How Horizontal Trajectory Affects BABIP

Batting average on balls in play has long been used as a yardstick for batters. Players whose BABIP is higher than the league average and their historic stats are said to be fortunate and likely to see a major drop in production when their luck runs out.

It occurred me that BABIP could be used as a defensive metric when Ben Lindbergh used it, calculated in groups by horizontal hit trajectory, to illustrate the improvement in the quality of Major League fielders over many decades. Considering an emphasis on defense (it’s said that it’s cheaper to prevent a run than to score one), the rise of the infield shift, and more athletic and able players, it’s certainly conceivable that defenses today are better than they were in the 1970s.

However, that’s not particularly useful for the task of roster construction today. Sure, Manny Machado may be a better third baseman than most of the professionals 50 years ago, but how much better than his active peers is he? And how much is he worth to the Orioles in run prevention and ultimately revenue generation (i.e., as the General Manager, how much I should pay him)?

To begin to answer these questions in a new way - good but volatile defensive metrics exist, remember - I sought to categorize balls in play by horizontal angle as Ben and partner Rob Arthur did, but also to build in the ability to separate fielding teams from one another. This would allow me to strip out BABIP against the Orioles (or any team) by horizontal angle - absolving Manny of balls hit to areas of the field that he isn’t responsible for, and better indicating which hits came in his domain - and compare it to other teams or the league generally. As such, we are able to see which teams are best at fielding in which locations around the diamond, and better ascribe that to specific players.

The wonderful and regularly-updated PITCHf/x database produced by Baseball Heat Maps was my primary source of information, as it included coordinates for the location of every ball in play since PITCHf/x was introduced in 2007. One thing to note - the coordinates are for where the ball was played by a fielder, not where the ball first hit the ground. A liner that hits the grass and reaches the outfield wall before being picked up in the warning track is shown as having a y-coordinate of the warning track, not of the grass. That’s a big issue if you’re measuring hit distances, but not so much if you’re measuring hit angle; rarely do balls take sharp turns in the field of play before being played.

The other critical source that needs recognition is this Hardball Times article by Peter Jensen from 2009. The PITCHf/x data contains no information to put hit coordinates into context, and the coordinates make little sense to someone used to the normal four-quadrant coordinate plane. Home runs typically have a y-coordinate of less than 20, while bunts have a y-coordinate of around 200. Peter Jensen answered a ton of questions by publishing his normalized estimates of home plate coordinates in each stadium, which allowed me to manipulate the data to orient the field of play in a way that made sense with a regular plane - and to calculate the horizontal angle of each ball in play.
If you visit that Hardball Times link, you’ll notice that the home plate coordinates were from 2008, before the Marlins, Yankees, and Mets began playing in their new stadiums. To avoid complicating things even further, I elected to use the 2008 context for all of the PITCHf/x information. The home plate in most stadiums is just about at the y-coordinate 200.0, so I can’t imagine the MLB Gameday scorekeepers would deviate much from that in three parks.

With all of this information (plus hits and outs) properly paired and grouped, we can see how well the Orioles do in the field against balls batted to different parts of the park:

They are, in this case, very much average. In fact, most teams are:

There just isn’t much variation in BABIP by hit angle across Major League Baseball, likely because certain balls in play are going to be hits nearlyall of the time. It doesn’t matter who’s on your team; a grounder to the shortstop is always going to be an out (or an error, which isn’t counted here), and a liner to shallow left center is always going to be out of reach of everyone running towards it.

Since Manny Machado’s call-up, we might expect to see the Orioles leading Major League Baseball in BABIP allowed by hit angle. He is, after all, one of the best fielding third basemen in baseball, with stellar range and a cannon for an arm - all while playing shortstop throughout the minors. Surely a player of Manny’s caliber in the field would hurt the ability of a batter to reach base safely if he were to put the ball into play, right?

Actually, Manny’s call up hasn’t materially affected the Orioles’ ability to prevent hits once the ball is in play.

To get a better sense of the Orioles’ ability to prevent hits when a ball in play might be expected to be played by their third baseman, I pared down the data to only include a hit distance inside or just out of the infield - ones that are played by the third baseman or by the outfielder rushing in to back up the third baseman once a ball gets through his area - and balls hit between -55 degrees (which is just foul) and -22.5 degrees (which is halfway between the foul line and second base). This seems like a good set of balls in play to assign as the responsibility of the third baseman. I included balls past -45 degrees because MLB’s data tracks where the ball is played, not where it lands, so a live ball that gets picked up in foul territory would be played at a less than -45 degree angle.

