Should the Twins shift less?

One of the well worn stories regarding the Twins during Derek Falvey’s tenure has been the club’s markedly increased emphasis on using analytics to inform strategy and on-field decision making. The area that might illustrate this best is on defense, where the Twins have enthusiastically embraced the deployment of infield shifts in recent seasons. For data collection purposes, MLB defines a shift as a defensive alignment where there are three or more defenders positioned on the same side of second base. By that definition, the Twins have increased their shift usage every season since Falvey took the helm after the 2016 season:

While the rest of the league increased its use of shifts, the Twins did too, staying steadily inside the top 8 teams for frequency of shifting each season.

Breaking down these overall numbers, there has long been a disparity in the frequency shifts are deployed against left-handed hitters and right-handed hitters. Historically, lefty swingers have been shifted against more often. But the gap is closing. In the 2017 season, there were nearly three times as many shifts deployed for left handed hitters than right. In 2018 that ratio dropped to about 2.3 times. In 2019, it was just a shade over 2.0 times. In 2020, it decreased again to about 1.75 times. This change occurred despite the fraction of left-handed hitting plate appearances facing a shifted infield increasing year over year each season. In fact, 2020 was the first season the shift was deployed against left-handers the majority of the time - 50.1%. The changing ratio isn’t because teams have been shifting lefties less. It’s because they’ve gotten much more aggressive in shifting righties. Minnesota has been one of the teams leading that charge:

You can play with the Statcast data for yourself, here. The Twins’ use of shifts against right-handed batters has tripled in Falvey’s tenure and has been well clear of the MLB average in each season. In 2020, though, the Twins’ three year trend of increasing year over year reversed and they deployed shifts against right-handers slightly less often than they had previously. With the shortened season, it’s difficult to know if this 3.4 point decrease is random noise, or intentional. In either case, new research from some of the public analytics community’s leading thinkers suggests the Twins would be wise to continue backing away from their shift-happy ways against right-handed batters.

Questioning the effectiveness of infield shifts

In general, the logic of infield shifting is pretty clear and empirically-based. Ben Lindbergh of the Ringer pointed out recently that batters, from both sides of the plate, tend to hit batted ground balls to their pull side. Aggregated over the past five seasons about 64% of batted ground balls have been pulled, a stat that holds true for both righties and lefties. In the same piece, Lindbergh noted that rate was close to 75% in 2020. With a pattern that clear, it just makes sense to position more fielders in the areas of the field where you can reasonably expect more batted balls to go.

We’re now getting to a point in time where we have enough data to analyze and determine if shifting is effective. To look at this I pulled MLB-wide data from Statcast for the past six seasons, 2015-2020, seeking to compare hitter performance when the shift was and was not deployed against them. First, let’s look at the results for left-handed batters:

In the plot above, the blue lines are batting average (BA) data and the red lines are weighted on base average (wOBA) data. The darker colored lines are the data when shifted infields were deployed and the lighter colored lines are data when infields were in standard alignments. The effects of the shift are evident when you compare dark red against light red and dark blue against light blue. When shifted, left-handed batters produce a much lower batting average, about 21 points lower on average the past 6 years, than when defenses are in standard alignments. wOBA is a more comprehensive measure of offensive production, in large part because it also accounts for walks and appropriately weights extra base hits. Interestingly, in terms of wOBA, the effect of the shift appears to be slightly detrimental to the defense. wOBA has been about 6 points higher when the shift is on.

Now let’s look at the same for right-handed batters:

Against right handed batters, both batting average and wOBA have been significantly higher when the shift is deployed than when it has not. Batting averages were, on average, 11 points higher across the six seasons. wOBA was 38 points higher. That’s not what we would intuitively expect given the growing popularity of deploying shifted infields against right-handed batters. More teams are shifting their infields against right-handed batters more often, yet the data suggests doing so makes the batters more effective?

This data and the plots above also track with recent, more rigorously conducted analytic research. As early as 2017, analysts began noticing the effectiveness discrepancy of the shift against left-handed and right-handed batters. The early data and analysis indicated the shift had been effective against left-handed batters but ineffective against right-handed batters.

In his recent article, Lindbergh pointed out that this data and the simple comparison I did above are inconclusive because we aren’t comparing the same hitter’s performance when they were and were not shifted. To get a definitive answer that stands up scientifically, the identities of the batters and pitchers need to be controlled to ensure the talent is consistent across situations and that any difference in results is attributable to the shift and not skewed because of the sampled batters and pitchers.

