If we broke ZIPS into percentiles

Yeah, I was thinking about that – I made my Official 2009 Prediction in 2 parts, based on SP performance (which turned out to be near-irrelevant due to the trades, but…) – but the bottom line is that it’s silly to act as if one set of numbers can be a useful projection (most of the time. Once players hit ~30, you need to think in terms of injury likelihood, while with other players (say Jones) you need to allow for what happens in terms of playing time (is he strictly platooned, does he get bounced between 1B and RF, etc.).

Getting into percentiles is a bit much for casual fans/blog commenters, but I think that laying out your assumptions (I usually think in terms of best (likely) case, most likely, and then what a crappy season looks like). I should add that (IMO) percentiles should be viewed as presumptions, not standalone evaluations- iow, I don’t think you should say that Freddy had a 75th percentile season, but that PECOTA thought he was pretty likely to be worse (for whatever reason). Given that he was having a good but not amazing season, then limped through August and then was done, I think it shows that PECOTA was underestimating him, not that 2009 was actually a really good year for him.

Sanchez wasn’t hurt due to a single incident; it was exactly the kind of nagging injuries that grow more likely over the course of a season. IOW, what happened, injury-wise, is exactly what was likely to happen.

That said, if the injury had been either less or more, his final numbers probably look better – worse injury, and his numbers get frozen at late-July, which was pretty damn good; less injured, and his August average (probably) doesn’t plummet. As it was, he was healthy enough to affect his numbers, not healthy enough to play well.

Well, as I said, it really looks like he just split the difference between ‘08 and ’07. I’m sure it was more sophisticated than that, but that’s what the splits look like (well, actually they look like maybe a 60/40 weighting). I assume that it’s derived from formulas, not from player-by-player consideration. I agree that “taking into account” injuries is a recipe for goofy numbers (what would Freddy’s ’08 have looked like without all that injury? who could possibly tell?), but it seems equally absurd to pretend that the numbers reflect ability, not circumstance.

It’s funny to think that you’d adjust Bay’s numbers based on going from PNC to Fenway, but you wouldn’t adjust Jack’s numbers the year he’d had appendicitis (I don’t think that made him appendicitis-prone, did it?). It’s the old bugaboo of the statistician – if you can’t count it, it must not exist. Just ask UChicago economists about irrational economic behavior.