Smart defense vs. smart offense

In an interview with mlb.com last June, Joe Maddon made a statement that you never would have expected to hear from a big league manager—he acknowledged the impact of sabermetrics on the game of baseball. True, Maddon is a “friend of the numbers people,” but his comments were especially surprising because they focused on the impact of numbers on defense.

Here is an excerpt from the article:

“All the stuff that’s going on and all this stuff that’s talked about, whether it’s data, matrixes, all the different stuff that’s out there is all slanted toward the pitching and the pitching and the defense,” the Rays manager said. “There’s no way to slant it toward the hitters.“…Thus, based on those truths about the games, Maddon believes offensive numbers have no place to go but down.

Maddon implies that, while pitchers and defenses have the ability to adjust to different situations, hitters are stubborn and have a harder time changing approaches. There is some evidence that this is true. Faced with defensive shifts, dead-pull hitters like Carlos Pena seem to be unwilling and unable to hit to the opposite field.

On the other hand, versatile hitters like Joey Votto (if any other hitter is “like” Joey Votto) are thriving. In an excellent interview with Eno Sarris on Fangraphs, Votto says, “I was a big pull hitter in high school, but when I tried to do that in Midwest League, I failed.” Votto changed his approach before reaching the major league level, and has posted consistently high BABIP levels since.

Votto argued that a strong the opposite field hitting “prevents shifting,” and Sarris concluded that hitters who possess this ability post higher BABIPs than those who don’t. If defenders attempt to take away hits in a particular area of the field, Votto can hit it where they ain’t. Carlos Pena will hit it where they are, and then he’ll do it again.

One way to evaluate Maddon’s claim would be to examine trends in batting average for balls in play, or BABIP. BABIP is a crude statistic, but it measures the success of a defense at the most basic level—the amount of balls in play that are turned into outs. The graph below displays this statistic over the last five years. On the surface, it looks like BABIP has been trending downward in recent years—with the exception of 2012. The decrease is so small that it could just be noise, though, because BABIP tends to fluctuate year-to-year.

image

The trend in overall BABIP didn’t tell us much about the impact of defensive metrics, but we can slice this data up in ways that might. Ideally, advanced metrics should help defenses turn more balls in play into outs. However, tradeoffs exist with regard to defensive positioning. Defenses can’t only care about limiting the number of hits per balls in play, because some balls in play are more dangerous than others.

We know that harder hit balls fall for hits more often than weakly hit ones, and we can reasonably assume that pulled balls in play are hit harder than balls hit to the opposite field. League ISO in 2012 for at bats in which a ball was pulled was .276. And for balls hit to the opposite field? .133.

Given this information (and given that some hitters pull the ball a majority of the time), it makes sense that a team would want to position its defenders to prevent a hitter from pulling the ball for a hit. So long as hitters can’t significantly adjust, this reasoning helps justify defensive shifts and shading. Opposite field hits simply don’t do as much damage as pulled hits.

The graph below depicts yearly aggregated values of BABIP, broken up by batted ball location. Pull BABIP considers all batted balls hit by right-handed hitters to left field, and all balls hit by left-handed hitters to right field. Opposite field BABIP considers balls hit by righties to right field, and balls hit by lefties to left field.

Historically, these splits have followed a relatively consistent trend in their relationship to aggregated BABIP. The values plotted on the graph below reveal that balls that are pulled fall for hits about 10-15 percent more often than balls that are hit to the opposite field.

image

If pull hitters can’t adapt to “smart defense,” we might have seen overall BABIP decline last year. However, if hitters were able to make adjustments, any decline in pulled ball BABIP should have been offset by an increase in opposite field BABIP. The graph below includes the 2012 numbers.

image

What happened to that 10-15 percent difference between Pull/Opposite Field BABIP? In 2012, it almost disappeared. A ball pulled was just about as likely to fall for a hit as a ball hit to the opposite field for the first time since we’ve been able to look at comprehensive hit location data (to my knowledge).

Why did BABIP increase in 2012?

Initially, it seems like Maddon’s claim is at odds with the overall BABIP data I first presented—because overall BABIP increased in 2012. However, the pull/opposite field splits above suggest that defenses might actually be gaining an edge.

It’s also possible that BABIP is actually trending downward, and that this trend is favoring defense. To this point, I have not considered the ability that defenses have to stop different kinds of batted balls from falling as hits. When we estimate a player’s expected BABIP (xBABIP), we place more weight on “tough to field” batted balls like line drives and ground balls than we do on fly balls and infield pop ups. Not surprisingly, the number of line drives, fly balls, and ground balls hit in play varies from season to season.

A particularly high number “tough to field” batted balls were hit in 2012 relative to recent years, and expected BABIP was relatively high in 2012. For reference, I have thrown up the 2010 season for comparison below. We saw identical BABIP levels in both years, but expected BABIP (based only on batted ball types) was significantly higher in 2012.

image

Combining the two splits

We know that defenses have started to take away pulled ball hits in favor of allowing hits to the opposite field, but what type of batted balls are falling as hits to each field?

