August 20, 2008

Prospectus Hit and Run

Park Adjusted Defensive Efficiency

by Jay Jaffe

Printer-friendly

Last week, in my look at the Twins' rotation, I made reference to this year's Park-Adjusted Defensive Efficiency statistics, and noted that the Twins were dead last in the American League in that category. I've been meaning to devote a column to the topic for quite a while, so at long last, it's time to get PADE.

For those in need of a refresher course, PADE is Baseball Prospectus alum James Click's method for bringing park effects to bear upon Bill James' Defensive Efficiency (DE) rate, which tells us how often a team turns a batted ball into an out. PADE compares each club's Defensive Efficiency at home to that on the road, incorporating a three-year park factor specific to DE (as opposed to runs, home runs, or another component stat) and the unique composition of each team's schedule. The end product tells us the percentage of balls in play above or below the major league average that each team converts into outs. A one percent difference may not sound like much, but extrapolating from Click's calculations, it's worth about 13 runs—more than one win in the standings.

When I dusted off PADE last fall (Click has moved on to a major league front office, but our data team of Bil Burke and Jason Paré has the special blend of herbs and spices necessary to whip up a fresh batch of figures), I discovered that the two teams which had just met in the World Series, the Red Sox and Rockies, finished first and second atop the year's list. That's a pretty decent advertisement for the utility of the number, particularly given that both teams in question play in very non-traditional ballparks that have the potential to obscure the quality of defensive play. According to the 2007 figures, the Red Sox, who led the AL in raw DE as well, were 3.22 percent above average in converting balls in play into outs after accounting for park effects. The Rockies were fourth in the NL in raw DE, and 3.07 percent above average when accounting for park. Bringing up the rear were the Brewers (-3.44 percent), Marlins (-3.86 percent), and Devil Rays (-5.64 percent).

I mention those numbers by way of introducing the 2008 PADE figures:

Team    DE   MLB DE  PkFactor  PADE
CHN   .7085   .692   0.9957    2.60
BOS   .6994   .692   0.9794    2.13
ATL   .7011   .692   0.9921    1.73
OAK   .7113   .692   1.0313    1.21
ANA   .6995   .692   1.0010    1.03
PHI   .6985   .692   1.0009    0.90
TBA   .7111   .692   1.0382    0.84
FLO   .6917   .692   0.9860    0.67
ARI   .6943   .692   0.9950    0.59
NYN   .7039   .692   1.0285    0.29
CLE   .6880   .692   0.9862    0.12
TOR   .7026   .692   1.0302    0.02
DET   .6923   .692   1.0008    0.01
BAL   .6960   .692   1.0133   -0.08
HOU   .6917   .692   1.0049   -0.28
MIL   .6976   .692   1.0245   -0.40
SLN   .6965   .692   1.0232   -0.50
SFN   .6832   .692   0.9933   -0.94
KCA   .6852   .692   0.9996   -0.96
CHA   .6897   .692   1.0134   -1.00
LAN   .6916   .692   1.0199   -1.03
COL   .6796   .692   0.9889   -1.23
SDN   .6940   .692   1.0307   -1.23
WAS   .6901   .692   1.0207   -1.30
NYA   .6843   .692   1.0240   -2.28
SEA   .6805   .692   1.0180   -2.54
MIN   .6844   .692   1.0436   -3.20
PIT   .6769   .692   1.0255   -3.41
TEX   .6708   .692   1.0157   -3.82
CIN   .6712   .692   1.0272   -4.30

A few notes before we dig in. First of all, the raw DE figures are displayed to the fourth decimal place because the teams are clustered rather closely, with a standard deviation of about 10 points. The DEs above are the ones from our Team Audit Pages, using the formula that incorporates Reached On Error numbers into the equation: DE = [1 - (H + ROE - HR) / (PA - BB - SO - HBP - HR)]. Our sortable stats report uses a simpler formula that's effectively DE = 1 - BABIP. We have the data to go beyond that, so I'll stop referring to the latter version to avoid confusion.