In September, such an analysis was completed and published by popular sabermetric analyst and MLB’s Senior Data Architect Tom Tango. In Tango’s study, he found left-handed batters, in bases empty situations against a shifted infield suffered a 24 point drop in wOBA. By contrast, right-handed batters experienced a 38 point gain in wOBA in similar situations. He did this same exercise with a few other factors and controls and concluded essentially the same findings. Around the same time, Baseball Prospectus’ Russell Carleton came to similar conclusions.

Twins’ Shifting Results

For their part, it seems the Twins have fared somewhat better than the league average results when deploying the shift. I re-ran the numbers from above, filtered only for the Twins. In rough terms, Twins’ left-handed hitting opponents have suffered a 31 point decline in batting average and 11 point decline in wOBA on average the past six seasons when the Twins have deployed a shift against them. The aggregated data masks the fact that the Twins’ shift had negligible impact on left-handed opponents’ batting averages and actually helped the opposing batters in wOBA in 2019 and 2020.

On the right handed side of things, opponents have hit 16 points worse when shifted, although their wOBA has been about 12 points higher:

While the Twins-centric data isn’t as clear cut as the league-wide numbers from Tango’s work, I think the upshot remains straightforward. Shifts against left-handed batters generally suppress offensive production. Shifts against right-handed batters have more limited benefits and may actually benefit the offense. All of this then begs the question, why are teams (especially the Twins) shifting against right-handed hitters at all?

Intuitively, it makes sense that teams would want to extend the shift to right-handers. After all, it’s proven to work well against lefties, so it stands to reason that it’s worth a shot against righties. But it’s not quite as simple as the side of the plate the batter is standing changing. The physical dimensions of a baseball diamond and the rules of the game present other variables for defenses to navigate. One very big issue that makes this more complicated is that the bases are run in a counter-clockwise route, making it necessary to devote defensive resources to covering first base. When a team shifts a left-handed hitter to pull, covering first base is not a problem because the defenders are near the first base bag. Pending the baserunner situation, the lone defensive player on the left side of second is often free to be deployed to locations that are most likely to see groundballs batted (usually in the vicinity of the traditional shortstop position). But shifting a right-handed hitter to pull means shifting defensive resources away from first base, with a key difference. Instead of the lone opposite field defender being free to roam to the most optimal locations on the right side of second base, he must stay near enough to the first base bag to field a throw before a runner can reach first safely. This necessity means that teams are often leaving a very large hole around the traditional location of the second basemen.

This logistical hurdle was (is) likely a big factor in shifting against right handed batters being more slowly adopted. Nonetheless, it was adopted. And now, the data is starting to make clear that the frequency of which some teams are shifting against right-handed batters is hard to objectively justify. Perhaps teams like the Rays, Dodgers, and Twins, that have been some of the most aggressive users of right-handed shifts (and are all regarded as some of the best and most forward thinking organizations in the sport), have proprietary data that supports using the shift so frequently. But there certainly isn’t anything in the public sphere that supports it being deployed with such regularity. There are likely to be cases where the shift makes sense against right-handers. Going forward, it seems clear they should be deployed in a much more targeted manner, if at all.

In Lindbergh’s article he quotes an analyst from Sports Info Solutions, Alex Vigderman, who suggests right handed hitters should only be considered candidates for a shift if they pull 80% or more of their groundballs and short line drives. By Vigderman’s analysis, only 43% of shifts deployed against right handed batters in 2020 were appropriate statistically. I searched the Statcast database for the ground ball pull rates of right handed batters with at least 50 plate appearances in 2019 and 2020. Only 19 of 212 players had rates at least 80%. Of those, only two, Chicago’s Edwin Encarnacion and Cleveland’s Jordan Luplow, were regular Twins opponents. The 20th listed player was Kansas City’s Jorge Soler (79.3%), making it only three frequent opponents of the Twins that, statistically, are good shift candidates in this manner of thinking. That’s quite a difference from the Twins deploying a shift in 32% of right-handed plate appearances in 2020.

While the Twins’ results from shifting righties over the past few years have been mixed and perhaps more positive than the MLB average, the overall conclusion seems clear. Shifting right-handers frequently seems to be a losing proposition. It will be interesting to see if the Twins agree with the recent public research and decrease their use of the infield shift against right-handed batters again in 2021.

---

John is a contributor to Twinkie Town with an emphasis on analytics. He is a lifelong Twins fan and former college pitcher.