We can take an even closer look at the data if we split batted ball hit rates up by location. The table below presents this information. All hit location data comes from a PITCHf/x database, and I split up sections for LF/CF/RF into something like 33/23/33 degrees to capture more of the possible effect of infield shifts. Each of these rates is not a traditional BABIP number, but instead the number of hits of a specific batted ball type divided by the number of that type hit in play.

The table displays the change in values from 2011-2012 (with negative values representing decreases in 2012). The results aren’t quite what you would expect.

image

(note: home runs are excluded from this table because they aren’t considered “balls in play”)

If “smart defense” continued its development in 2012, we should have seen less dangerous batted ball types fall as hits to less dangerous areas of the field (note: fly ball hit rate does not include infield fly balls). Again, the most dangerous batted ball type is a line drive (and a fly ball, if we count out infield flies). Pulled batted balls are more dangerous than opposite field batted balls. Thus, pulled line drives (and fly balls) should be the most dangerous split in the chart above, and opposite field ground balls should be the least dangerous.

Changes in ground ball hit rates by location are most apparent in the chart. The number of pulled ground balls that became hits in 2012 decreased about 2 percent (or five hits per thousand ground balls). This change really doesn’t seem like much, especially given that significantly more opposite field ground balls became hits in 2012. The infield shift should have had the most visible effect on pulled ground balls, but it doesn’t look to have had much of an impact (on a league-wide scale).

The increase in opposite field ground ball hit rate is significant, and can be interpreted as an accidental change or a causal one. If infields shade or shift completely to protect against pulled ground balls, more hits will obviously fall in vacated spots. Evan Longoria is a solid infielder, but he isn’t good enough to play two positions at once. This is true regardless of any adjustments that hitters make. However, it is also possible that hitters are putting extra effort into hitting harder ground balls to the opposite field. To explore this possibility further, we’d have to look at hit velocity data for ground balls. Whatever the cause may be, it is obvious that opposite field ground ball hits drove the increase in overall opposite field BABIP.

We didn’t see the pulled line drive hit rate change much in 2012, but overall line drive hit rate did decrease slightly. I should note that line drives fall for hits a lot more often than other batted ball types do, so a .9 percent decrease is equivalent to something around 15 less hits per thousand line drives. The implications of these numbers are mixed for a defense. Yes, it is beneficial to limit line drive hits—but it would be most beneficial to limit pulled line drive hits. In 2012, the bulk of the change in overall line drive hit rate came from balls hit to center field.

The decrease in overall fly ball hit rate also suggests that outfield positioning may have improved (or that the quality of outfield play has improved). Ideally, defenses should be placing a greater emphasis on limiting pulled fly balls than on opposite field fly balls (if this is possible). The overall rate did decrease, and again, batted balls hit to center field explain most of the change.

The real question that this chart presents us with concerns the tradeoff between hit rates. From a defensive perspective, is it worth it to allow seven percent more opposite field ground balls to fall as hits in exchange for a one to two percent decrease in fly ball and line drive hits? These estimates aren’t perfect, but they frame an interesting discussion.

Conclusions

This research suggests that new advanced metrics are improving defensive positioning, and that hitters may not be adjusting to these changes. Other defensive metrics can also be used to highlight any of the above trends, but BABIP serves as a simple and easily understandable measure.

It is possible to conclude that 2012 is a random outlier for a few reasons. First, BABIP is a statistic that is full of randomness. It is difficult to predict for hitters and for pitchers on a year-to-year basis. I also may have wrongly assumed that we have had some year-to-year consistency for these splits until 2012, because we only have data on pull/opposite field statistics going back to 2003.

While there may be other explanations for these trends in batted ball data, it is likely that defenses are prioritizing positioning based on a hitter’s tendencies. This is an interesting question, and should be explored further. It seems like harder hit balls are being turned into outs more frequently, and that defenses giving up more opposite field hits in return.

2013 will be a telling year for trends in batted ball hit rates. Will the “opposite field premium” be reinforced, or will it disappear? The answer to this question may depend on how seriously the league takes Joey Votto’s advice—hit the ball the other way, and hit it hard.

Notes:
The expected BABIP formula I used can be found here.

Print Friendly
 Share on Facebook0Tweet about this on Twitter0Share on Google+0Share on Reddit0Email this to someone
« Previous: The daily grind: 4-25-13
Next: Cooperstown Confidential: The wild life of George Brunet »

Comments

  1. David said...

    Perhaps since players can’t get away with steroids like before, teams are rostering more defensive players, and less home run types?

  2. David P Stokes said...

    @the other David:  It’s possible that teams are valuing good defensive players more highly even without bringing steroids into the discussion.

  3. guruE6 said...

    I am no saber-geek, but I could be.

    Needless to say, my first question would be this:

    Is -.02 * VALUE_OF_HARD_HIT_PULL > .07 * VALUE_OF_SINGLE_OPPOSITE

    By forcing hard-hitting players to go oppo, you eliminate a LOT of extra-base hits. So even thought their AVG in such a situation may be .340, their OPS may see a reduction. THIS should be your next article.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Current day month ye@r *