Ahead in the Count: Th... (09/15)
the archives are now free.
14 comments have been left for this article. BP Comment Quick Links
| |
|
|
September 23, 2009 Ahead in the CountSlugging on Balls in Play
Two of the first things you learned about sabermetrics were that batting average is not a very good measure of a hitter, and that fielding percentage is not a very good measure of a fielder. Not all hits are created the same, and extra-base hits cause a lot more damage than singles, which is why we know that slugging is better than batting average at approximating hitting ability. We also know that fielding percentage is a poor measure of defensive skill, because it only considers the balls that a fielder can reach, and therefore does not account for range-being able to get to more balls, all else equal, implies that a fielder is better. Being able to record outs on more balls in play is indicative of better team fielding. That is why the most widely used measure of team defense is Defensive Efficiency, which is effectively one minus your opponents' BABIP (including errors). However, there is a fairly obvious disconnect between those two facts: on the one hand, we avoid using batting average to measure a team's offensive ability, but then we encourage the use of batting average to measure a team's fielding ability. However, teams frequently guard the lines to avoid doubles late in a game-at the expense of making singles in the hole more likely-and play their outfielders deep to avoid extra-base hits-at the expense of letting singles drop in more frequently. These can be wise tactics in certain situations, since there are times when an extra-base hit would be particularly costly. Sometimes acquiring fielders who make doubles and triples less likely is wise, even if it means not acquiring fielders who can prevent singles. Teams that prevent hits are not necessarily the teams that prevent extra-base hits-and therefore are not necessarily the teams that prevent runs. For this reason, I am introducing a new statistic we can use to measure team defense: Slugging on Balls in Play. It's measured exactly how you think it should be: it is simply slugging on all at-bats that do not end in home runs, strikeouts, walks, hit by pitches, or sacrifice bunts, but it counts reaching on errors as singles. Here are the list of all teams' 2009 SLGBIP and RBIP (slugging on balls in play, and reaching on balls in play-which I define as batting average including errors), through Saturday's action:
SLGBIP RBIP
Team Rank SLGBIP Rank RBIP
Mariners 1 .354 2 .289
Dodgers 2 .360 1 .285
Reds 3 .371 4 .297
Giants 4 .371 3 .293
Cubs 5 .372 6 .299
Rangers 6 .374 5 .297
Cardinals 7 .379 11 .305
Yankees 8 .379 7 .302
Padres 9 .380 12 .306
Phillies 10 .381 8 .302
Rays 11 .383 9 .303
Tigers 12 .383 10 .304
Twins 13 .386 15 .309
Angels 14 .391 18 .312
White Sox 15 .392 21 .313
Athletics 16 .394 19 .313
Mets 17 .395 17 .311
D'backs 18 .397 20 .313
Rockies 19 .397 16 .311
Brewers 20 .397 13 .307
Indians 21 .400 25 .318
Marlins 22 .402 26 .317
Astros 23 .406 27 .320
Pirates 24 .406 14 .308
Red Sox 25 .406 29 .320
Braves 26 .406 23 .315
Blue Jays 27 .406 24 .318
Royals 28 .406 30 .324
Orioles 29 .410 28 .320
Nationals 30 .411 22 .315
I have also included each teams' rank at preventing slugging on balls in play, and at preventing hits on balls in play. Perhaps with an eye towards the example set by the Rays' worst-to-first shakeup in part through their defensive improvement in 2008, the Mariners are one the latest teams to stress defense as a means to improvement. In particular, they've stressed outfield defense, and after putting together an excellent defensive outfield, they lead the league in SLGBIP. Another interesting example is the Pirates, who are 14th at getting outs on balls in play, not bad, but they are 25th at slugging on balls in play, at .406. As a result, their ERA is a little bit higher than their FIP. Note that I have used FIP, fielding-independent pitching, which is a statistic that approximates what pitchers' ERA should be based on their home-run, walk, and strikeout rates. As has been discussed previously, FIP is not as stable a statistic as xFIP or QERA, but it is what we are looking for here because we're looking for a statistic that accounts for the fact that home runs have been hit, and delivers an expected ERA conditional on the number of home runs, strikeouts, walks, and hit batsmen occurred, assuming average defense and average luck. Much of the difference between ERA and FIP is therefore team defense. A statistic like xFIP or QERA might be better at measuring a pitchers' run prevention, but the difference between realized FIP and ERA will be highly correlated with defensive abilities. Looking back at 2008, we see the following rankings for SLGBIP:
SLGBIP RBIP
Team Rank SLGBIP Rank BIP
Brewers 1 .372 7 .302
Mets 2 .373 6 .302
Rays 3 .373 1 .290
Padres 4 .376 9 .304
Blue Jays 5 .378 3 .296
Red Sox 6 .378 5 .301
Athletics 7 .381 4 .300
Cubs 8 .382 2 .295
Dodgers 9 .383 16 .309
Angels 10 .383 15 .308
Orioles 11 .383 18 .312
Marlins 12 .385 13 .307
Phillies 13 .386 11 .305
Indians 14 .388 22 .314
Twins 15 .389 19 .313
