Hey everybody, read some interesting discussion lately here on the topic of "getting your stats" vs. "helping the team win". It seems that most of this is centered around Tyreke, but it is an interesting question in general. As follows, a quck analysis of PER and win shares per 48.
Before the analysis, let me give a personal story. Growing up my best friend was hands down a better ball player than me. I played him way too much 1 on 1 and lost way too often. But this guy never gave any effort in games; pick-up or organized. Long story short, I noticed over time that my teams seemed to win more often than his, despite the obvious talent discrepencies. Now, I am not a bad player myself, but my thing has always been constant hustle and game smarts. So, this is a particularly interesting topic from my end.
To the analysis. I took all players from last year that (a) qualified for the minutes per game leaderboard and (b) were not traded midseason. This gives 185 players, or about 6 per team. So, basically these are the heavy rotation guys.
A major problem with most amatuer stats analysis is the lack of context. Really, saying a guy has a 17.5 rather than 16.5 PER only tells you one player performed better. It says nothing about HOW MUCH better. This is why most stats people use means and standard deviations. It tells us if one subject is SIGNIFICANTLY outperforming the rest of the sample. Therefore, I have taken the raw data and recoded it in such a context. Players are now rated on the following scale: -3 (Very bad), -2 (bad), -1 (below average), 1 (above average), 2 (good), 3 (very good). A score of 0 would be exactly the sample mean; no players land there. This is done for PER and ws per 48..... Check out how the Kings come out....
J.T.: PER = 1 ws48 = 1 .... Cousins: PER = 2, ws48 = -1 .... I.T.: PER = 1 ws48 = 1 .... Thornton: PER = 1 ws48 = -1 .... Evans: PER = 1 ws48 = -1 .... Jimmer: PER = -2 ws48 = -2
These results are really amazing. They exactly reflect what I see as the majority opinion on here. JT & IT are guys who help out in multiple ways they are net plus players. Cousins, Thornton & Evans have high PER numbers, but fail to contribute in other ways. Jimmer is bad all around. FYI, James Johnson comes out as "below average" (score of -1) on both PER & ws48.
What is possibly most interesting to me, as a stats person, is that these two measure of player quality that I have constructed are correlated very high (.739 or 73.9%). In other words, it is a general truth around the league that players with high PER have high win share rates. Somehow, this is not translating for some of the big minute players on the Kings.