The Other Fifteen

Eighty-five percent of the f---in' world is working. The other fifteen come out here.

Another look at projection systems

Here is a color-coded chart of the projection forcasts from PECOTA, CHONE and ZiPS using either EqA or wBA (wOBA on the batting average scale). At the very least it gives you an idea of who some systems like more than others and, really, how close all of the systems are.

Derrek Lee0.3020.3160.3110.3100.311
Kosuke Fukudome0.3030.3040.2970.3010.303
Aramis Ramirez0.2930.3070.3030.3010.303
Alfonso Soriano0.2860.2960.2940.2920.294
Matt Murton0.2760.2850.2880.2830.285
Geovany Soto0.2740.2830.2860.2810.283
Daryle Ward0.2710.2820.2680.2740.271
Mark DeRosa0.2700.2730.2690.2710.270
Felix Pie0.2730.2650.2610.2660.265
Micah Hoffpaiur0.2570.2640.2700.2640.264
Mike Fontenot0.2670.2530.2630.2610.263
Eric Patterson0.2610.2570.2550.2580.257
Josh Kroeger0.2470.2540.2540.2520.254
Ronny Cedeno0.2510.2640.2490.2550.251
Sam Fuld0.2430.2450.2530.2470.245
Ryan Theriot0.2370.2490.2420.2430.242
Henry Blanco0.2040.2400.2300.2250.230

Labels: , ,

3 Responses to “Another look at projection systems”

  1. # Anonymous DeRoMyHero


    Maybe you could post a primer sometime for those of us who are statistically challanged. How many years of data go into these projections? How much minor league data? My eyes told me that Felix Pie was only slightly better as a ML hitter than Jason Marquis last year. (Actually, Marquis has a better swing.) His ML stats weren't pretty. Yet he is predicted to hit .266? I hope these guys are correct, but I sure wouldn't put any money on it.

    They also seem to be mixing in DeRo's pre-Jaramillo stats. (Obviously, computers can't tell the difference.) His last two years as an everyday player are probably much more reasonable estimates of what he will do in 2008 than his years as a "utility guy" getting 300 ABs a season.

    I realize that these are projections -- which is really a polite way of saying "crap shoot".  

  2. # Blogger Colin Wyers

    DMH: I'm actually working on a lengthy explanation of exactly how these projections work; I meant to have it done a lot earlier but I got sidetracked.

    The short version: Generally at least three years of data, and information beyond that isn't given a whole lot of weight. Generally as much minor league data as is available is used, although obviously you have to perform some wizardry to make it relevant to the majors. (You don't just look at a guy hitting .300 and use that to say that he's going to hit .300 in the majors.)

    In the specific case of Mark DeRosa - a .260 EqA/wBA is right about league average, and that's right about where DeRosa has been the past two seasons. So I don't know if that forcast is really out of line with what you would expect just looking at his stats for the past few seasons and making an educated guess.

    I should note that these aren't actual batting average, but a rate version of Linear Weights that are supposed to LOOK like batting average.  

  3. # Anonymous DeRoMyHero

    OK, I thought they were supposed to be equivalent to BAs, which I would disagree that DeRo would hit only .26x -- but if it shows him at league average, that sounds about right. I'm just not convinced on Pie, however. That is one ugly swing...  

Post a Comment

Links to this post

Create a Link