Stats stuff - Original-ish content!

By Jason Wojciechowski on April 7, 2005 at 7:27 PM

I've spent the last few weeks writing and tweaking a script that will add a new bit of interest to this site. I got interested in using play-by-play data to determine the cumulative run values of players' actions on the field ever since I first saw a run expectancy table (like this one at Baseball Prospectus [Premium Subscription required], though I think I saw my first one in Earnshaw Cook's book).

With nice computer-readable play-by-play data not readily available for free, however (damn the baseball community's lack of Open Source spirit), I can't create these stats on a large-scale basis. What I can do, though, is create these stats for the A's. I have to enter the data by hand from Internet-published play-by-play accounts of the game, but that's the point of my script: it cuts down an arduous task to one that takes just a few minutes for any game.

To find out more about my method, see this paper by Jim Albert. It's the most basic method possible, and it doesn't adjust for park or position or anything else, but I see it as, essentially, a toy anyway. It's not predictive, it's largely situation-based, and it's not especially elegant, but it is fun to look at and can be useful for certain situations: relief pitching, for instance.

A probability-based stat is more elegant and, in my mind, more useful for relief pitcher evaluation, but, because I largely conceived of this project being only for hitters (it will include pitchers, though), I went with the runs-based model.

For each game, I'll update the stats and post the MVP and LVP, ideally along with the usual witty and insightful (and sarcastic, particularly toward myself) commentary (all three of) you are used to.

Just to give you a taste, the offensive MVP of last night's 9-0 win over the Orioles was, unsurprisingly, Nick Swisher, with 2.902 ARC (Actual Runs Created?). The offensive LVP was, on a night when no one was very bad, Erubiel Durazo, with -0.6225 ARC.


Don't forget about my other relief pitcher project, because I haven't. I'll be updating that every day as well, though to save bandwidth, I'll hide the images and force you to click to get them if you want to see, for example, the running average of run differential for Huston Street so far.

A taste of that as well: Octavio Dotel figures, by PECOTA's VORP/IP projection, to be the most valuable of the five A's who've pitched out of the bullpen so far, is also the pitcher who's thrown in the most meaningless game, coming in for Oakland last night when the score was already 9-0.