Finally, park factors below 1.0 reflect parks that are more difficult than average to defend, while factors above 1.0 represent parks easier to defend. The amount of foul territory, size of the outfield, irregularities in the field's dimensions, altitude, and surface of play can all have an impact on the degree of difficulty when it comes to defending one's home turf. Fenway Park, Dolphin Stadium, Progressive Field (formerly Jacobs Field), and Coors Field rate as the toughest to defend according to the current three-year factors, while the Metrodome, Tropicana Field, McAfee Coliseum, and Petco Park rate as the easiest.

Given the 2007 year-end results, it's not terribly shocking to find the most recent Hit List's top two teams, the Red Sox and Cubs, occupying the top spots even if they're flip-flopped here. The Cubs have produced the league's top raw DE in Wrigley Field, a park whose ivy-covered walls and small foul territory place it as slightly harder than average to defend. It's worth noting that their pitching staff has the majors' second-best strikeout rate, since high strikeout rates tend to have some impact as far as lowering BABIP figures. On the other hand, the team has overcome the slight disadvantage of playing more day games than any other, as the splits show BABIPs as about four points higher in the sunshine. In any event, the Fielding Runs Above Average numbers show the Cubs playing above-average defense at every position, particularly at second base (Mark DeRosa and Mike Fontenot) and the outfield corners (Alfonso Soriano and Kosuke Fukudome). Given that strikeout rate and FRAA themselves factor into the Secret Sauce recipe, it's no wonder that Nate Silver is ready to crown them champions.

The Red Sox, of course, have to defend the Green Monster and Fenway's other assorted quirks, and while they're seventh in the majors in raw DE, they look much better after accounting for those idiosyncrasies. The Coco Crisp/Jacoby Ellsbury center-field tandem is a combined 13 runs above average, even with Crisp failing to live up to last year's astounding/fluky +41 showing. FRAA is extremely—overly—sensitive to the distribution of discretionary plays divided up between fielding neighbors, and Crisp's 2007 showing is almost certainly an artifact of his covering for Manny Ramirez, who improved to -5 last year after four seasons of being in double-digit negatives. The long-gone slugger rates at about average this year thanks to the help of Crisp and Ellsbury, and while there's not a lot to love about the sight of Manny playing defense, its impact on the team's performance has been dramatically overstated by the jury—as has virtually every other aspect surrounding his acrimonious departure from Boston. Elsewhere for the Sox, Mike Lowell and friends (particularly Kevin Youkilis, who's held down the hot corner admirably while Lowell's been sidelined) are 18 runs above average at third base, and the only spot on the diamond where the team is below average is at shortstop, where Jed Lowrie (-1) has replaced an injured Julio Lugo (-14).

The A's and Braves, two teams whose Pythagorean records put them around .500 instead of 10 or more games under it, are the only other non-contenders in a top 10 that includes both of the Sunshine State teams that dragged bottom last year. The Rays, you may recall, posted a raw DE that was the worst single-season rate in our database (which now goes back to 1954) whether you use either version of DE; the ROE version has them at .656. This year they're a whisker out of first place in that category, thanks primarily to changes at second base (Akinori Iwamura for B.J. Upton), shortstop (Jason Bartlett for Brendan Harris), and third base (Evan Longoria for Iwamura). Their raw DE has improved by 55 points, which would blow away the year-over-year improvement record of 48 points pulled of by the 1980 A's. While that would leave me with egg on my face for having suggested the unlikelihood of such a feat, I'm content to have gotten a bead on a factor that's no small part of the game's most compelling story this year. Over easy on that egg, please.

The Marlins' defense didn't look to be any great shakes as the year began, particularly with Dan Uggla at second, Hanley Ramirez at short, and Jorge Cantu at third, none of them threats to become go-to guys at Web Gem time. All three had established themselves as below-average fielders coming into the year, at least according to FRAA, and of course Uggla made a gruesome spectacle of himself at the recent All-Star Game by committing three errors. Yet their raw DE is right around the major league average, in part because the Marlins' pitchers keep the ball away from those infielders via the league's third-lowest ground-ball percentage, and their center fielders (Cody Ross and speedster Alfredo Amegaza) have done a good job of chasing down fly balls in that spacious outfield to overcome a rather uninspiring group of cornermen (Jeremy Hermida, Luis Gonzalez, and Josh Willingham). Having Willingham (-27 FRAA last year, -2 this year) on the sidelines for an extended period of time didn't hurt their cause on that end.

