Phillies Defense Part 1: Just out of Reach

This season, as in so many others this past decade, the Phillies defense was a weakness. But it wasn’t supposed to be. In 2019 the defense graded out well by publicly available metrics, especially the outfield. Then, during the offseason, the front office made moves to improve the infield defense. Coming into the 2020 season we had reason to hope that the defense would become a strength: one that could bolster a good rotation and buttress a questionable bullpen. What went wrong?

Let’s once again start by looking at BABIP. As I discussed previously, the Phillies surrendered the highest full season BABIP-against in the post-integration era by a wide margin. In that article I tried to measure how responsible the pitchers were for that BABIP. At first, I concluded that they were minimally responsible. But after lorecore told me to use Statcast’s xBA (expected batting average) filtered for BIPs (balls in play) only, I found that the pitchers might be responsible for as much as half of the extra hits produced by the elevated BABIP.

I say “might” because xBA is based on the launch angle and exit velocity of individual BIPs. Small differences in angle and velocity on contact can make a big difference in outcome in the field. We don’t think the small differences are caused by a pitcher’s performance. As a result, we shouldn’t charge the pitcher with the outcome. When those small differences aggregate and the outcomes snowball, perhaps, we should attribute some of that conglomeration to the pitching. But not all of it, and I don’t know how much, if any. A true xBABIP for pitchers would be based on pitch characteristics such as velocity, movement, and location. Contact characteristics are too closely related to batted-ball outcomes to serve the purpose.

Fortunately, xBA on BIPs is perfectly suited for evaluating the defense. xBA gives us the hit rate allowed by an average defense given the actual contact made on each ball in play. If a team allows a higher BABIP than its xBA on BIPs, then the defense is most likely the culprit. And, as I’m sure you’ve anticipated, the Phillies defense indeed allowed a higher BABIP than its xBA on BIPs. The Phillies BABIP-against was .343, 15 points higher than its .328 xBA on BIPs. The difference, then, between the Phillies defense and an average one amounts to 25 hits over the 60 game season.

How did the defense reach 25 extra hits allowed? Where were the holes? A simple question requiring a complex response. On the one hand, the defenders’ performances after the ball is hit bear some responsibility. On the other hand, the defenders’ positioning before the ball is hit determines how effective any performance could be in the first place. I’ll start with the individual defenders.

We have three prominent defensive metrics: UZR (ultimate zone rating), DRS (defensive runs saved), and OAA (outs above average). I’m going to justify relying primarily on OAA; so, if you don’t care about methodology, skip this paragraph. Of these, Rob Arthur found that only OAA correlates significantly with team spending on defense. Since teams have access to more Statcast data than the public, they have better proprietary defensive metrics. If they aren’t paying for it, it’s probably not measuring defense accurately. The reason for that is DRS and UZR depend on heavy regression to the mean over small samples and bucketing parts of the field, quality of contact, and the role played by the defender (SS, 2B). As Mitchell Lichtman explains for UZR, the regression and bucketing entail that these metrics measure true talent more than what actually happened on the field. In contrast, OAA measures continuous quantities rather than discrete ones, which makes it a more reliable indicator of what happened. Finally, only OAA screens out defensive positioning effectively. OAA bases its model on batted-ball quality, location, and, importantly, defender location, whereas DRS and UZR only consider batted-ball quality and location. Although DRS and UZR retro-fit shifting back onto their models, it is an insufficient stop-gap. So, I’ll be using OAA for player performance.

(Aside: on a playoff broadcast I heard the play-by-play caller name DRS as the most reliable public defensive metric. That seems to me precisely opposite of reality. From what I can tell, DRS exaggerates the extremes of performance, good or bad. OAA and UZR are similar in magnitude of defensive impact. But DRS tends to put the extremes at 3x what OAA does. Part of this might be shift effects, but then it shouldn’t be so out of step with UZR. Anyway, I’m shocked and horrified that a pbp guy isn’t well read about analytics.)

So, OAA allows us to measure the performance of a defender regardless of his position before the ball is struck. We can distinguish between the defenders and the defense on the whole. By OAA, the Phillies defenders were among the worst in the league in the aggregate. They gave away 8 outs relative to league average, good for 26th in the league. That translated to -7 runs or a little less than a loss subtracted from a .500 record. In other words, if the Phillies had league average defenders they probably would have made the playoffs over the Brewers. The Phillies, led by so much talent, could only be kept from the largest playoff field in MLB history by a conspiracy of flaws.

First, let’s look at the good performers on defense. There’s really just the two: Jean Segura and JT Realmuto. After poor results at SS in 2019, Segura became the Phillies infield Swiss Army knife, which is especially impressive because moving down the defensive spectrum is not as easy as the positional adjustment for WAR makes it seem. To begin the season he took to 3B like a cat to an empty box. Then, when Alec Bohm was called up, he reestablished himself at 2B with ease. He even graded out well at SS in his limited time giving Didi Gregorius a breather. Overall, OAA credits him with 6 outs saved, just behind Fernando Tatis, Jr. and Nolan Arenado, whom Segura used, along with Matt Chapman, as a model for how to play third. Segura told reporters that from them he learned to play a deeper third to give himself time to react and trust his arm to make up for the distance. It must have worked. Indeed, despite playing back Segura saved an extra out when charging in.

