BY SAM LINKER

Defensive shifting has greatly affected baseball since it became a common practice amongst all the teams. A lot of emphasis is placed on a hitter’s capability to beat the shift to keep base hits from being stolen. Pitchers are also greatly affected by the shift, as they now have to pitch with many different defensive alignments that yield different outcomes for the same types of batted balls. 

This leads to two questions concerning the merits and practice of pitching with the shift:

1.    Do pitchers change their approach when a shift is in place?

2.    Does changing their approach prove beneficial to the pitcher?

I chose to answer these questions using data from 2019-2020, so the data reflect recent trends in pitch usage and in the outcomes of the pitch. 

By looking at how pitchers utilize their repertoire when there is and is not a shift in place, the difference between each pitch type usage percentage will serve as the basis for defining change in approach. 

Data

At Sports Info Solutions, a shift can be classified as either a full or partial shift depending on the positioning of the fielders. For this study, both full and partial shifts will be grouped together as “shifts”.  

To ensure a sufficient sample of pitches with and without a shift, I chose pitchers who have thrown at least 100 pitches with the shift and 100 pitches without the shift over the span of the 2019 and 2020 seasons. This condition resulted in a sample of 554 pitchers.

I chose the span of 2019-2020 because I wanted to work with data that reflected the recent state of the shift and how pitchers approach it. Also, I wanted at least a full season worth of data since the 2020 season was shortened. 

Along with the pitch type and presence of a shift, I also recorded whether the pitch was a strike, whether the ball was in play and whether it was a hit. These three outcome-based stats serve as a way to measure the effectiveness of a pitcher’s change in approach when a shift is in place. 

Analysis

The difference in Fastball Usage Percentage is the key pitch type since a positive increase in fastball usage corresponds with a decrease in off-speed pitch usage and vice versa. I chose the fastball since generally, a pitcher’s main pitch is his fastball with breaking balls and off-speed pitches (curveballs, sliders, changeups…) supplementing the fastball. Every pitcher has his own arsenal of off-speed pitches, but the universal use of the fastball makes it the ideal pitch to focus on to help measure change in approach. 

Table 1 shows the breakdown of fastballs and non-fastballs for a shift or no shift for all pitchers across the two seasons. 

Table 1: Fastballs v. Non-Fastballs Pitch Usage

Pitch	Shift	No Shift
Fastballs	50%	53%
Non-Fastballs	50%	47%

Without a shift in place, fastballs are thrown the majority of the time. However, when a shift is in place, fastball usage percentage drops to 50% as non-fastball usage rises. Clearly, Table 1 shows that pitchers are using their fastball less when a shift is in place. In fact, non-fastballs have a usage percentage that is approximately 0.4% greater than fastball usage percentage for shifted at-bats. 

I also calculated the difference in strike percentage, Ball in Play (BIP) percentage, and Batting Average on Balls in Play (BABIP) to see how the shift and pitch usage affected the outcome of  the at bats. 

The pitchers can be broken into three groups based on their fastball usage difference: pitchers who decrease fastball usage with a shift, pitchers with no difference, and pitchers who increase their fastball usage. To account for normal variance in pitch usage differences, I decided pitchers with no differences will be pitchers who have a 2% difference in fastball usage in either direction of 0%.

The difference in pitch-usage percentage between a pitch with a shift and a pitch without a shift serves as the key stat in determining the magnitude of a pitcher’s change in approach by pitch usage. A positive difference represents an increase in that pitch’s usage percentage when a shift is in place, and the opposite is true for a negative difference. 

Table 2 represents the three groups along with some notable pitchers in each grouping. 

Table 2: Groups by Fastball Usage Difference

Fastball Usage Difference	Number of Pitchers	Notable Pitchers
Less than -2%	258	Max Scherzer, Clayton Kershaw, Jacob deGrom
Between -2% and 2% (No Difference)	194	Justin Verlander, Zack Wheeler, Trevor Bauer
Greater than 2%	102	Chris Sale, Mike Clevinger, Gerrit Cole

Using the three groups from Table 2, I found the average differences for Strike%, BIP%, and BABIP% for each group. Table 3 contains this information.

Table 3: Average Result Percentage Differences by Group

Group	Strike% Difference	BIP% Difference	BABIP Difference
Less than -2%	-2%	-2%	.009
Between -2% and 2% (No Difference)	-2%	-2%	.011
Greater than 2%	-3%	-2%	.033

Based on the data, the magnitude of change in fastball usage has almost no effect on Strike% and BIP% as the values are exactly the same except for the Strike% Difference in the “Greater than 2%” group. However, the difference is only 1% which seems almost insignificant. The BABIP increases with a shift for all three groups, but there is a clear difference in the size of the increase. 

The “Less than -2%” group and the “No Difference” group have only a 2 point difference in the increase in BABIP when a shift is in place. However, both of these groups’ increase in BABIP is at least 22 points less than the increase in BABIP for the “Greater than 2% group.

Conclusion

Using the data and analysis, we can look back to the two questions presented at the beginning and try to answer them. 

1.    Do pitchers change their approach when a shift is in place?

In terms of fastball and non-fastball usage, Table 1 implies that the usage does not experience much change. However, Table 1 is based solely on the number of pitches thrown, so some pitchers, mainly starting pitchers, most likely dominate the data set which influences the usage percentages. 

Table 2 offers a clearer answer to the question as it looks at the usage differences for when a shift is or not in place which lessens the degree of influence that high-volume pitchers have on the data set. Of the three groups, the “Less than -2%” group has the most pitchers which means they are throwing their fastballs less often compared to non-fastballs when a shift is in place. Overall 360 pitchers are in the groups that signal a change in fastball usage while 194 pitchers are in the group that represents generally no change in usage. 

2.    Does changing their approach prove beneficial to the pitcher?

According to the data, either having relatively no change in fastball usage or less usage in fastballs yields more favorable results when compared to the pitchers who have a higher usage of fastballs when a shift is in place. 

The magnitude of the increase in BABIP is much greater for the pitchers who tend to rely on their fastballs more when the defense is shifted. Because of the similarity between less fastball usage and generally no change, this study cannot definitely say that a change in approach is strictly beneficial. 

However, the data does highlight that recently, pitchers who are increasing their fastball usage with the shift are having less favorable outcomes than those who decrease fastball usage or generally have no change in fastball usage. 

Further Study

I mentioned that the data would only focus on the 2019-2020 seasons and provided my reasons why. Thus, the analysis cannot be used to make hard recommendations as it only contains a portion of the data on shifts. The results presented in this study are used to identify recent trends and potential areas to further observe to form recommendations for pitch usage changes. 

Also, I based the groupings and results on the fastball usage and grouped all other pitches as non-fastballs. 

For pitchers who have been classified as “No Difference”, this only refers to their fastball usage. They could have different usage percentages among their secondary pitches which would classify them as pitchers who do show change in usage. 

The same thinking applies to all three groups’ outcome-based statistics. The differences could be explained by changes in secondary pitches, and the groupings could be remade when factoring in usage change on a more specific pitch type level.