In terms of year-over-year improvement, the Florida teams are head and shoulders above the rest:

Team   2007    2008     +/-
TBA   -5.64    0.84    6.48
FLO   -3.86    0.67    4.53
MIL   -3.44   -0.40    3.04
ANA   -1.56    1.03    2.59
OAK   -0.46    1.21    1.67
HOU   -1.13   -0.28    0.85
PHI    0.15    0.90    0.75
ATL    1.11    1.73    0.62
BAL   -0.67   -0.08    0.59
SEA   -3.00   -2.54    0.46
CLE    0.02    0.12    0.10
CHN    2.62    2.60   -0.02
SLN   -0.40   -0.50   -0.10
CHA   -0.71   -1.00   -0.29
LAN   -0.72   -1.03   -0.31
KCA   -0.47   -0.96   -0.49
ARI    1.19    0.59   -0.60
DET    0.82    0.01   -0.81
BOS    3.22    2.13   -1.09
TOR    1.12    0.02   -1.10
PIT   -2.21   -3.41   -1.20
NYN    1.51    0.29   -1.22
WAS   -0.03   -1.30   -1.27
SDN    0.38   -1.23   -1.61
MIN   -1.40   -3.20   -1.80
TEX   -1.76   -3.82   -2.06
CIN   -2.12   -4.30   -2.18
NYA    0.03   -2.28   -2.31
SFN    1.61   -0.94   -2.55
COL    3.07   -1.23   -4.30

Third on that list, and now a competent middle-of-the-pack team defensively, are the Brewers. They remade their defense over the winter, moving center fielder Bill Hall (-16 FRAA in 2007) to third base, and third baseman Ryan Braun (-23 FRAA in about two-thirds of a full season's worth of butchery) to left field, and signing defensive whiz Mike Cameron (0 FRAA last year but historically seven runs above average per 100 games) to take over center. Hall has been below average at the hot corner (-7), still an improvement on Braun (Russell Branyan has been respectable there as well), while in the outfield Braun in left (+14) has actually outdone Cameron in center (+1). Again, with FRAA's sensitivity issues, what we're seeing here is likely an illusory result of the more experienced (and possibly collision-shy) Cameron simply staying out of Braun's way on balls that could be taken by either. Nonetheless, it will suffice to say that the team's outfield play has improved.

The bottom third of the 2008 PADE list features only a few contenders, most prominently the Dodgers and Twins. The former has resorted to playing Nomar Garciaparra at shortstop to overcome the absence of Rafael Furcal. They've also had to deal with the continual erosion of Jeff Kent's range, and their outfield configurations without Andruw Jones—whether they stack up as Ramirez-Matt Kemp-Andre Ethier or Ramirez-Juan Pierre-Kemp—don not inspire confidence. The latter has wound up with far too much Brendan Harris in the middle infield given injuries to Alexi Casilla and Adam Everett, and they've also had to face the fact that Delmon Young's outfield play has been rather brutal; he was 19 runs below average last year in Tampa Bay, and is 18 below now; perhaps it's a turf thing that will improve with experience, or maybe he's just a DH in the making.

Finally, it's worth noting the fall from grace of last year's defensive darlings, the Rockies. To be fair, the loss of Troy Tulowitzki for about two months has been critical to that decline; he was at 35 FRAA last year, but rates as only average this year. As inflated as that former number probably is, his importance to the team's defense is backed by strong showings in alternate systems such as the Fielding Bible's Plus/Minus system and Dan Fox's Simple Fielding Runs (worthy of a dust-off itself). For a team that puts the ball into play as often as Colorado's staff does, particularly on the ground, Tulo's injuries are a big reason why the Rockies are looking up at an NL West race that left the station without them.

There are 30 stories to be told with a list such as this one, and I haven't got the space to cover them all in one fell swoop. Without fuller input from a current play-by-play based metric, some of these mini-narratives are almost certainly oversimplified, but rest assured, this isn't the last time we'll shine a light on team defense in his space. Speaking of which, can I get a side of bacon with those eggs?