Paul DeJong shifted right to go left

Last week, Statcast released their long-awaited new defensive metric, Infielder Outs Above Average (OAA). For the uninitiated, Infielder OAA estimates the chance that an average infielder converts a batted ball to an out based on how far the fielder is from the ball, how much time he had to get there, how far he is from the base he’s throwing to, and the runner's average speed. As with most defensive metrics, it’s not without issues. Given how difficult it is to quantify defensive value, though, it’s helpful to have another reference point.

With an eye on St. Louis, JP Hill of Viva El Birdos followed up the introduction of OAA with a must-read walkthrough of what the new metric could tell us about the Cardinals. Overall, it seems like OAA lines up well with what we saw on the field. As a team, the 2019 Cardinals ranked first in MLB at +42 outs above average after coming in at 28th (18 outs below average) in 2018 . Individually, Kolten Wong (+10) and Paul DeJong (+13) both ranked in the top 15 last year, with DeJong tied for the second-largest improvement year-over-year.

Looking back at the stat’s makeup, arguably the biggest difference between OAA and existing metrics like Defensive Runs Saved and UZR is that OAA considers the fielders location at the start of each play. Eno Sarris of The Athletic posited that by comparing a team’s OAA rating to its UZR, you could estimate which teams were better at positioning their fielders. Specifically, a team which positioned its fielders well would receive less credit under OAA. Think about these two ground balls up the middle:

In the play on the right, DeJong barely moves a few steps. The average infielder makes this play about 92% of the time when they start where DeJong started, so DeJong receives credit for “0.08 outs” above average on this play. In the play on the left, he has to dive. Based on where he started, the average infielder would make this play somewhere between 60% and 80% of the time. The more often a player has to make the diving play, the higher his OAA but the worse the team was at positioning him in the first place.

Based on that framework, Eno generated a leaderboard to estimate which teams were the best (or worst) at positioning their defenders. The Cardinals came up 29th, meaning they positioned their infielders worse than everyone except the Blue Jays.

The Cardinals being analytically hesitant (and thus worse at positioning) wouldn’t have come as a surprise under former manager Mike Matheny. Under Mike Shildt, however, the Cardinals were supposedly different. Derrick Goold reported in September that the team’s defensive improvement was in large part due to “Pocket Oquendos,” which suggest where the fielder should stand for each hitter. For DeJong, better positioning was thought to be one of the drivers behind his improvement. What gives?

To start, let’s look at DeJong’s range profile for the last two years:

Source: Baseball Savant

In 2018, DeJong was bad going to his right (toward third base) and OK going to his left. In 2019, he was still bad going right but great going left. Knowing he’s better going left than right (which presumably the scouts knew before the 2019 season), you might try to get DeJong more chances going left (or less going right) and, consequently, easier opportunities going to his right (leaving more difficult ones to his left). It looks like the Cardinals made some effort to do just that:

Source: Baseball Savant

Against right-handed hitters, DeJong positioned himself about three feet further to his right in 2019 compared to 2018. Now, that doesn’t mean on every play in 2019 that he stood three feet right of where he would have stood in 2018. He stood in a variety of places throughout the infield, and, on average, those locations were three feet further to the right. That might not seem like much, and it isn’t. Three feet is only three feet. But, on the average play, it was three feet less that he needed to range to his right, and it decreased the chance that on a given ball he had to go right. For balls hit to his left, there was more OAA to be gained since he was closer to third base but a better chance that he reaches the ball and makes the play.

To test the first theory that better positioning would get DeJong easier opportunities overall and more opportunities to his left side, I broke down the DeJong’s Statcast defensive data for 2018 and 2019 into the buckets and scale used by Inside Edge, another 0-100% framework which estimates how often a play should be made. If DeJong’s positioning in the field was really improved in 2019, we’d expect to see that he got easier opportunities more often.

Statcast shows DeJong receiving a small bump in the frequency he received Almost Certain chances. Adjusted for playing time, DeJong received 6 more Almost Certain plays in 2019. DeJong made a small conversion improvement to his Almost Certain chances, which, coupled with more opportunities and playing time, led to a larger jump in OAA. (Inside Edge disagrees on the magnitude of the increase, estimating his Almost Certain opportunities rose from 80% to 86% which may suggest his positioning was more markedly improved).

The more notable improvements come in the Likely and About Even categories, where DeJong improved from -6 to average and -2 to +2, respectively. Breaking these categories down further to isolate DeJong’s chances by direction reveals a possible trend:

Source: Baseball Savant

On About Even opportunities, DeJong was challenged with two fewer plays to his right in exchange for two more chances to his left. On Likely plays, DeJong went to his left more frequently. On Almost Certain plays, DeJong received more chances to his left. The differences, as you can see, aren’t huge. Accounting for the number of opportunities, it’s about nine more plays to his left and three fewer to his right. Maybe the Cardinals didn't position DeJong in the most efficient location for the average fielder, but they seem to have positioned DeJong in a better spot for DeJong.

Obviously, twelve plays don’t explain all of DeJong’s improvement in OAA from 2018 to 2019, but they’re part of the picture. It means he gets a few more opportunities like this ground ball off the bat of Francisco Cervelli:

And a few less like this one hit by JB Shuck:

DeJong makes the play to his left to nab Cervelli look much smoother than when he ranges right and fails to get Shuck. Assuming I’ve narrowed down the plays accurately on Baseball Savant, though, Cervelli’s ground ball was actually the tougher play (76% chance to get Cervelli vs. 87% for Shuck). To get Cervelli, DeJong had to travel a couple feet further after a harder hit ball. Cervelli is slower than Shuck and the throw may have been shorter, but those are already factored in. DeJong just makes one play look easier than the other in part because he’s better going left than right.

Scrolling through the OAA leaderboard, the glove-side bias isn’t contained to just DeJong. Infielders all throw right-handed, and for most righties it’s easier to make a play to the left than to the right. DeJong’s sample size in any category is small. We know we need a lot of defensive data to conclude anything, and that’s especially true for a brand new, untested metric. The number of opportunities and higher success rate on those opportunities might have just been a favorable, temporary swing in DeJong's favor. It’s also entirely possible that after a year at the position, DeJong simply got better regardless of positioning – he hadn’t played shortstop for any meaningful amount of time before 2018.

I’d also speculate that there are some benefits of improved positioning that wouldn’t show up in this data. If DeJong really is better going left than right, there might be intangible benefits and comfort if he feels that he’s in a better spot on the field. That’s not going to show up in a number. Maybe his positioning helped him convert other types of plays at a higher rate. It’s possible it didn’t do much at all, since DRS and UZR rated him similarly in both 2018 and 2019.

The problem with defensive numbers is often that we don’t have a large enough sample to make conclusions with certainty, and that’s still the case here. I might have written all these words for nothing! If you want to believe what we have, though, you can add OAA to the list of metrics which support DeJong’s premium value at shortstop.

Credit to Baseball Savant and FanGraphs for the stats used within this article, and especially the Statcast team for the development of Infielder OAA. Thanks as always to @cardinalsgifs for his contributions.