Spray Charts from Statcast Data

Review of Statcast Posts

In last week’s post, we discussed how to obtain a spray angle from the Statcast data scraped from Baseball Savant and we used the spray angle (together with the launch angle and launch velocity) to obtain reasonable predictions of hit probabilities.  This week, we’ll discuss several things:

1.  Baseball Savant will produce spray charts of batted ball locations (such as the one for Carlos Santana here) where the color of the location depends on a different variable such as the batted ball type.  We’ll suggest a modification of this spray chart that adjusts for the batter side.  We’ll illustrate it for two hitters who have different characteristics in pulling balls.
2. These spray charts help us understand that batters can have different pull characteristics depending on the batted ball type.  We’ll introduce several “average spray angle” measurements that may be useful in understand batted ball tendencies.

An Improved Spray Chart

When we were modeling a hit probability, it was helpful to define an “adjusted spray angle” that adjusts for batting side, so a negative spray angle always corresponds to a ball that is pulled (a positive spray angle is a batted ball hit the opposite way).  In a similar way it makes sense to adjust the horizontal batted ball location by the batter side so that a negative horizontal location corresponds to a pulled ball.

Using this idea, we produce an adjusted spray chart.  To make it more obvious that this is not the usual spray chart, I color the “pull” area using a yellow color.   I’ll use this to display the spray hitting tendencies of two “extreme” hitters Carlos Santana and Joe Mauer.

In this adjusted spray chart, we see below that Carlos Santana is a strong pull hitter as most of the ground balls are hit in the pull region.  The direction of line drives and fly balls seems more evenly divided in the pull and opposite field directions.  As one might expect, the foul popups (green color) are all hit on the opposite side direction.

By the way, it is interesting to contrast my display with the following Baseball Savant display that does not adjust for side of the plate (Santana is a switch hitter, so it is unclear which balls are pulled).

Joe Mauer has a very different spray pattern than Santana.  His ground balls are more equally divided in the pull and opposite directions.  A majority of the line drives and fly balls are hit in the opposite side direction.

It is a little hard to compare the sprays of the two players by looking at the individual graphs.  Below we use the facet feature of ggplot2 to compare the two players.  Since the scale is the same on each graph, it is much easier to compare the sprays.

Measuring “Pull”

By looking at several of these spray graphs, we see that the spray angle distribution  of one type of batted ball like a ground ball for a player like Joe Mauer can be different from the spray angle distribution from a line drive or a fly ball.  For each player with at least 200 batted balls in the 2017 season, I find the mean adjusted spray angle for (1) all ground balls, (2) all fly balls, and (3) all line drives.  (By the way, Fangraphs compute the percentage of batted balls that are pulled, hit the opposite way, and are hit in the center.  In my work, it seems that Fangraphs says that a batted ball is pulled if the adjusted spray angle is smaller than minus 17 degrees.)

Below I construct a scatterplot of the mean spray angles for ground balls and fly balls where we remember that negative spray angles correspond to balls that are pulled.  I’ve labeled Carlos Santana and Joe Mauer — Santana has a strong tendency to pull his ground balls (mean spray angle about -0.4) but his average fly ball angle is 0 which indicates that his fly balls are equally divided between the pull and opposite field directions.  Mauer has a modest tendency to pull his ground balls (mean spray angle about -0.2), but a strong tendency (mean spray angle of 0.4) to hit his fly balls to the opposite direction.   Actually all players tend to pull ground balls and a majority of players tend to hit fly balls in the opposite side direction.

Below I graph the mean spray angles for ground balls and line drives.  Generally we see that a majority of players tend to pull their line drives, but the pull tendency is stronger for ground balls.  There is a positive association here, so players that tend to strongly pull ground balls also strongly pull line drives.

Questions to Explore

Obviously, teams care about these spray charts since they help in deciding on the positions of fielders.  Given that infield shifting is currently very common among MLB teams, the teams must be familiar with spray tendencies of hitters.

I’d be interested next in understanding how pitchers affect the spray angles.  Specifically,

1. How does the pitch location affect the spray angle?  I would think that batters are more likely to pull inside pitchers, but the size of this effect is unclear.
2. What about pitch type?  It would seem easier to pull off-speed pitches than fastballs.
3. What about the effect of the pitch speed?  Again, I would think it would be harder to pull a Chapman fastball that is thrown at 100 mph