Dunn’s low BaBIP can also be attributed to him hitting into the shift so often. The Dunn shift has taken away at least 20 would be hits if fielders were in there normal position. One could say that he should try hitting the ball the other way more often but that’s really not his game and would probably take away from his home runs. I can deal with the low AVG as long as he keeps hitting 40-50 HRs a year and walking 100+ times which I don’t think think are unrealistic expectations for Dunn.
Don’t forget that it should also take into account bunting tendencies, IFFB% (and really HR/FB% with HR distances to estimate if some HR should have been in-play outs, and vice-versa), not to mention whether a player is likely to face a shift and hit “in-play” balls off the Green Monster or the like that will always yield a 1.000 BABIP.
I have to image this sort of thing is what in-house stat guys have built for MLB teams… but even better, since I’m sure there are publicly-available factors I’ve missed in addition to proprietary information they have at their disposal.
So it’d be a huge mess to do properly, but it seems like a good xBABIP model that would output to every FanGraphs player page (or whatever site) would be a huge hit — I’m really surprised it hasn’t happened already.