Most crazy splits are the result of statistical anomalies. If you're only using this season's worth of at bats, a few lucky or unlucky bounces is the only reason it's noticeable. If you look at splits for players with long careers you'll notice that very few have crazy splits. Maybe slightly, but definitely not more than .200 difference. I remember I had researched splits for certain players and I noticed stuff like splits significantly favored lefties for a hitter last season, and then I went back a year and it significantly favored righties so from that point on I stopped giving a shet about 99% of splits. The sample sizes are too often simply not large enough.
Grady Sizemore will have more value than Jason Bay regardless of draft position
Aramis Ramirez in 155 G will hit over .300 40 HR 110 RBIs
Brian McCann will have more value than Jorge Posada regardless of draft position