Astros 16 .389 8 .302
Nationals 17 .390 14 .311
Royals 18 .393 17 .310
D'backs 19 .394 21 .314
Cardinals 20 .396 10 .305
Yankees 21 .396 25 .318
White Sox 22 .399 20 .314
Braves 23 .400 12 .306
Giants 24 .401 23 .315
Tigers 25 .403 24 .315
Mariners 26 .405 26 .318
Reds 27 .415 29 .327
Pirates 28 .416 28 .325
Rockies 29 .416 27 .322
Rangers 30 .421 30 .330
Perhaps the most interesting result here is that Milwaukee tops the list, even though they were only seventh in getting outs on balls in play. However, they led the league in the difference between their ERA and their FIP. This was partly due to Mike Cameron, who had a UZR of 11.3 in just 119 games. Surprising nobody, the Rays were also particularly good at defense in 2008; while going from worst to first in Defensive Efficiency, they were also very nearly the best in SLGBIP too, thanks to superb outfield defense. The Orioles are also an interesting team. They were 18th in the league in getting outs on balls in play, but were 11th at keeping their club SLGBIP down, so their ERA and FIP were nearly identical. Something particularly interesting about the O's is that they had such a difference between their infield and outfield defense: their infield defense had a combined UZR of -32.2, but their outfield defense had a combined UZR of +38.9. Nick Markakis, Jay Payton, Adam Jones, and Luke Scott combined to help them prevent a lot of extra-base hits, as the team gave up twelve fewer doubles to center than the league average, and four fewer triples to center than the league average. They also gave up ten fewer doubles to right than league average, and about 1.5 fewer triples than average. Melvin Mora at third base and the handful of players manning shortstop combined to allow a lot of hits on balls in play-but many were singles. Let's have a look at the SLGBIP tables for 2007:
SLGBIP RBIP
Team Rank SLGBIP Rank BIP
Blue Jays 1 .363 1 .294
Padres 2 .375 4 .297
Red Sox 3 .375 3 .296
Cubs 4 .376 2 .295
Nationals 5 .378 6 .301
Braves 6 .381 7 .302
Mets 7 .382 5 .299
Rockies 8 .387 8 .303
Orioles 9 .388 20 .316
Cardinals 10 .389 9 .303
Dodgers 11 .390 19 .315
Athletics 12 .391 11 .308
Indians 13 .392 18 .315
D'backs 14 .393 14 .310
Tigers 15 .393 12 .308
Twins 16 .394 17 .314
Yankees 17 .395 15 .310
Giants 18 .397 10 .307
Rangers 19 .399 22 .318
White Sox 20 .400 23 .319
Phillies 21 .401 16 .313
Astros 22 .402 13 .309
Angels 23 .402 24 .320
Brewers 24 .410 26 .325
Pirates 25 .412 28 .330
Reds 26 .412 25 .325
Royals 27 .414 21 .317
Mariners 28 .415 27 .327
Marlins 29 .418 29 .337
Rays 30 .437 30 .343
Toronto led the league in SLGBIP and Defensive Efficiency in 2007. Interestingly, although they stayed near the top of the league in 2008, they fell quickly towards the bottom in 2009. Looking at 2007-09 together, we can see some patterns emerge. Preventing hits on balls in play seems to be largely related to infield defense, but preventing extra-base hits on balls in play is largely related to outfield defense. The correlation between UZR of infielders and reaching on balls in play was .39, but the correlation between UZR of outfielders and reaching on balls in play was slightly lower at .36. The correlation of UZR of infielders and SLGBIP was .22, however, and the correlation of UZR of outfielders and SLGBIP was .40. It certainly seems that the Mariners' 2009 charge towards improving outfield defense is the way to go. Of course, the correlation between UZR of a team's infielders combined from year-to-year is much stronger than that of outfielders at the team level (.34 versus .10), so it may be tougher to help your team consistently prevent those extra-base hits. But what we have learned here is that looking at slugging on balls in play is important, and is a helpful additional way to look at team defense. As defensive metrics get more and more refined at the individual player level, it remains true that these are often difficult to truly interpret, and separating the effect of different players is still a challenging endeavor. These metrics at the team-wide level remain far more reliable and important to use as a benchmark, and slugging on balls in play is another way to measure it.
Matt Swartz is an author of Baseball Prospectus.
|
Do you have any explanation or theories on why the Orioles, who were 9th in SLGBIP in 2007, and 11th in 2008 have fallen all the way to 29th in 2009? Jones has still been the predominant center fielder, and Markakis is still in right. Does the change in left field really make that much of a difference?
To what degree do team numbers stay consistent from year to year? We know, for instance, that measures of "clutch hitting" such as BA with RISP can be generated, but are not considered to measure an "ability" because they fluctuate so much from year to year.
To what degree is this pitching rather than defense? For example, one would assume that a ground ball pitcher would have a lower SLGBIP than a fly ball pitcher.
These are interesting concepts. I look forward to greater refinement.
I'll look into this. I tried to do some tests and it seemed like the correlation of BABIP and SLGBIP were similar, but I'll try and dig up a larger sample to check.