skyking162

Ok, let’s pretend you’re convinced that fielding really is uber-important to winning baseball games. And you’ve realized that traditional stats like errors and assists just aren’t worth your time. Where can you go to find good fielding data? After all, mainstream websites certainly don’t have it, Baseball Prospectus’ FRAA isn’t based on play-by-play data, and The Hardball Times only has zone rating data back through 2004.

Thankfully, there are some industrious individuals who have done the heavy lifting. Here’s where you can go, depending on the time frame and quality demanded:

1956 through 1986

Check out Sean Smith’s TotalZone system, which uses play-by-play data from retrosheet. Hit location data isn’t precise or complete for many of these seasons, but Sean came up with a brilliant way to address that problem. While this approach isn’t as accurate as other methods for a single season, we have the benefit of looking at these players’ entire careers, hopefully removing any flukiness from specific seasons.

1987 through 2007

You’re going to want to use a zone rating system here, and SG over at Replacement Level Yankees Weblog has compiled that data from the past 21 seasons. From 1987 to 2001, data is pulled from ESPN and chances are estimated based on Chris Dial’s research. From 2002 through 2007, the exact zone rating data from STATS is used. (I’ve been using combined STATS and BIS zone data for my 2007 player values. Sean Smith has the same data in a nicely formatted file.)

2000 through 2006

Ultimate Zone Rating is probably the best publicly available fielding measure. Well, at least for the seasons that MGL has made it available. UZR starts with a basic zone rating, but also adjusts for batter handedness, pitcher handedness, ballpark, and a number of other potential biases. It also doesn’t treat every ball hit into a fielder’s zone the same — fielders get more credit for hard-to-field balls and less credit for easy plays. There are two files: 2000 through 2003 and 2003 through 2006 (with partial 2007 data).

Dave Pinto’s PMR system is worth paying attention to, but nobody’s compiled all the data in a single place or converted the data into runs. Hopefully we can convince Dave to do that one of these days.

2008

Sean Smith has the only fielding projections I know of, using zone ratings from previous seasons to project 2008 performance. He knows what he’s doing, even using speed scores to help the outfielder regressions. There are two files: infielders and outfielders.

Catchers 1957 through 2006

Catchers are a whole different beast, because instead of fielding batted balls, they prevent wild pitches and help control the running game. Sean Smith has career data available, in a per-season format.

Outfield Throwing Arms 2004 through 2007

Most outfielder ratings only measure range, so you’ll need to sprinkle in some information on throwing arms. John Walsh has done some great research in this area, not just measuring assists, but how valuable the reputation of an outfielder’s arm is in keeping runners from taking an extra base.

First Basemen Scooping Ability 1985 through 2007

One aspect of a first baseman’s fielding skillset that’s almost never measured is his ability to corral errant throws, either by scooping bounced throws, stretching to the side, or being tall enough to reach high throws. Sean Smith has done some simple work on the careers of first basemen from 1985 through 2007. The methodology is explained at his blog.

Scouting Data 2003 through 2007

Do you want to know what a specific player’s strengths and weaknesses in the field are? Check out TangoTiger’s Fans Scouting Report, conducted yearly since 2003. Hopefully he’ll have each season’s data converted to runs in the near future.

There you go. Now, whenever you want to rate any position player after 1957, you know where to find the fielding data. Sure, fielding analysis isn’t perfect, but what’s available is a much better option that assuming everybody is average, counting up Gold Glove awards, or asking Steve Phillips. Oh, and remember to apply a positional adjustment when comparing two different positions.

Popularity: 9% [?]