The Dodgers Might Have a Shifting Strategy of Their Own

Last week, I wrote about the Padres and how their usage of the infield shift stands out. To recap: They shift almost exclusively against left-handed hitters to great success, neglecting right-handed ones in the process. This decision is backed up by public research, which casts doubt on the efficacy of shifts against righties.

As a few of the comments noted, though, the Padres aren’t the most interesting subject when it comes to shifts. If anything, they’re conformists! The Dodgers and Rays, in contrast, are the rebels who defy convention by shifting more against righties than against lefties. We still don’t have a clear answer as to why. Leading up to this article, I did take a crack at the problem, and in the process, unearthed something about the Dodgers.

Before that, some context: Much of our discourse regarding the shift is focused on the dynamic between the hitter and team shifting against him. Kole Calhoun has a tendency to pull the ball, so the Dodgers have prepared this alignment. If Calhoun could go the other way, he’d earn himself a free knock, and so on.

But what about a version of the dynamic that includes the pitcher? By the same logic applied to hitters, if a pitcher could alter his approach to induce pulled grounders that are tailor-made for infield shifts, he’d probably be successful. We know pitchers can control the types of batted balls they allow to some extent: Last season, our Alex Chamberlain wrote about the relationship between pitch location and launch angle. As it turns out, a lower pitch will yield a lower launch angle compared to one located higher up, irrespective of pitch type.

Applying that knowledge to investigate if and how pitchers can maximize their success with the shift made sense. With the caveat that we’re looking at right-handed hitters only, I downloaded every ball they hit during the 2020 season. Each row of data came with information about a pitch’s characteristics, such as velocity and movement, which I used for two logistic regressions: One estimated the probability that a batted ball is a groundball, and the other estimated the probability that a batted ball is pulled. I could have run one regression to model pulled grounders, but doing so left me with a small-ish sample, and besides, I wanted to isolate which variables affected each component.

With the technical jargon out of the way, these are the variables that matter the most in determining the probability of a groundball:

Variables For GB Regression
Variable Est. Coeff. p-value
Pitch Velo 0.069 <0.1
H Mov -0.195 <0.01
V Mov -0.906 <0.1
V Loc -0.613 <0.001
Spin Rate 0.001 <0.01
The model with the lowest AIC was chosen.

There’s a lot on display, but for this article, let’s narrow in on vertical pitch location. It possesses not only the lowest p-value of the bunch, but also the second-highest estimated coefficient. In other words, a change to vertical location leads to a relatively big swing in the model’s output; the lower it is, the more likely a batted ball is a groundball. Notice, too, that pitch category isn’t a significant variable. There’s maybe a minor difference between a curveball and a fastball, but not enough for it to matter, all of which falls in line with Alex’s findings.

Next, here are the variables that matter the most in determining the probability of a pulled ball. Some of them look familiar, and for good reason:

Variables For Pull Regression
Variable Est. Coeff. p-value
Pitch Velo -0.064 <0.001
Vertical Rel Pos -0.125 <0.05
H Mov 0.247 0.111
H Loc -0.99 <0.001
Spin Rate -0.001 <0.05
The model with the lowest AIC was chosen.

Again, pitch location stands out. A negative value for horizontal location indicates that a pitch ended up in the inner half to right-handed hitters, and since the coefficient is negative, the conclusion passes the smell test for me. Watch enough baseball, and you’ll know that hitters have a much easier time pulling inside pitches than outside ones. The model, with good certainty, thinks so too.

Based on these findings, a pitch that fares well (in theory, mind you) against right-handed hitters with the infield shift is one that’s down and in, along with sharp movement in both directions. This brings us back to the beginning of the article. Could it be that the Dodgers and Rays are shifting frequently against righties because they can reliably manipulate the outcome of a batted ball? Below is a graph that shows the number of down-and-in pitches teams have thrown since 2020, with a shift and against righties. Let’s see where those two crafty teams stand:

The Rays elude us, but this is, indeed, something about the Dodgers. Maybe we’re being misled, though. Shift as often as Los Angeles – a whopping 3,765 times during that span – and chances are that pitches end up in that location even by mere chance. But even if you look at the rate at which teams are throwing down-and-in pitches to dampen the effect of volume, the Dodgers are still on top, rocking a 20.7% clip.

