clock menu more-arrow no yes

Filed under:

Fun With Pitcher Projections

New, 10 comments

Since this is essentially what we're going to be trying to do for the next four months, I might as well get the ball rolling. This one's a pretty simple system, based on elementary statistics, in which I estimate a range of possible 2006 FIP's by assuming that nothing about the pitcher's ability will actually change. By this I mean that I'm treating each pitcher's 2005 strikeout and walk rates as his "true performance level," and then re-calculating his FIP with his rates reduced or increased by one standard deviation.

(For those curious, the standard deviation can be estimated by sqrt(pq/n), where p = probability of something, q = 1-p, and n = the sample size. So, for example, Gil Meche struck out 13% of the batters he faced last year, so p = 0.13, q = 0.87, and stdev = 0.013 (1.3%)).

Hernandez: 2.51-3.30 FIP
Washburn: 4.13-4.67 (not park adjusted)
Moyer: 4.20-4.69
Pineiro: 4.25-4.76
Meche: 4.78-5.41

Now let's run the projections again, this time assuming that each pitcher's home run rates regress towards the mean (which is that 11% of all outfield flies go for homers):

Hernandez: 2.35-3.15 xFIP
Washburn: 4.28-4.81 (not park adjusted)
Moyer: 4.40-4.88
Pineiro: 4.11-4.62
Meche: 4.78-5.41

What if we take this a step further and attempt to estimate ERA's? It's a little more hazy, and the FIP numbers will be more reliable, but it's worth a shot. For example, Joel Pineiro's career ERA is 4.11 while his career FIP is 4.18, so we'll knock his estimated range down by 1.7%. Same procedure repeated for all of them except Felix, who we'll just assume for the sake of simplicity stays the same:

Hernandez: 2.35-3.15 ERA
Washburn: 3.70-4.15 (not park adjusted)
Moyer: 4.16-4.61
Pineiro: 4.04-4.54
Meche: 4.59-5.20

Although it would be the same process, I didn't bother doing this for the relievers, since their smaller sample sizes last year result in much wider ERA ranges.

So, there you go - five preliminary ERA projections for the Mariners' 2006 starting rotation. Note that Joel Pineiro appears in line to improve rather significantly.

What these projections don't take into consideration is the possibility of a real (read: statistically significant) improvement or decline by each pitcher. Standard deviations are just used to estimate the amount of noise around a mean you can expect given a consistent level of performance. If Gil Meche strikes out 13.5% of the batters he faces next year, that won't represent a significant improvement, because it's within one standard deviation of last year's strikeout rate. If he comes out and posts an xFIP in the low-4's, though, then you know he's made a real improvement, because that would be out of line with even the best-case scenarios from last year's numbers.

It's also worth noting that, although outfield flies leave the park 11% of the time over a long enough period of time, single-season rates can fluctuate like anything else based on a limited sample, so one of these guys' ERAs might fall outside of his projected range even though he didn't pitch any better or worse than expected. That's just the nature of the beast, and I can't really do anything about it.

Enjoy. Doing this with hitters is a lot easier, so I might get around to that tomorrow.