CJ vs. Curry, Overanalyzed

This springs from a GSOM FanPost arguing that CJ should be starting over Curry, due to his outplaying the rookie. This has drawn outrage, some justified, some not... I have three strong, unconflicted takes on this.

1) Yes, I agree: CJ has been outplaying Curry. In fact, right now, CJ is the better player by a comfortable margin.

To wit, their numbers per 36 minutes:

(i know its small look just click on it)

CJ scores a bit more despite shooting a bit less often; he shoots better from everywhere. Rebounding is (weirdly) an exact push, and CJ actually racks up steals a tiny bit more often, despite getting whistled a good bit less. Curry’s only advantage is in his two extra assists per 36 minutes, which is more than canceled out by his 1.9 extra turnovers. Curry’s passing numbers are not stellar; he ranks 40th amongst point guards in A/TO ratio, behind combo guys like Arenas and Devin Harris and Ramon Sessions. Those turnovers really hurt.

Now, there’s a counterargument, one to which I myself have sometimes subscribed: that with Curry, you have to go beyond the basic stats. Maybe his playmaking has an alchemic effect on our offense that’s not reflected in his numbers. And maybe so! Thus far, our offense is 4 points per 100 possessions better when Curry’s on the floor than when he’s not. That’s an impressive figure. But our offense is 7.5 points per 100 possessions better when CJ’s on the floor than when he’s not… lineups featuring CJ shoot better and score more often than lineups featuring Curry. Any vague offense-improving powers you want to ascribe to Curry, you have to doubly ascribe to CJ. There’s just no way to spin Curry as CJ’s equal on offense thus far, let alone his superior.

Then there’s defense. Neither guy is a defensive whiz, but CJ is two years farther along in learning how to play perimeter NBA defense, and it shows. We’ve defended a bit better with CJ on the floor this year, despite the fact that he’s usually in a hypersmall lineup that cripples our defensive rebounding. Plus-minus numbers don’t paint Curry as an outright defensive disaster, and he creates enough positive plays so that he might not be, but opponents shoot better and score more easily when Curry’s out there. And the average lineup he’s played in is not nearly as undersized.

Grains of salt abound, of course; the sample sizes for all this, especially on CJ’s side, are not big. On the other hand, the margins are pretty big. By Win Shares, CJ’s got a 1.1 to 0.6 edge, despite playing less than 60% as much; projecting them both out to 2500 minutes, CJ’s the better player by more than five wins, 6.6 to 1.0. Basketball Prospectus places CJ at a .521 winning percentage and Curry at .423; they’d expect a team of CJ and four average players to be eight games better than an analogous Curry-led team. So while it’s by no means a given that C.J. Watson is several games better than Stephen Curry right now, it’s certainly plausible. And it’s more plausible than the idea that the two are roughly equal.

2) Yes, I agree: we should absolutely give CJ the starting job tomorrow.

Right now, of course, there are plenty of minutes to go around... as things stand, both guys will probably keep getting at least 25 minutes a night. But we should pencil CJ's name into the starting lineup in Philly, and indefinitely thereafter.

3) These two things have nothing to do with each other. We should not start CJ because he's better, but because we should be showcasing him for a trade.

Like dozens of other NBA players, CJ becomes tradeable starting Tuesday. And as a low-mistakes three-point marksman who scores well and plays the passing lanes, CJ is a type of player that a lot of playoff hopefuls -- Utah, Miami, Philly, OKC, Chicago, Detroit -- might be interested in. It's not an accident that the Magic fought hard to get him; CJ is a perfect second or third guy off the bench for a playoff team. CJ is not long for this world, so we should trade him and get value while we can. But he'll have more value if he's got a still "starter" sheen to him, and maybe a few 20-point nights that GMs hear about. Circumstances forced the Rockets to start Rafer Alston for the bulk of last season; the subsequent increase in his notoriety and counting stats raised his value to the point where they could trade him for Kyle Lowry, a better, younger and cheaper player, who's a nice part of their foundation going forward. So it's time to pump up CJ's value as much as possible. Curry will get his job back by the deadline, and plenty of floor time in the interim; for now, our motto should be "CJ Watson, Superstar."

So, to recap: NO, it doesn't seem like CJ has been a lot better than Curry, but, YES, a closer look reveals that he actually has. But, NO, that doesn't mean we should prioritize a free agent over a rookie in a lost season. But, YES, we should pretend to, to up his possible return in a trade.

IMO, anyway. Blog co-proprietor though I am, I won't pretend that those are, on some objective level, the truly current answers. I just thought it was worth walking through the issue, because it's sort of a fun microcosm of the various balancing acts that player evaluation sometimes requires. You have to account for personalities, visual observations, data of varying levels of extraction; you have to account for salaries, competitive timelines, fan expectations, PR. It can be, in fact, quite a complicated thing to figure out who the starting point guard of a shitty team should be. And we beleaguered Warriors fan can only hope that our front office approaches such situations with the attention to nuance they require.