Horizontal movement may not mean as much to Pittsburgh Pirates pitchers as you may think

In continuing with our recent series of articles identifying the improvements Pittsburgh Pirates pitchers can see by varying eye level of pitches, it’s important to paint the obvious other half of the picture as well: horizontal variation.

To do this we’ll be utilizing similar methodology outlined in this article ,which looked into the benefits for Pittsburgh Pirates pitchers to mix up vertical pitch locations.

One key distinction, and an improvement on the original methodology, is that we will only be looking at plate appearances with more than one pitch thrown. This affords us a more accurate look at the actual effect in a given at bat of changes in positioning since there is obviously variation in consecutive pitches that doesn’t occur with a one pitch plate appearance.

Similar to the original article we will be looking to maximize the likelihood of a “successful plate appearance”, defined as either a strikeout or an xwOBA of less than .280 for balls in play. Additionally, the statistical model will be similar to the original one, utilizing a logistic regression with the form:

PA Success=f(Avg Horizontal Difference, Avg Horizontal Difference², Avg Vertical Difference, Avg Vertical Difference², Average Speed of Pitches, Average Changes in Speed, Total Pitches in the PA, Strike Zone Size)

The only difference between our previous studies and this one is addition of horizontal difference.

Intuition isn’t always right

The primary takeaway for Pittsburgh Pirates pitchers here is that, effectively, changing horizontal location of pitches only leads to a decrease in the likelihood of a successful batter faced. This is a fascinating finding as it seems to run contrary to the intuition about baseball, namely that keeping a batter guessing on whether a pitch will be inside or outside should make a pitcher more effective.

The exact reasoning for this is not immediately clear; it could be that batters can more easily pickup on horizontal variation out of the pitcher’s hand than they can for vertical variation. There is some work being done on this idea using Pitch Tunnels over at Baseball Prospectus, but the results aren’t in yet. It could also be that pitchers are much less likely to venture horizontally and potentially leave a pitch in a hittable zone, preferring instead to attack one edge of the strike zone.

That being said, there is a bit of intuition that does point to why this might be. A batter looking for a pitch inside that ends up being outside is still swinging his bat horizontally through the strike zone, covering at least some of that horizontal variation. Even if they miss the sweet-spot on the bat, they’re still more likely to make contact with a horizontally varied pitch than if they swing low and the pitch is high. This leads to the next point.

Horizontal or vertical?

According to the data, horizontal variation is much less effective on a per-inch basis than vertical variation. Increasing the average horizontal difference by 1 inch on average yields a 0.02% boost in the likelihood of a successful PA, albeit a statistically insignificant one. Increasing the vertical variation by 1 inch yields an average 0.3% boost in the likelihood of a successful PA, meaning that vertical variation is about 15 times more effective at getting batters out than horizontal variation.

What makes this point even more surprising is that the strike zone is a rectangle where the vertical height is about 22.5 inches on average, and the horizontal width is 17 inches. Theoretically, an inch of movement in the horizontal direction should, proportionally, be “worth” a lot more than in the vertical, by nature of there being fewer inches horizontally. The fact that is so demonstrably untrue, lends additional evidence to the theory that batters can more easily adjust to horizontal variation than they can to vertical.

Under league average conditions for the other 5 variables, a pitcher optimizes their likelihood of a successful batter faced with a horizontal variation of only 2 inches. Mind you a baseball has a circumference of roughly 3 inches, so this is realistically an unachievable goal. Any more variation and a pitcher would reduce their likelihood of striking a batter out or generating weak contact.

Elsewhere in the model, we see that average speed has a negative impact on the likelihood of a Plate Appearance being successful, likely due to a reduction in control at those higher speeds. The variation in pitch speed is a key to a pitcher’s success having roughly the same effect per-MPH as eye-level changes has per-inch. There is no more obvious example of this than Filipe Rivero, who has an average fastball velocity nearly 10 mph faster than his changeup and ends up making a lot of batters look silly as a result.

There is also the, somewhat obvious, observation that the larger the batter’s strike zone, the less likely they are to be successful at the plate, likely due to the decreased likelihood of walks.

Finally, and somewhat surprisingly, the number of pitches in the plate appearance had a statistically insignificant effect on the likelihood of a successful batter faced. Baseball intuition tells us that the more pitches in a PA, the more the batter gets to see, and the less likely the pitcher will come out on top, this is, however, not proven to statistical significance.

How did Pittsburgh Pirates pitchers Do?

In short, not great in terms of their horizontal changes, but they are beating the trends.

ERA & FIP

The Pirates had the third highest average horizontal difference last year at 8.17 inches, but were middle of the pack in terms of ERA. This may be evidence that the Pirates are quite effective at generating weak contact. As I said before, a batter swinging through the zone covers some of the horizontal variation of the pitch by nature of the fact that they swing mostly horizontally through the zone. There is however, the possibility that increasing this variation gets more contact off the end of the bat or closer to the handle, creating weaker contact. It would therefore make sense as to why the Pirates, who posed a 4.22 ERA, were able to beat the trend which would’ve put them at about 4.60 ERA.

Here’s a similar graph with similar interpretations for FIP:

SO/BB

If we look to slightly more advanced metrics as an indicator of performance we can look at strikeouts per walk as it relates about as well as any pitching metric to a team’s wins.

Here the Pittsburgh Pirates are again beating their expected SO/BB ratio at about 2.5 strikeouts per walk issued over an expected 2.3. While this could point to the Pirates striking out more, it’s more likely that they are reducing their walks by substituting them for weak contact. This plays out in the numbers as their SO/9 is almost exactly the same as expected and their BB/9 is less than expected, meaning this exchange of walks for weak contact is almost certainly what is happening.

The Takeaways

While it’s pretty clear that the more horizontal variation a pitcher has, the less successful they will turn out, I wouldn’t say that a pitcher ought to eliminate it from their repertoire, just that they should limit its use.  The Pirates are a prime example of this, despite being one of the league leaders in horizontal changes in pitch location; they manage to leverage it to good effect by generating weak contact, which gives them a reduced ERA.

What is clear, from this article as well as my previous ones (here and here) is that increasing variation is in vertical location is much more important to a pitcher’s success, along with speed variations.

Moving toward a staff that has improvements in eye-level changes and speed changes, as well as either a maintaining or reduction in horizontal changes should result in an even better pitching staff for the Pittsburgh Pirates in 2018.

Photo credit – Daniel Decker