JT Realumto, of course, continued to be a very good defensive catcher, although he didn’t contribute as much as he did last season. Fangraphs gives him 2 runs of framing but just league average stolen base suppression. When we look at the numbers, we see that JT suppressed the running game as measured by attempts but not as measured by success rate. In 2019, runners attempted to steal 92 bases off of him and were successful just half the time, well below the break-even point of 75%. In 2020, however, runners only chanced it 20 times, which prorates to about 55 over a full season, but managed to be successful at the break-even rate. Teams have learned how to pick their spots against JT, but I’m hesitant to draw any strong conclusions from this small sample. Except this: JT’s mere presence behind the plate controls the running game.

That exhausts the good fielders on the Phillies in 2020. Sure, Alec Bohm and Jay Bruce are credited with 1 OAA but that looks like a fluke. Certainly, if Bruce were given the opportunities, he would become an albatross in the outfield. As for Bohm, although I thought he showed soft hands and a strong arm, his movement looked slow and stuttered. I doubt he’ll continue to be an above average defender at third. Passable but not above average.

So, three players combined to produce 8 OAA. By dint of arithmetic, the rest of the fielders provided -16 OAA. Yikes. So, where were the biggest holes? Well, the -16 OAA was evenly divided between the outfield and the infield. In the outfield Bryce Harper, Roman Quinn, and Scott Kingery received passing grades. Adam Haseley and Mickey Moniak managed to deliver a combined -2 OAA in the corners given limited action. (Poor Moniak: he doesn’t have a carrying tool.) And Haseley also lost an out in center over a larger sample. But I’m sure you know that the biggest hole in the outfield was in left field as patrolled by Andrew McCutchen.

With McCutchen we all know the story, and the stats don’t add anything beyond quantification. He’s a sure handed, decent armed, smart outfielder whose lower half had been slowly betraying him coming into 2019. Even in 2019 he remained an average fielder in LF. But he tore his ACL last season, and his lower half aged rapidly through the recovery. He retains the athleticism to maximize his range despite his clear loss of speed, as this sliding catch demonstrates. But if I recall that ball had a catch probably above 50%. In fact, he only made one play on a ball with less than a 50% catch probability (out of registered 11 opportunities).

After 2020 McCutchen no longer looks like a starter on a playoff team. If the Phillies play him in LF for a full season, they risk giving away a win just from his defense. Since he’s now about a league average hitter, he’s better suited to a bench role or the short-side of a platoon, sad as that is to say. McCutchen exudes so much joy playing baseball that I wish he could play all the time and forever.

Unfortunately, for the Phillies they don’t have a ready replacement for him, nor the ability to move his contract and sign someone in his stead. Haseley has turned himself into a slap hitter and hasn’t look good in either corner outfield position. Quinn can probably handle LF defensively but hasn’t hit well enough to be, well... to be on a competitive roster. Most likely, leftfield will be a problem in 2021.

On to the infield. Didi Gregorius gave his usual performance, just below league average. So, overall, the left side of the infield played well enough. Unfortunately, when Jean Segura wasn’t on the right side, it was a fairly gaping hole. Both Rhys Hoskins and Scott Kingery gave away 4 outs each. For Hoskins part, the poor performance matches his reputation, even though the stats from the last couple of seasons did not. In 2018-19 he graded out as a league average 1B; in 2020, no longer. I myself have never seen Hoskins as particularly poor, although I’ve talked to enough people who do that I am willing to regress my own perceptions. So, maybe OAA is finally reflecting what others have seen all along. If that’s the case, Hoskins is a candidate for full-time DH, despite the penalty associated with DHing. If the NL retains the DH for 2021, the Phillies should consider moving Bohm to 1B, playing Segura at 3B, and, per force, signing a SS. But perhaps Hoskins just had an off-year and will return to being league average. In that case, the Phillies can stick to their ostensibly preferred route: keeping Bohm at 3B at least one more season to see if he can learn to play the position well enough to stick there through his twenties.

Perhaps, however, they should prefer moving Bohm to 1B. Doing so could improve the team’s overall infield defense, even if Hoskins returns to past norms. Bohm showed good range and hands at first and can obviously make the throws to second. But that move would force the Phillies to play Kingery at 2B, and he was the other big hole. Kingery struggled to range both up the middle and towards first. His legs, which were supposed to be his sine qua non, failed him. When Kingery was coming up his defense at 2B was a plus tool. Many argued that his development stalled in MLB because he was playing everywhere except his “natural” position. (I don’t like the concept of a natural position.) Well, he didn’t do anything to bear that out. Now, we have to take this Kingery season with a lot of salt, despite his protestations, because he was likely suffering from the long term effects of COVID-19. At the end of the season, specifically in the series against the Rays, Kingery looked much better, even laying out to steal a hit on the sort of looping line drive that fell all too often against the Phillies’ pitching staff.

So, what should the Phillies do? Ultimately, that depends on the sort of knowledge that comes with close acquaintance. The sort of knowledge we don’t have. From where we sit, we can’t tell whether Kingery’s struggles at second were extraneous or essential. If the Phillies keep Bohm at 3B, it would be best not to move Segura back to SS, in which case Kingery is best suited for CF. Defensively, he’s the Phillies best option in center anyway. If the Phillies move Bohm to 1B, then Kingery has to take 2B. If Kingery is as bad as he looked this season, that’s not an option. (And if the NL does not retain the DH it is even more not an option.) Plus, Kingery needs to hit much better to secure a starting job on even a mediocre team. All this to say, the Kingery question is deeply unresolved.

On that cheerful note, I need to close this post. I haven’t yet covered the Phillies shifting quality. That will have to wait for a second part. For now, know that the Phillies allowed approximately 8 extra hits due to the performance of the defenders on the field, given their positioning before the pitch. That leaves up to 17 extra hits allowed due to shifting. In a subsequent article I will dive into that problem.