Printer-friendly

October 27, 2003

A TAD Here or There

A Closer Look at Team Defense

by James Click

Over the last two articles, I've looked at various methods for removing some of the complicating factors when looking at team defense. Based on the idea that team defensive metrics were really a measure of three separate factors (park, pitching, and actual defense), we determined one way to remove park factors (PADE: Park Adjusted Defensive Efficiency) and another to remove pitching factors (PIDE: Pitching Independent Defensive Efficiency). By removing these outside influences in our defensive metrics, we've leveled the playing field, allowing us to better judge which teams have the best team defense, based simply on the percentage of balls in play that they convert into outs.

With both PADE and PIDE, we removed one factor, but not both. We were able to see either how a pitching staff and defense together looked compared to the league or how a defense and park looked against the league. What we did not have was one metric that simply measured defense versus defense, our ultimate goal.

To produce that metric, we have to move back a step. Rather than looking at the final result of PADE and PIDE, we need to look at their baselines--the expected performance of the defense based on either the pitching or park. These baselines were included in the PIDE column as Team Expected In Play Average (tExIPAvg), but were not included in the PADE article. Here are both baselines, side by side:

Team                   BasePIDE       BasePADE
Anaheim Angels           1.0026         0.9992
Arizona Diamondbacks     1.0087         0.9893
Atlanta Braves           0.9917         1.0042
Baltimore Orioles        1.0113         1.0093
Boston Red Sox           1.0085         0.9905
Chicago Cubs             0.9924         1.0062
Chicago White Sox        1.0045         1.0049
Cincinnati Reds          0.9939         1.0051
Cleveland Indians        1.0007         0.9990
Colorado Rockies         1.0110         0.9794
Detroit Tigers           0.9883         1.0028
Florida Marlins          0.9989         1.0025
Houston Astros           0.9786         0.9951
Kansas City Royals       1.0009         0.9886
Los Angeles Dodgers      0.9958         1.0127
Milwaukee Brewers        0.9953         1.0072
Minnesota Twins          1.0055         0.9942
Montreal Expos           0.9886         0.9999
New York Mets            0.9988         1.0064
New York Yankees         0.9952         1.0082
Oakland Athletics        1.0111         1.0074
Philadelphia Phillies    0.9781         1.0117
Pittsburgh Pirates       0.9905         0.9988
San Diego Padres         1.0004         1.0076
San Francisco Giants     1.0097         0.9987
Seattle Mariners         1.0310         1.0104
St Louis Cardinals       0.9971         1.0020
Tampa Bay Devil Rays     1.0209         0.9961
Texas Rangers            0.9927         0.9910
Toronto Blue Jays        0.9972         0.9963

As before, the higher the number, the easier things are on the defense. Typically, when applying park factors or other adjustments to statistics, we simply multiply the park factor to get the adjusted statistic. However, our park and pitching factors are reversed, with higher numbers representing a higher standard for the defense. In order to make our measurements accurate, we'll multiply by the inverse of the baseline. This calculation makes our formula for determining Team Adjusted Defense (TAD):

TAD = Def_Eff * 1/(BasePIDE) * 1/(BasePADE)

It should also be noted that when generating adjusted statistics, the park factor is halved before multiplying. This adjustment is made because we're using the player or team's home park factor, yet teams only play half of their games at home (excepté les Expos). That adjustment doesn't need to be made to either PIDE or PADE. PIDE involves pitching; teams play all 162 games behind the same pitching staff. PADE was already park adjusted for each team's entire schedule, weighted by the number of games they played in each park, so again, there's no reason to halve the park factor.

Before we get to the final numbers, one more note should be made. As mentioned last time, the BasePIDE range is significantly larger than the BasePADE range. The likely explanation for this is that teams play behind a unique pitching staff (save for trades) all season long while they play in many of the same parks. Sharing the same venues tends to keep the range tighter because extreme parks have only half the effect of extreme pitching staffs. Therefore, our final metric will usually be altered more by pitching staffs than by park factors. In the battle of the lame acronyms, PIDE wins.