, etc.