You could argue that none of this is deliberate. An alternative explanation is that the team happens to roster pitchers with a penchant for down-and-in pitches. Take Clayton Kershaw, for example. Even before his team acquired a taste for infield shifts, the southpaw built a career out of pounding the inside to right-handed hitters with his slider. A heatmap with all his sliders to date reveals just how aggressive and consistent he’s been:

Think of it this way, though. Rather than disrupting how Kershaw pitches, it makes much more sense for the Dodgers’ front office to devise a shift around his tendencies. And there’s evidence that this is occurring. Since 2020, there have been 413 pitchers who threw a minimum of 200 pitches to right-handed hitters. Of them, I found what percentage of their pitches were thrown with an infield shift. Here’s the top ten:

Top Ten Shift Rates vs. RHH, 2020-21
Player % of Pitches
Clayton Kershaw 88.1%
Victor González 66.2%
Dennis Santana 63.1%
Trevor Richards 61.1%
Tony Gonsolin 60.7%
Noé Ramirez 59.6%
Chad Kuhl 58.6%
Felix Peña 58.0%
Josh A. Smith 55.3%
Yimi García 54.4%
SOURCE: Baseball Savant

Kershaw’s lead here is staggering. And check out the others who’ve popped up: González, Santana, and Gonsolin, all members of the Dodgers’ pitching staff. What complicates this investigation is that those three aren’t attacking the zone the way you’d expect. González’s and Santana’s pitches are mostly down, but they don’t have a consistent horizontal location; what’s more, Gonsolin locates down and away against righties, not in. Maybe the Dodgers are only focused on generating lower launch angles, believing that batted ball direction is more in the hitter’s domain. They’re also frequently aided by the shift against lefties, so batter handedness might not even be a relevant factor.

As for actual results, the Dodgers are finding success. Their .314 wOBA against righties with the shift is ninth-best since last season, and that’s with teams like San Diego whose small amounts are misleading. Exactly how and why is still a mystery, but there’s nothing additional research can’t solve. One possibility is examining the regression models’ unexpected variables, such as pitch velocity. According to the estimated coefficients, a higher velocity increases the probability of a groundball. That’s a bit counterintuitive, but consider Blake Treinen, whose turbo sinkers have translated into a 2.98 FIP and 62.0% GB rate since joining the Dodgers. So maybe it isn’t! Plus, the Rays have yet to be exposed; there’s plenty of work ahead.

Good organizations don’t do things without reason. Data collected by the front office informs the team that it should shift often against righties, despite what public analysts and rival teams may think. The Dodgers, concurrently, are throwing the most down-and-in pitches against them. Maybe this is a coincidence, but there’s also a good chance it isn’t. We might have a glimpse of what the Dodgers are accomplishing through public data alone. That alone is exciting.

Justin is a contributor at FanGraphs. His previous work can be found at Prospects365 and Dodgers Digest. His less serious work can be found on Twitter @justinochoi.

Leave a Reply

Please Login to comment
newest oldest most voted
Brad Johnson

When I evaluated the Dodgers bullpen for fantasy purposes this spring, I noticed they have a definite *type* – relievers who run high GB rates and modest K-rate (by today’s standards). Most of them work down-and-in consistently as part of their approach.

That was before they traded Dylan Floro who also fit this archetype.


Floro (like Kolarek, who was also traded) lacks an additional property many of the LA relievers have – rather high velocity fastballs.

It seems the team used to (say 2015 – 2018ish) have one type – high velo, high FB% guys who could work up in the zone with 2 strikes to induce weak flyballs or Ks (Jansen, Baez, Garcia, & Fields) – and more recently (2018 – present) they have another type – high GB% preferably also with high velo fastballs – Floro, Kolarek, Kelly, Knebel, Graterol, Treinen, May, Santana, Alexander, & May.

Sort this leaderboard by GB% or vFA to see what I mean:,d&page=1_50 . The former is pretty striking.


This is probably an obvious inference but their bullpen also features a ton of sinkers. As a Twins fan, it made me understand a little more why they wanted Brusdar Graterol in the Maeda trade.