July 3, 2007
Measuring Team Defense
This week, we're going to take a look at a controversial issue from its least controversial angle. The topic? Defense--a subject that, when we approach it statistically, tends to cause fistfights and sour feelings.
Why is defense such a problem? Because it's hard to reconcile what's actually happening on the field with what we see, and it's even harder to assign responsibility for defense among the team's fielders. The first problem is the one that draws the big emotional responses. We tend to grade fielding based on how difficult and/or graceful the plays look. So when we see Derek Jeter unleash one of his patented jump-throws in Friday's game against the A's, and it creates an immediate impression of elite defense. The play is spectacular, acrobatic, and it takes a single away from the batter.
Not to denigrate the accomplishment of making a difficult play deep in the hole, but it's worth noting that we don't place such an emphasis on aesthetics when a player is at bat. At the plate, we care much more about the end result--single or groundout, homer or fly ball--than we do about how stylish the batter looks while getting there. There's also the issue of whether defensive stylishness actually adds anything to the quality of play. Often when players make spectacular dives or leaping grabs, it's because they had a slow first step, or were badly positioned to make the play. Comparing this to hitting again, it's a lot harder to hit a pitch in the dirt or way off the plate than it is to hit one that's down the middle. A handful of times a week, you'll see a player successfully golf a slider off his shoe tops for a home run, or slash a pitch that's headed for the opposite batters' box down the line for a double. Sometimes the guy with that hacktastic hitting style is Vlad Guerrero, sometimes he's Shea Hillenbrand, and you need better tools than just a highlight reel of their most difficult hits to discern the vast difference between those two batters.
The other issue, once you get past the aesthetics, is how to break down credit for the team's defense between the pitchers and the fielders, and among the individual fielders. Unlike offense, defensive opportunities are not doled out on a rotating basis, and those opportunities are not all created equal. Maybe that nifty throw Jeter made was on a ball that the third baseman should have made. Should the guy at the hot corner get a demerit because the shortstop fielded the ball? Do we give extra credit to a center fielder like Ichiro Suzuki, who spends most of his time bracketed by a pair of fairly immobile cornermen?
One way that we can gloss over some of the issue of assigning individual players credit is by looking at the defensive performance of a team as a unit. That way, we don't care whether it was Ichiro's responsibility to field a fly ball to left center or Raul Ibanez's, we just worry about whether the team converts that fly into an out. The most basic tool to assess defense in this manner is Defensive Efficiency Rating, also abbreviated DER or Deff-Eff. If you remember our conversation about batting average on balls in play (BABIP) a few weeks ago, DER is, loosely, the inverse of BABIP--it simply measures how often a team's defense converts a ball in play into an out.
As a statistic, DER doesn't possess much nuance-it doesn't track double plays, assists, or errors. However, it's a good, simple way to see if a team is doing its job afield. A number of reports on the stat page keep track of DER, including the Team Defensive Efficiency report, which keeps track of the overall team numbers, and the Pitcher Defensive Efficiency report, which tracks the balls in play information for each pitcher separately (however, this report is not updated during the season, so you can't see this year's numbers yet).
One improvement that has been made to DER is adjustment for ballpark, something BP alumnus James Click incorporated in a statistic innovatively named Park Adjusted Defensive Efficiency (PADE). PADE takes a team's DER, compares it to the MLB average, and then applies an adjustment factor based on the team's ballpark's effect on batting average. This helps account for the fact that it's much more difficult for a team to maintain a good DER in Boston, where there is next to no foul territory and the Green Monster in left field, than in a more spacious park, like you have in Oakland or San Diego. This gives us a scale where the league-average defense is the zero point, and a positive or negative number represents the percentage of balls in play converted into outs above or below the MLB average.
Team Year DER PADE FRAA ---- ----- ------ ------ ------ SDN 2006 .7130 2.81 22 DET 2006 .7020 2.10 35 HOU 2006 .7011 1.99 22 NYN 2006 .7035 1.31 9 SFN 2006 .7009 1.19 -8 ANA 2006 .6916 1.05 9 CHN 2006 .7011 0.65 -13 ARI 2006 .6850 0.23 1 ATL 2006 .6867 -0.01 -7 COL 2006 .6838 -0.14 36 CHA 2006 .6942 -0.14 12 FLO 2006 .6858 -0.36 -8 NYA 2006 .6949 -0.64 19 OAK 2006 .6894 -0.66 -9 PHI 2006 .6817 -0.81 -3 SLN 2006 .6972 -0.91 12 LAN 2006 .6812 -0.95 6 TOR 2006 .6933 -1.00 17 MIL 2006 .6874 -1.01 -17 BOS 2006 .6801 -1.18 -17 KCA 2006 .6764 -1.39 -9 WAS 2006 .6918 -1.59 -27 CIN 2006 .6818 -1.64 -23 SEA 2006 .6895 -1.70 -6 MIN 2006 .6862 -2.13 16 TEX 2006 .6802 -2.16 17 BAL 2006 .6802 -2.57 -18 PIT 2006 .6745 -2.71 -13 CLE 2006 .6754 -2.74 -10 TBA 2006 .6716 -4.46 -55
As you can probably guess based on the chart, FRAA is the third team defensive metric I'll introduce you to today. You might remember a brief discussion of Clay Davenport's Defensive Translations, in the second edition of Toolbox. Basically, FRAA is a component-based defensive measure using basic game information we have throughout the game's history: putouts, assists, double plays, ballpark factor, and the handedness of a team's pitchers. The DT system scores each player's defensive performance, but some of you might not realize that it also keeps track of team totals. This information is not available as a sortable report on the Statistics page, but we can cull it out of the Davenport Card Team pages, which are maintained for every team in our database going back to the 19th Century. The easiest way to access a team page is through the team audit toggle at the top of this page-you pick your team, and the first link on the audit page will be the team's 2007 "DT Card listing." To get prior years, you either have to play with the URL in your browser, or navigate using the links on the individual player cards. Each player DT card carries on the left-hand side links to each team on which the player received time in the major leagues.
With a little elbow grease, by looking at the team pages you can produce a current defensive ranking of teams by FRAA:
Year Team DEF-EFF FRAA ---- ----- --------- ------ 2007 NYN .734 34 2007 COL .696 26 2007 CHN .721 25 2007 NYA .711 23 2007 ANA .709 23 2007 OAK .719 16 2007 PHI .692 10 2007 SDN .714 9 2007 BAL .713 8 2007 BOS .710 8 2007 MIN .699 7 2007 DET .699 4 2007 CLE .692 4 2007 WAS .706 2 2007 PIT .684 2 2007 SFN .715 1 2007 TOR .713 1 2007 MIL .699 -2 2007 CHA .701 -3 2007 KCA .690 -3 2007 ATL .697 -5 2007 SLN .707 -6 2007 LAN .697 -7 2007 HOU .692 -9 2007 ARI .694 -12 2007 TEX .688 -15 2007 CIN .681 -21 2007 SEA .679 -22 2007 FLO .681 -29 2007 TBA .666 -40
So that's what we have for this week. If you have any questions, or topics you want me to cover in this space, you can send me a message using the links at the bottom of the page.
James Click, "Prospectus Basics-Defense": A slightly more technical overview of defensive metrics.