Defensive Runs

This page contains all the various Defensive Runs (DR) files I have generated. DR is a way of calculating how many runs above average a given Major League Baseball player is worth, defensively. It uses Sherri Nichols's Defensive Average (DA) numbers and Palmer and Thorn's Linear Weights to value each event.

Here is a description of the DR method (plus the meaning of some of the header abbreviations in the long forms),and here are some comments and qualifications I have about the DRs.

Many of the data files which follow have not been put into HTML format yet and are just regular ASCII text files. However, you can use simple UNIX tools like grep and sort to do useful formatting - for example, putting together a list of all CFs ordered by average DR/G - which is nice when you are writing an email or rec.sports.baseball article.

These are the important final quantities created by this analysis. Remember that these are runs allowed, so negative numbers indicate better than average fielding:

The player names are given as "first_initial. lastname". This causes a problem for John and Jose Valentin, since they play the same position. So I have changed Jose's first initial to "X" in these files; I figure John was here first and has seniority.

Short Form

These files contain just the Defensive Runs information for every player from 1988 to 1995. They are ordered by position, player, and year.

Long Form

These files contain DA, DR, and auxilliary information, such as the number of extra base hits allowed or double plays made above average, pro-rated to a full season. There are also statistics for regular players and extras. Each year and league has its own file; players are ordered by position and then DA.

Mean DR over many seasons

These files list Defensive Runs per game for each season, plus an average DR/G over that player's career (since 1988), and an average Total Defensive Runs for a typical full season given that average DR/G. (I assume that a full season is 150 games played, so the column is just 150 times the column). Combining many seasons in this way cuts down the error from sample size, and helps balance out some biases due to pitchers or neighboring fielders. The players are ordered by position, DR/G, and name.

(If your newsreader can't handle tables, or if you want simple ASCII files to do greps and sorts with, click here: AL, NL. Unfortunately, these files are more than 80 columns wide, but I couldn't compress eight seasons worth DA data any further).

Tim Shippert, shippert@alumni.caltech.edu
December 2, 1997