Plympton91 said:
There's a thread in the MLB forum where Jeff Passan deconstructs fWAR and I don't see how anyone could ever cite it with a straight face. Reading that article just reinforced my opinion that it is a low information context dependent stat like RBI or plus/minus ratings in hockey. There's no way it should be given equal weight to offensive stats. Doing so makes a mockery of serious analysis.
They absolutely should have traded for Fowler.
Fine, okay; I'm aware of the valuation issues. I literally just said that Fowler should have been acquired last season. But at a certain point a quantitative distinction becomes a qualitative one, and we have to ask if Fowler is really a CF. How do you propose comparing a player like Fowler (good bat, terrible glove) and Bradley?
Here's a thought experiment, leaving out extra base hits on both sides of the ledger for simplicity's sake:
Offense first: Fowler made 269 outs in 432 PA, posting a .372 OBP. Bradley's OBP of .285 would extrapolate to 309 outs in the same number of PA, but he actually had far fewer batting 9th. Here, it's easy: Fowler is ahead 40 outs offensively.
Fowler played 835 innings in CF this season. He made 198 putouts, of which 155 were judged to be in zone and 44 out of zone. He failed to make plays on 26 balls in zone. He made four assists. (I'm aware that there is a degree of zone bias.)
Bradley played 898 innings in CF this season, about 7% more. He made 283 putouts (45% more), of which 204 (31% more) were judged to be in zone and 77 (75% more) out of zone. He failed to make plays on 19 (27% fewer) balls in zone. He made 13 (225% more) assists.
Between the putouts and the assists, we're looking at a difference of 94 outs between Bradley and Fowler. Now some of this difference stems from opportunities — different pitching staffs, different corner outfielders, different positioning by coaches, different ballpark quirks — but some is likely predictive, and all of it really happened. Pitchers deserve a degree of credit for some of those outs, and corner outfielders some discredit: maybe Bradley catches some of those OOZ balls because Gomes is a lumbering doofus that a better LF catches, and that inflates Bradley's numbers; maybe Lester's cut fastballs were climbing bats to send cans of corn out to CF. But we watched the games, we know that he's made some plays this season that very, very few players make.
(We average these game states out using linear weights, which Passan objects to because they are context
independent, and which you object to — unless I've misunderstood you — because, like RBI, they are context
dependent.)
But I don't know how to assign credit for those extra outs. UZR's answer FWIW is that Bradley's range was worth 10 runs (or about 34 outs) above average, and Fowler's 12 runs (about 41 outs) below. So they say that the difference in CF was worth (34+41)/94=80%. I'm not sure I buy that it was that extreme, and I expect that you don't either. But neither is it credible that the difference is nil. What's the "real" difference, imagined in neutral circumstances? 30 outs? 20? 10?
I don't see how it makes sense to just say that the values are incommensurable and therefore we know nothing. Because the difference in the number of defensive outs between the two players is actually larger than the number of offensive outs and presumably symmetrically valuable. But within these bounds, it's wholly possible that Bradley had the better season of the two.