The U Files #34: Errors of Defensive Evaluation

It was well documented that the Mets had a terrible defense last year. As proof, it was offered that the Mets led the Major Leagues with 144 errors. While the experts may have come to the right conclusion, the implicit assertion that errors are the way to measure team defense is incomplete at best. So, just how bad was the Mets defense last year? We need to trek through some more complex statistics to find out.

Errors and fielding percentage are likely popular for one reason: they're easy. This is the same reason home runs and batting average are also popular. It is increasingly widely known that batting average is a limited statistic. In telling how often a player gets a hit per at bat, it leaves out more than it makes clear. It ignores walks, and says nothing about how often a player reaches base, a much more vital concern than pure hitting ability. It ignores also the matter of how many (extra or total) bases the batter picked up. Home runs fill in as a measure of power, but not adequately so. A home run hitter can be less valuable then he seems if he hits few doubles, or a player can be underrated who is a doubles machine.

Similarly, errors and fielding percentage are limited measures of fielding. The principal concern of a team defense is to get to balls in play and convert them into outs, yet the most popular measure of fielding completely ignores this issue. There is indeed an adage about a player who makes a lot of errors because he gets to more balls than his peers, and on the flip side the player who makes few errors because he reaches few hard to play balls. You will occasionally hear this when watching a ball game on TV, but when the time comes to measure team defense, the only stat used will be errors.

Just as On Base Percentage and Slugging Percentage and more advanced measures based on on-base and slugging factors are available, so are there measures that fill in the measure of fielding. None is as well known as OBP or SLG, however. The two that can be found on mainstream sites like ESPN.com are Range Factor and Zone Rating.

Range Factor, RF, will be completely and utterly useless in our discussion of the Mets defense, because it is strictly a player's statistic. It is a measure of how many outs a player records per defensive game (nine innings in the field). It has flaws in that it ignores factors such as the number of balls put in play against a team (which varies as the inverse as the team's strikeout rate), the handedness of the team's pitching staff, and the ground/fly ball tendencies of the pitching staff. It fails as a team measure because every team records exactly 27 outs per nine innings, a fact built into the very rules of the game.

Zone Rating is a more mysterious measure created by STATS, inc. It apparently is the measure of how many balls in a player's "zone" that he reaches. Applied to players, it has a flaw in not considering how many balls the adjacent fielders take away. In assigning a zone to each position, ZR calculations leave out certain "no mans land" parts of the field where no fielder is "expected" to make a play, such as the short outfield where bloops fall in. Nevertheless, different teams may allow different numbers of balls into hard-to-reach zones, leaving zone rating inaccurate as a measure of team defense.

Defensive Efficiency, DER, is a statistic created by Baseball Prospectus that measures the percentage of balls put in play against a team that are fielded. It is the inverse of a team's hit rate per balls in play against. It does not account for how difficult the balls the team gave up were to field, which depends on factors like the speed of the batted ball and the placement of balls in play. The latter is a function mostly of the batters faced by the team; pitchers do not control exactly where a ball is hit. The Mets had placed sixth in the NL in DER with .7141. The league average was .7120, but Shea Stadium raises DER by about .005 points. The Mets park neutral DER would then be about .7091. Adjusting DER to account for errors, the Mets fall to ninth in the NL at .6909. The league average error adjusted DER was .6945. This statistic is not park adjusted, so the Mets true ranking should be lower than ninth.

