Pumpsie said:
If you're going to put Mookie in left, you lose all his defensive value, which is considerable and that's just not worth it. At that point, your best bet would be to trade him to someone else who could use him at second or center, and that would be a shame because he's going to be an All-Star for another team. Or, you can trade Pedey for pitching and put Mookie at second. But the last place he should be playing is left field at Fenway.
What about the wasted value of Castillo playing LF? Right now it's too early to use rate stats like UZR but Castillo has made 43 "out of zone" plays (per fangraphs) in 348 RF innings with 5 "in zone" plays not made. So 348 innings, +38 plays, an average of .109 positive plays per inning, basically one a game.
Jackie Bradley meanwhile has 159 innings in RF, has missed 2 in zone plays and made 13 out of zone, so +11, or .0691 per inning. Betts is at 77 innings in RF, missed one in-zone and made 5 out of zone, +6, .077 positive plays per inning.
Extending that to CF, Castillo has 10 OOZ, missed 3 in zone, and has played 130 total innings for a .0538 rate. Bradley is 103/23 out/in over 1213 innings, a .0659 rate, while Betts is 107/18 over 1266 innings, a 0.0703 rate.
Obviously this isn't a hard and fast analysis of who is the best defender, but there is one clear outlier in the early zone scoring sample and that is Castillo in RF. He's made a few bonehead plays to be sure but he's got a strong arm of his own and has CF range himself. We're discounting how good he is defensively because both Betts and Bradley cover (slightly) more ground and Bradley has a slightly stronger, definitively more accurate arm.
As for the concept of "lost value" in having Mookie play LF, Alex Gordon has consistently achieved dWAR values of 1.9, 1.3, and 2.6 from LF (on pace for another good one this year) while Lorenzo Cain has been putting up mid-2's dWAR alongside him. Bradley in 2014 had a dWAR of 2.0. Cain's UZR numbers parallel Bradley's quite well with similar >20 UZR/150 highs in partial seasons from 2012 and 2013 but with over 3200 innings played (a sample where UZR/150 gets to be more worthwhile) he's averaged out to a 15.3 UZR/150. Bradley's current average over his 1213 innings in center is 13.5 (drug down by two very small samples in 2013 and 2015).
So Gordon and Cain prove that two elite defenders in left and right can each still put up very large fielding values, even within the same year, without cannibalizing the other. There would be some missed opportunity on Betts' part because of Fenway's dimensions in particular, but he's the most naturally athletic and instinctive fielder in the bunch and a damn bright kid. He could develop an entirely unique style to playing the close confines of Fenway's LF that changes our perception on the position far into the future. For example, if the front office commits to that alignment and modifies the scoreboard with some padding Betts could play very shallow and rely on his speed going back to track down balls that land in the warning track, while stealing a ton of shallow hits that just make it over the infield. The only gap that would really open would be in deep left center where Bradley and Betts would likely see defensive overlap anyhow.
There's more than one way to assemble this OF, but I'd argue the last two things we should do when discussing it is: 1. discount Castillo's own defensive value and 2. let the reduced value of Fenway's LF weigh so heavily that it dictates a trade to resolve a non-problem in having three young, talented athletes with center fielder range. The Red Sox haven't had good production from all three OF positions since the end of 2013 and that required a five man assembly with three of the five rotating in LF and two of those having massive outlier seasons.