The infield shift is a much-maligned defensive strategy, hounded as one of the worst analytics-based changes to baseball. Multiple times each season there will be some conversation about banning the shift, and each time pros, ex-players/managers, commentators, and analysts will chip in with their two cents. But for now, the shift is here, and it is as popular (with the fielding teams) as it has ever been. Just under 26% of all pitches were thrown with some form of infield shift in place in 2018, 22% of at-bats had a shift for the entirety of it, and 30% had at least one pitch shifted.
As you can see, left-handed hitters are far more likely to be shifted than their right-handed counterparts, with 46% of left-handed ABs seeing a shifted pitch versus 19% for righties. This makes rudimentary sense as the shifted players for a left-hander are closer to first base, so they have a greater chance of impacting the play to first and therefore stopping a potential single.
I have taken players who have 100-plus at-bats in 2018 both against a shifted and non-shifted infield, then I compared the outcomes. There were 132 such players, and their combined number of at-bats was 72,389 (39% of the seasons total). I have split these up into four categories based on the handedness of the batter and the pitcher.
Let’s start with the top and LHH vs. LHP, where we can see that that walk rate and home run rate have increased, but for all other factors the shift has had a positive outcome for the pitching team. The increased walk rate is outstripped by the increased strikeout rate, and the overall impact (wOBA) is down just over 6%. Shift working.
As for LHH vs. RHP, this follows a very similar pattern to the one versus LHP but with slightly higher increases to walk rate and home run rate. The positives for the pitching team are a little bit smaller as well, but the overall impact is still a reduction in wOBA of 4%. Shift working.
RHH vs. LHP does not look good for the shifting teams, as the only stat reduced is BABIP. More walks, less strikeouts, and a higher home run rate is not what teams would be wanting off shifting these players, and the overall difference is an improvement of just under 5% in wOBA. Shift not working.
Finally, RHH vs. RHP. As with versus LHP, they walk more and strike out less, but thanks to a bigger dip in BABIP and a decreased home run rate, the overall impact of the shift is a slight decrease in overall production of 1.7%. Shift just working.
The walk rate went up across the board when shifted, but this was countered for left-handed hitters by them being struck out more but not so for right-handers. Now while this is interesting, there are various factors which could impact these outcomes, including the pitcher himself. To remove this impact, I calculated the same rates for all pitchers in 2018 and prorated them to give expected rates from these at-bats. Using those figures, I then calculated how much above or below this expected outcome the results were. I scaled this to 100, with 100 being inline, less than 100 being below average, and greater than 100 being above average.
As before, you can still see the same patterns in the difference between shifted and non-shifted at-bats. The percentages have changed but most not significantly. This still shows that shifting is overall (wOBA) working against left-handed hitters and isn’t against right-handed hitters when they face left-handed pitchers.
How a team shifts and where they position their infielders can be affected by the number of runners on base. To truly interpret if these changes are the impact of the shift, we should look at cases where the bases are empty. The table below does so, with the same logic of comparing it to pitcher averages.
Accounting for bases-empty scenarios only, we now see that shifting isn’t working at all against right-handers. It is reducing BABIP but with increases in walks and home run rate, and decreases in strikeout rate, so the overall impact on batter performance is an increase and not the expected decrease. Left-handers remain the same as before with the impact more significant for left vs. left encounters. Also, this now shows that in all scenarios, the home run rate is higher when shifted.
Further to the adjustments made to account for pitchers and the impact of batters on bases, we should account for any selection bias of the hitters. To do that we weight number of at-bats and shift to non-shift ratio. If a player was shifted 100 times and not shifted 500 times, then their weight would be 100. If they were shifted 300 times and not 300 times, their weight would be 300. Repeat this for all hitters and come up with the weighted average. Using that and previous adjustments, you get the following.
These adjusted results still align with previous findings, as shifting looks like it is working against LHH but not against RHH. Next is to look at the teams who are shifting to see if there is any influence there.
Let’s start with looking at how good the teams are defensively. The two main metrics which are used to describe defensive performance are DRS (Defensive Runs Saved) and UZR (Ultimate Zone Rating). These both try to measure the same thing, but they do it differently, so this analysis uses both to approximate which teams have the best infield defense. The table below ranks the teams based on 2018’s DRS and UZR for infield players only and combines the ranks for an overall ranking.
With this list of which infields are defensively best, let’s look at which teams shift more that others and see how good they were at it. As with the previous piece, this will focus on bases-empty scenarios only, so that there is no impact of runners on base with the positioning of the infield. The outcomes have been adjusted for batters and pitchers the following ways:
- To begin, it takes the statistics as before and calculates them for shifted and non-shifted at-bats of each team.
- For these at-bats, I have also generated a season version of each stat, split by handedness, for the batters and prorated them to give expected rates from these at-bats.
- Using those batter figures, I have then calculated how much above or below this expected outcome the results were.
- For pitchers, I calculated if there was any difference between ability of pitchers with and without the shift within a team.
- I then adjusted the expected outcome, for the shift, based on that difference (max pitching team difference for wOBA was 3.6%).
- Finally, I scaled this to 100, with 100 being inline. As we are looking at wOBA from the pitching side now, less than 100 is better than expected and greater than 100 is worse than expected.
