So does anyone have offensive stats with and without Irving? Small sample size but that should say something.
So far this year. But, But We don't need him.
Off. Rat. Def. Rat. Net Rat. TS%
Kyrie on court 112.7 103.6 +9.1 58.0
Kyrie off court 100.4 101.4 -1.0 53.9
Kyrie on court 116.1 104.0 +12.2 58.9
from Nov.1st
until now
The TS% numbers when Kyrie is on the floor are what championship contending teams put up and there is still room for improvement as many players have under performed to start the year. The numbers without Kyrie should improve greatly as Hayward gets his wheels under him and can shoulder the load when Kyrie is out.
See, the problem with this is that the "Kyrie off court" includes both games in which Kyrie started and games in which he did not. Yet the "Kyrie off court" numbers are very different in those two sets of games. Think about it: Your data shows a negative net rating for the "Kyrie off court" ... yet the team is 3-1 in games without Kyrie.
Here is a breakdown of the "Kyrie off court" minutes this year:
Kyrie starts, but on bench (1140 possessions):
ORtg 101.9
DRtg 105.9
Net -3.95
Kyrie doesn't play (387 possessions):
ORtg 118.1
DRtg 101.8
Net +16.28
So as you can see, the numbers for the two scenarios are very, very different. I haven't run them yet for last year, but I expect there is a similar disparity.
4 game sample against 4 western conference teams who are all currently outside the playoff picture. 3 of the 4 games were at home. You have commented about being careful with analysis of statistics and then you quote a ridiculously small sample like that? That wouldn't pass muster in any 1st year statistics course as evidence of anything.
If you look at the entire 45 game sample you get +85 if I did the math correct (not doing it again lol)
Those numbers would be indicative of a 45-50 win team IF the sample was an accurate representation of the C's without Irving (it's not). That is probably just about what I would suspect, maybe a 4-6 seed in the Eastern Conference with potential to be a bit better with a healthy Hayward for an entire season. With Irving this team should obtain home court at least for a round or 2, start playing like a 60+ win team and have title aspirations when all the parts come together.
This is conflating the different questions one asks when analyzing data.
If the purpose is to figure out how the team performed, on average, in situations when Kyrie started and then rested in-game, then the first sample is relevant. That question is useful for wondering: How did our bench perform while Kyrie was resting in games that he started?
If the purpose is to figure out how the team performed in games in which Kyrie did not play, the the breakdown I gave you is the accurate partitioning of the data. That question might be more useful in regards to how well the team might played when Kyrie should not be able to play any minutes. I.E., should he be lost to injury for a long stretch, as has happened as recently as last year.
If the purpose is to figure out how the team performed, on average, in all situations when Kyrie was not on the floor this year, then the aggregate mash-up of the 2 samples is appropriate. That question is actually only really useful for wondering: Wouldn't it be great if Kyrie could play 48 minutes per game in all 82 games?
Yes, a 4 game sample is small. A larger sample would be preferable and if I have some spare time I will try to do a breakdown of last year's data. But the data is what it is. You can't say that the team performed differently in games in which Kyrie did not start than how they performed. In particular, you can't say how they performed in those 4 games using data from a completely different set of games.
The point of fact that I'm trying to make clear here is that over-sampling, taking in too broad a swath of data is not necessarily always a better thing. The data should be taken from the thing which you are trying to measure. Adding in data from other datasets just to make your dataset larger is not helpful if it doesn't actually pertain to the question you are trying to answer.