Select a player: A B C D E F G H I J K L M N O P Q R S T U V W Y Z

Defensive efficiency at the player level



Defensive efficiency (DER) was invented by Bill James. It is simply the proportion of balls in play that are turned into outs. I have always considered it as somewhat of a sanity check on defensive metrics. For example, if you have a metric which shows the fielders on a team combine to be 50 or 100 runs below average, but the team's defensive efficiency is better than average, you probably are doing something wrong. This is not always true. Ballparks have an effect on DER. The Rockies and Red Sox generally have worse DER in Coors field and Fenway Park than they do on the road. People in and around baseball have come to accept that the theories of defense indepenent pitching are mostly true, however it is not 100% true that pitchers have no effect on balls in play.

Stats like TotalZone generally track well with defensive efficiency. In 2020, the St. Louis Cardinals (.731) and the Los Angeles Dodgers (.729) had the top DER in baseball. Both teams also score well in TotalZone (+28 and +37) and Defensive Runs Saved (+33, +29). The Phillies were at the bottom in DER (.642) and rate poorly in both TZ (-41) and DRS (-33). On the other hand, the Pirates had a much better DRS (+26) than the Angels (-26), but the Angels actually had a slight edge in defensive efficiency (.698 to .697, both slightly above average).

TotalZone will not perfectly track with DER, even with a park adjustment, for several reasons. Popups and line drives fielded by an infielder are not counted in TZ. In addition, TZ has no fielding component for catchers and pitchers. These positions don't have a large responsibility for ball in play defense. Catchers are mostly responsible for catching popups and fielding squibbers and bunts in front of the plate. Pitchers field fewer balls than infielders in general, but in some cases their fielding can distort the defensive records of the fielders behind them. For example, in 2005 Mariano Rivera fielded more outs than his shortstop did during his time on the mound.

My implemenation of DER at the fielder level is not intended as a replacement or correction for the fielding metrics that are published on sites like Baseball-reference, Fangraphs, BaseballProspectus, or others. I did not park adjust anything, although I give the home and road ratings for people to see for themselves where a park adjustment might be needed. DER at the individual level does not make any adjustment for difficulty of play, so popups for infielders are included. If you add up the plays made and responsibility for hits for every fielder on the team, you will get to the team DER. I started out with a calculation similar to what is used in TotalZone. For example, a single on a ground ball fielded by the left fielder would be split between the SS and 3B. Since it's a ground ball, no outfielder had a chance to field it. Since it was hit hard enough to reach the outfield, we can assume the catcher and pitcher had no chance. The 1B is on the other side of the field. While in modern baseball it's possible that the 2B could have moved over on a shift, the responsibility mostly falls on the SS and 3B. Such an approach works well for 2003 to 2020, but before that information such as who fielded a ball or whether it was a ground ball or fly is not always there. What's worse is that such information is available more often for some teams than others, making it very difficult to fairly compare players in the same league.

In order to implement a rating that works for every team back to 1912 (Retrosheet now has play by play data going back that far) I had to ignore batted ball type and who fielded the ball on hits. Instead, partial hits are attributed to a fielder based on only two pieces of information: Type of hit, and the handedness of the batter who hit it. Doubles and triples will be assigned to outfielders at a higher rate than singles, and hits by LHB are more likely than those by RHB to be assigned to the 2B and 1B.

Here are definitions of the headings on player pages:

  • Outs - balls fielded by player. Includes proportion of balls turned into outs where information on the responsible fielder is missing
  • nonouts - Proportion of hits assigned to fielder. Also includes reached on error plays, and fielder's choices where no out is recorded
  • OAA - Outs above average. Each player is compared to the league average, separately for balls hit by LHB and RHB. This is outs, not run value of outs.
  • OAA_hm - OAA in home ballpark
  • OAA_rd - OAA in road ballpark
  • BIP - Balls in play
  • Plays_plus - Plays made above average. Think of this as a range factor done right, with the denominator BIP instead of total outs
  • GB_plus - for infielders only, Plays_plus on ground balls
  • Home - In home park. For infielders, this is GB_plus. For outfielders, Plays_plus
  • Road - In road park. For infielders, this is GB_plus. For outfielders, Plays_plus
  • vsLeft - Vs. left handed batters. For infielders, this is GB_plus. For outfielders, Plays_plus
  • vsRight - Vs. right handed batters. For infielders, this is GB_plus. For outfielders, Plays_plus
  • New, added with 2023 Retrosheet release

  • Arm_opps - Opportunities for an outfielder to throw out or hold a runner
  • Arm_runs - Runs saved by outfielders on baserunner advancement

With OAA, fielder ratings are available for pitchers and catchers. I am disappointed to see that Jim Kaat (16 gold gloves) does not rate highly in this metric, but most of the pitchers who won the most gold gloves do rate well here. Greg Maddux (+86, 18 gold gloves) does rate as the greatest fielding pitcher of all time.

Again, I don't think these metrics are superior to other published defensive metrics, especially Statcast Outs above average, which I think is the gold standard of the publicly available metrics. These might be of interest as metrics that use the same code, same methodology, for over 100 years of baseball history. Defensive efficiency at the player level, for better or worse, forces defensive ratings to add up at the team level and therefore might be of some interest as an accounting exercize.

This page was last modified 12/8/2023


All content Copyright © 2008-2023 by Sean Smith.