This data is only split by batter handedness, due to a small sample size when splitting the data further by pitcher handedness.
All the following charts are color-coded with red being good, blue being bad, and white being no difference.
Shifting vs. Left-Handed Hitters (LHH)
The table below is ordered by the teams which shift against LHH most. It shows the team’s defensive ability (DEF Rank), the volumes and percent which a team’s shifts, and the outcome for the at-bats.
Looking at the shift percentage and the teams’ defensive ranks, you can see that the teams with worse defenses shift more often, with only two of the top 10 defensive teams in the top 10 for shifting and six of them in the bottom 10 for shifting.
That may make some sense managerially; if you had a infield which was very good, you might be less inclined to make changes as you expect the improvements to be more marginal for your team. But this may also indicate that the standard defensive metrics are also skewed towards defenders who play in the traditional defensive positions.
With that in mind, let’s look at the performance of the shift. For the teams which shift most, there are three clubs with significant improvements when shifting – the Astros, Yankees, and Twins all show an improvement of over 15 points.
It looks like most teams are getting positive results out of shifting left-handers, and teams who have better defensive infielders are shifting less. But how are they getting these results? Is this from a reduction of BABIP? Let’s look at the same factors as we did for the batters.
The table below is ordered by the overall performance difference (wOBA) between shifted and non-shifted at-bats, compared to expectations.
Reducing BABIP is thought to be the principle component of what the shift is trying to accomplish, and 18 teams managed to do this. Of those, 15 managed to bring down the overall production (wOBA). But as you can see, there are other factors which contribute to the change in overall production of the batters. Changes in the K-BB and HR rates also have massive impact on the batters’ overall performance.
Looking at the Astros and the Yankees, you can see that they haven’t decreased the BABIP significantly, but they have in increased their K-BB rate and decreased their HR rates. Both these teams have inflated HR rates (over 60% above expectation) when not shifting and decreased rates against expectation when shifting (Astros -20% & Yankees -45%). Conversely you have the Mariners, who did reduce BABIP but their HR rate increased (from -47% to -1% against expectation).
But the Rockies…
There is only one team which shifted often and had significant deterioration – the Rockies at 10 points worse off. What makes this more interesting is that the Rockies are the top ranked team defensively of teams which shift more than 50%.
As we all know already, whenever the Rockies or their players appear to stick out in a stat, you should check their home/road splits. In 2018, the Rockies shifted similarly home and away but the impact of the shift was dramatically different. The chart below is for LHH only.
The impact of shifting in Coors Field was a 36 point increase in batter performance. When the shift is on, it is 16 points up from expectation, but when they don’t shift, they reduced the expected batter performance by 19 points. When at Coors Field, HR, BB, and XBH rates all nearly double when the Rockies put a shift on. That should be a lesson for Bud Black in 2019.
If we look at how other teams do when shifting in Coors Field, we also see an increase in batter performance, although a much smaller one. We have decreased sample sizes, so this will require further investigation to see if this impact is significant, but this does suggest that shifting at Coors Field may not be a worthwhile endeavor.
Shifting v Right Handed Hitters (RHH)
In the chart above, you see that vast usage difference in facing RHH compared to shifting against LHH.
Tampa Bay shifts the most but their rate of 36.6% would put them 24th on the LHH shift rate, so teams have clearly not introduced this as widely. The defensive ranking is split out a bit more evenly here, but there is still the general trend of teams with worse defenders shifting more. Additionally, there are similar teams at the top, with the exception of Tampa Bay, who was 16th for LHH. However, unlike shifting against LHH, most of the teams which are shifting a lot are not decreasing hitters performance when they do so.
Of teams who shift more than 15% of the time, only the Twins have a significant decrease in batter performance (9 points), whilst the Yankees have a significant increase (24 points).
The Yankees have seemingly been using the shift against RHH experimentally, with them shifting against 65% of batters (78% when at home) in May but averaging around 20-25% in other months. During that time period, they played at home against the other four teams in the AL that reached the postseason, as well as the Angels, so it looks like they perhaps were trying to see if there was any impact and stopped doing so once they concluded it wasn’t effective.
Like with the LHH, let’s look at what is going on to see if we can see why. This time I have kept it in shift percentage order, so it is easier to see what is happening for the teams that shift the most.
Of the teams that shift more than 15% of the time (12 clubs), eight of them managed to decrease BABIP when shifting, nine of them drop the HR rate, and only three of them managed to decrease wOBA. The difference here, compared to LHH, is that the K & K-BB rates drop significantly for all bar one of these teams, the Royals. This means the overall impact for most of these teams is an improvement in batter performance and not a decrease.
The Yankees are an extreme case here, with their walk rate being 38% lower than expectation when not shifting and 72% higher than expectation when they shifted. Their strike rate dropped from 37% higher than expectation when not shifting to 22% lower than expectation. Combined these lead to a change in the overall batter production from 12% below expectation to 12% above, even though there was only a marginal difference in BABIP and HR rate.
This is the analysis which shows that the shift, for both RHH and LHH, is something way more than just moving some infielders about to stop some singles. It suggests that batters, pitchers, and catchers are all changing their mentality and approach to what is going on in front of them. If everyone was playing the game the same, you wouldn’t see these drastic changes in categories outside BABIP.