First, here's my work, so to speak. It's not supposed to be self-evident as I plan on breaking down what I did. Anyway, after all that math, I end up with a projection of .241/.311/.415 for Josh in '09. So what did I do?
What I really wanted to look at was Fields' true talent from 2006-2008. His 2007 was perhaps the only bright spot outside of Buehrle's no-no and, it seemed, bode well for a Crede-less 2008. Those 23 HR and the accompanying ISO had me fairly well convinced that despite his lack of glove and love of strikeouts, he could be slightly below average at 3B in 2008 and that we'd forget all about Joe Crede as soon as Kenny could let him loose. Then the offseason and the actual season happened and Fields was more or less forgotten and ended up battling injuries to the point that he didn't even break 300 AB in Charlotte. According to my 3B FA post, it's pretty clear I don't think that much of Fields at this point, but I skipped around actually quantifying his value.
Since then, Chone Smith released his 2008 defensive projections. Fields rates nearly a full win, -9 runs, below average. Remember that the replacement level player plays average defense, so replacement level and average are functionally equal here. So that's half of his 2009. What about the bat?
That's what all the math was about. I started with his raw stats from 2006-2008 and used the MLE calculator to get everything normalized to a neutral MLB setting. That's the first set of data on the google doc. The next thing was to get a coherent season out of 2007 between the Charlotte and Chicago parts. I took an average weighted by ABs and got his 2007. Next, I regressed each part of the slash line from each season to the AL 3 year mean. To show specifically, I used Actual*[(ABs)/550] + Mean*[(550-ABs)/550]. "Actual" = whatever part of the slash line I was using. So after all that, we have a regressed true talent slash line for 2006-2008.
The only thing left to do to make a Marcels' projection would be to weight each season (such that the more recent the season was, the more weight it receives) and then tack on an age component. The only thing I don't have is the age component, which may or may not be critical. 2006-2008 corresponded to Fields' 23-25 age seasons, so he stands to improve, on average, up until age 27. My projection, then, assumes that he's going to plateau at age 25 without even getting incrementally better through age 27. The .241/.311/.415 line completely neglects upside. You might think of it as the 50th percentile projection rather than the weighted mean, to use PECOTA terms. Or even maybe the 40th, depending on how high on Fields (of grass) you are.
Translate that to wOBA and it's around .320, which comes about to about 10 runs below average for 600 PA. All told, that's .5 wins above replacement, or 2 wins below average. What if we throw some upside back into it? What if he manages a .260/.330/.490 line, as per the old community projection? About .350 wOBA, which gets him all the way to -.5 wins above average.
Lastly: this is me doing math. We all know how well that turns out. So, obviously, there may be some glaring errors that potentially invalidate the specifics of what I've done here. But no matter! What's intriguing to me about this is the method. Front offices do the same thing every day in their evaluation of players from the Rule V draft to CC's future deal. With the vast array of resources baseball-crazed webfolks have made available, and for free no less, we can do the same damn thing. Pretty cool, I say.