Team

Adjusted DER

Los Angeles Dodgers

.7158

Atlanta Braves

.7117

San Francisco Giants

.7051

Philadelphia Phillies

.7043

St. Louis Cardinals

.7042

Arizona Diamondbacks

.6964

Milwaukee Brewers

.6949

Pittsburgh Pirates

.6922

New York Mets

.6909

Cincinnati Reds

.6891

Colorado Rockies

.6887

Florida Marlins

.6870

Houston Astros

.6866

Chicago Cubs

.6854

Montreal Expos

.6845

San Diego Padres

.6750



Team

ZR

Adjusted DER Rank

Atlanta Braves

.886

1

Milwaukee Brewers

.882

7

St. Louis Cardinals

.880

5

Florida Marlins

.876

12

Cincinnati Reds

.868

10

Colorado Rockies

.867

11

Philadelphia Phillies

.866

4

Houston Astros

.864

13

San Francisco Giants

.863

3

Pittsburgh Pirates

.862

8

San Diego Padres

.861

16

Arizona Diamondbacks

.857

6

Montreal Expos

.856

15

New York Mets

.853

9

Chicago Cubs

.841

14



Differences in the rankings for ZR and DER can be attributed to teams allowing more or fewer balls in hard to reach zones. The Mets ranked much higher in DER than in ZR, so they must have allowed few balls hit into tough spots (raising their DER). Park factors affect DER but not ZR to as great a degree. Park influences would tend to inflate the Mets DER, so their true defense lies below this number.

Ultimate Zone Rating, UZR, is a largely unknown statistic that adjusts ZR to account for some of it's failings. UZR has recently been reconfigured by Mitchel Lichtman of BaseballPrimer. It fills in the gaps in the field left open by ZR, and adjusts ZR to account for the difficulty of fielding balls hit to a particular sub-zone, and accounts for park factors, the speed of the batted ball, the handedness of the batter (which affects positioning), the ground/fly ball tendencies of the pitchers (ground ball staffs allow fly balls that are more difficult to field and vice versa), and baserunner/outs situations. It takes into account a players errors. For a full description of the statistic, see parts one and two of this article at BaseballPrimer.com.

Unlike ZR, it does not leave no-mans lands, and does not treat all chances to a zone equally. Unlike DER, it accounts for park effects and ball placement, and does not treat all chances equally. Thus it is a better statistic than either.

Here is a breakdown of the UZR runs above or below average for the Mets in 2002:

Player

UZR Runs

DP/Arm Runs

Edgardo Alfonzo

-5

-1

Roberto Alomar

-9

-1

Jeromy Burnitz

-14

-2

Roger Cedeno

-15

-1

Joe McEwing

3

1

Rey Ordonez

4

0

Timo Perez

7

0

John Valentin

-3

0

Mo Vaughn

-15

0

Total

-47

-4



In total, the Mets defense cost the team 51 runs compared to what average fielding would have done. UZR is not counted for catchers. Mike Piazza's defense cost the Mets an additional 19 runs. In total the Mets defense cost the team 70 runs. This is enough to increase the Mets RA .4367 runs from 3.8678 (the Mets defense neutral RA) to 4.3045. In other terms, the Mets defensive play cost the team seven wins.

Go here to download a spreadsheet with UZR totals from 1999 to 2002. Or, look for the numbers to be posted at baseballprimer.

Team

UZR Runs

DP/Arm Runs

Total Runs

Errors

St. Louis Cardinals

59

6

65

103

Colorado Rockies

38

-5

33

112

San Francisco Giants

21

2

23

90

Los Angeles Dodgers

27

-5

22

90

Atlanta Braves

18

-5

13

114

Cincinnati Reds

4

5

9

120

Chicago Cubs

0

5

5

114

Milwaukee Brewers

10

-6

4

103

Montreal Expos

-9

2

-7

139

Florida Marlins

-15

1

-14

106

Philadelphia Phillies

-13

-3

-16

88

Houston Astros

-24

5

-19

83

San Diego Padres

-19

-1

-20

128

Pittsburgh Pirates

-31

9

-22

115

Arizona Diamondbacks

-13

-11

-24

89

New York Mets

-47

-4

-51

144



The Mets were indeed the worst defensive team in the National League. It was not simply because they made more errors than any other team, but because they didn't catch balls average fielders would have caught, and because Mike Piazza can't throw to save his life. Note the error totals compared to the team rankings in UZR runs above. Note also that the variance (range) in UZR runs is almost twice that of errors.



Do you have an opinion on the Mets? Be sure to let us know on the message board. NYfansonly.com is always looking for die-hard Mets fans who would like to be writers for the site. Click here to learn more on how to become a Mets beat writer for NYfansonly.com.


Amazin Clubhouse Top Stories