David Pinto has begun to release data for PMR in 2007, starting with 2007 team rankings. As you'd expect from my and others' work on Reds defense, the Reds fared poorly: they ranked 4th from the bottom in the National League, ahead of only the Marlins, Brewers, and Pirates. Pinto estimates that they made 28 fewer outs than expected, which is the equivalent of ~22 runs. That means that the Reds' defense alone cost them about two wins.
I'm surprised to see the Brewers that low, though Fielder, Braun, and Weeks must make for rough infield defense.
----
It hasn't shown in posts over the past week, but I've been diligently working on the player value series. Expect a piece on evaluating catchers in the next day or two, with the position player wrap-up shortly thereafter. Both are largely finished, just waiting of a few small details. FWIW, my own work has the Reds at about 14 runs below average overall.
As a side note, I've learned a ton of Excel in the past several weeks, which has saved an enormous amount of time. But the result of all my concatenating, pivot tableing, subtotaling, and vlookuping is that my 13.5 mb 2007 fielding spreadsheet is getting Darn Slow. ;) It's pretty neat, though, when you finally get five or so independent datasets all talking to each other without some handy commonality like a lahman id.
I look forward to the post on catchers' defense (as well as the other stuff). I know Dial or someone calculated it on BTF a little while back. I tried it and got it down for the most part, but I couldn't get the total above average to equal 0. I assumed it's because the runs above/below average are from a different year, but anyway I didn't "publish" it. It'll be interesting to see how you tackle it.
ReplyDeleteThanks for the tip, as I hadn't thought to search BTF for this kind of thing. Chone Smith actually did something that looks awfully similar to what I came up with:
ReplyDeletehttp://www.baseballthinkfactory.org/files/primate_studies/discussion/tweaking_zone_rating/
My approach more explicitly works in a comparison to average within the dataset (yay for pivot tables), so I like it a bit more...but it's more or less the same. I basically just compare CS's, WP+PB's, and error rates among catchers and then assign run values to them. I don't correct for pitcher handedness because I haven't figured out an efficient way to do that yet, unfortunately.
-j