I calculated the league average BABIP using this set of data that could reasonably considered specific to third basemen, as well as the BABIP of each of the thirty teams in order to determine the standard deviation from that average. The data follows a roughly normal distribution, although Toronto is apparently so bad at third base as to be a significant outlier that made me think I had very heavy skewed data.

League average BABIP on balls to third base and in the infield is 0.117 - very low! Which is to be expected, given that those are balls played in the infield and therefore did not get past the third baseman. The Orioles, since Manny’s callup, have allowed a batting average of 0.122 on similar balls, or about 0.41 standard deviations above average. This is well within the range of possibilities via random chance, meaning that there might not be any cause for alarm regarding infield defense.

Rather, what this means to me is that the dazzling, rangy plays that it seems like Manny makes on a regular basis might not have a material affect on outcomes. Even though Manny can field a bunt on the grass and fire to first from an impossible angle to garner a putput, he doesn’t do so with enough regularity to dramatically affect the Orioles’ ability to suppress batting average on balls hit to him - and really, nobody in Major League Baseball does.

But wait - that was only on balls fielded in the infield. Perhaps Manny is unusually good at stopping balls hit towards him from reaching the outfield, and makes more outs on those than normal. In fact, the Orioles fare slightly worse in this category, with a BABIP allowed of 0.335, a full standard deviation above the league average of 0.322. Here, again, I have to consider the possibility that despite his incredible range, Manny doesn’t have supernatural talent at preventing an unusually high number of balls in play hit towards him from becoming hits.

There is one final way to slice the data that might shed a more optimistic light on Machado’s fielding. Since his callup on August 9, 2012, Manny has missed large chunks of time for two separate knee injuries. By limiting the sample further to include not only games since Manny’s callup but also only games in which Manny was an active player, we see exactly how his presence on the field has affected the Orioles’ defense. At the same time, this limits the sample size significantly, the information is far more prone to drastic swings, and our conclusions should be couched to recognize that our understanding of Manny’s defensive ability is still developing.

The first thing I notice is that the Orioles are about as adept at preventing balls in play from becoming hits with and without Machado on the field - and they’re about league average at it. Minor fluctuations visible in this chart are more than likely the result of small samples rather than Manny being better or worse than fielding balls in play than average.

Remember, this is not to say that Machado is somehow overrated or that we should reevaluate our understanding of defensive ability. Not one team was noticeably well above average in preventing hits on balls in play - but some were noticeably worse, particularly at first base. Manny’s greatness comes from many things, and although he can snag a hard grounder up the third base line arguably better than most players, those web gems are few and far between for a reason: they’re incredibly difficult plays to make. Turning them once in a while is awesome, and proves that the ability is there, but doesn’t materially affect how often a player who hits them gets on base against the Orioles. Some balls in play are just impossible to turn into outs.

* * *

In addition to slicing this data up by pitching team, it’s possible to examine in-play occurrences in each park. The results of doing so are expected to be pretty similar to splitting the data up by pitching team - after all, nobody plays at Camden Yards without the Orioles, so roughly half of the balls in play in Camden Yards occur with Adam Jones, Manny Machado, and J.J. Hardy on the field.

Where this is interesting is on the margins. All Major League infields are identical in shape and size, but outfield corners can be in any variety of shapes and any distance from home plate.

For instance, the right field corner in Fenway Park, where the outfield continues past the foul pole, seems to cause fielders some trouble. There’s even evidence that BABIP is unusually high when the ball is hit at a 20- to 30- degree angle, which for alignment is about where the second baseman would usually play. Whether that abnormality is due to liners over the second baseman’s head dropping in because the right fielder is in position to cover Pesky’s Pole, or due to teams rarely shifting at Fenway (doubtful, seeing as how the Red Sox are among league leaders in runs saved via the shift, the Orioles and Yankees are two of the teams that most commonly employ the shift and play in Fenway often, and because the Red Sox employ David Ortiz, one of the most shiftable players in baseball).

Unlike Fenway, Yankee Stadium features a below-average BABIP on balls hit sharply to right field. Perhaps the very plain dimensions of Yankee Stadium are easier to play than the quirky ones of Fenway - or perhaps long fly balls that would fall on the grass or hit the wall in most stadiums carry into the right field bleachers in the Bronx, lowing the number of hits and at bats considered in the BABIP calculation while leaving the balls in play that are easily fielded.