Better defense should help these Philly arms

2018 Phils finished 30th in Defensive Runs Saved

January 7th, 2019

The Phillies defense in 2019 is going to be a stronger unit than it was in 2018. That much should be obvious, because the 2018 group just put up the weakest defensive numbers on record (dating back to 2002), and several of the moves Philadelphia has made so far this winter are clear defensive upgrades -- namely shipping out and so that can return to first base, and importing from Seattle to play shortstop rather than .

"I do expect our team defense to improve next year," Phillies general manager Matt Klentak said after trading for Segura in December. "I don't think you go from the bottom of the pack to the top of the pack in one offseason, but I do think you have to make every effort to improve wherever you can."

That's exactly right. It might not be a good defense, but it will be a better one, and it's reasonable to expect that Philadelphia's pitchers will benefit from it. But which pitchers, and by how much? Can we really assume improved defense will allow to do better than his already fantastic 2.37 ERA? 

Another way of saying that is "just because the defense was bad, it doesn't mean it was equally bad for every pitcher," because the numbers show that's not exactly true. Let's look at the five primary Philadelphia starters -- Nola, , Zach Eflin, , and Vince Velasquez -- to see exactly how the weak 2018 Phillies defense affected them, and what a better version may do in 2019.

Traditionally, there's long been two ways to look at how defenses affect pitchers: batting average on balls in play and unearned runs. Those are each imperfect for reasons we'll get to in a second, but as a quick-and-dirty indicator of what happened behind a pitcher, they're a decent start.
As you can see, the five Phillies pitchers were not affected equally. Let's start with BABIP, which is simply "on non-homers, how often did balls land for hits?"


Phillies BABIP, 2018
Pivetta: .328
Velasquez: .316
Eflin: .309
Phillies average: .303
MLB average: .293

Arrieta: .287
Nola: .251

Now, this ignores the quality of the contact a pitcher allows, and that obviously matters when we're talking about the difference between rocket line drives or softly hit grounders. We'll get back to that with Statcast™ data.

But to start with, you can see a pretty sizable difference between the five Philadephia pitchers. There's a 77 point difference between Pivetta and Nola, and it's difficult to believe that's entirely due to contact quality. Among 140 pitchers with 100 innings, Pivetta's mark was 10th-highest. Nola's was 10th-lowest. That's interesting. File that away.

Looking at unearned runs is interesting, too. For example, when looking at Nola, how can you possibly have the fourth-best ERA among qualified starters if you're pitching in front of a terrible defense? The easy way to do that would be to collect a ton of unearned runs -- that is, if your fielders make a large amount of errors behind you, then few of the runs you allow will be "earned" and inflate your ERA.

That's not an unreasonable expectation. For what little value errors have -- you can't get charged with one if you're too slow to even get to a catchable ball -- the Phillies had the second-most in baseball, and they allowed 63 unearned runs, the seventh-most. When they were torched 24-4 by the Mets on Aug. 17, a full 11 of those runs were unearned thanks to poor defense -- the most by a National League team in a single game in 35 years.

But despite that, Nola allowed only a single unearned run all season, that coming on Aug. 28 when a wild Santana throw home allowed to score. Arrieta, by contrast, allowed 17 unearned runs, the most of any pitcher in baseball.

Unearned runs, 2018
Arrieta: 17
Eflin: 7
Pivetta: 4
Velasquez: 4
Nola: 1

Regardless of the reasons why -- Arrieta would have you believe it's the shift, though it's unclear if that's actually true -- the traditional metrics show that the effects of weak defense were not evenly distributed in 2018.

We can do better than traditional metrics, of course. What do the advanced numbers show?

As we mentioned above, we can't completely give blame or credit to the defense for good or poor BABIP numbers, because that ignores what kind of contact the pitcher is giving up in the first place, and among qualified pitchers, Arrieta, Nola, and Pivetta all finished in the Top 15 in ground-ball rate.

In order to account for that, we can look at Expected Batting Average, a Statcast™ metric that looks at the exit velocity and launch angle for each batted ball and describes how often similar batted balls turn into hits. If we look at it on ground balls only, we can see how our five regular Phillies starters did.

As expected, Pivetta, Eflin, and Velasquez saw real-world outcomes worse than the quality of contact they allowed would indicate. But surprisingly, Nola and Arrieta each collected more outs than their already-strong quality of contact would have suggested. "Being good and having favorable outcomes on top of that" is a good way to get to those low BABIP numbers we shared above.

Expected average vs. actual average, ground balls, Phillies SP, 2018
+.096 // .231 expected, .327 actual // Pivetta
+.051 // .243 expected, .294 actual // Eflin
+.037 // .258 expected, .295 actual // Velasquez


-.014 // .222 expected, .208 actual // Arrieta
-.024 // .218 expected, .194 actual // Nola 

Since defense is more than just about grounders -- and since Hoskins's presence in left was a big factor as well -- let's also look at this on all non-homer batted balls.

Expected average vs. actual average, balls in play, Phillies SP, 2018
+.037 // .199 expected, .236 actual // Pivetta
+.019 // .219 expected, .238 actual // Eflin
+.016 // .213 expected, .229 actual // Velasquez


-.006 // .239 expected, .233 actual // Arrieta
-.012 // .192 expected, .180 actual // Nola

The numbers are slightly different, but the story is the same. Not every Phillies pitcher suffered from poor defense. Some may not have suffered at all.

That's partially how Nola managed to put up that sparkling 2.37 ERA, down from 3.54 in 2017 and 4.78 in 2016, without any major improvements in his underlying skills, aside from the fact that his four-seamer went from 91.0 MPH to 92.7 MPH.

2016: 4.78 ERA // 3.08 FIP // 25% K // 6% BB // 32% hard-hit // 55% grounders
2017: 3.54 ERA // 3.27 FIP // 27% K // 7% BB // 31% hard-hit // 50% grounders
2018: 2.37 ERA // 3.01 FIP // 27% K // 7% BB // 31% hard-hit // 51% grounders

That's oversimplifying, of course, because he improved his first-pitch strike rate and saw a notable drop in his left-on-base percentage. But a large difference is that his BABIP dropped from .334 to .309 to .251. The lousy 2018 Phillies defense may not have actually hurt Nola too much. It might have actually helped him.

ERA tells one story, but more advanced metrics tell others. We can look at Fielding Independent Pitching, which focuses on strikeouts, walks, hit batters and home runs; DRA, which tries to express a pitcher's "deserved" contribution based on a wide variety of inputs; and a Statcast™ version (let's call it xERA) which converts defense-independent expected batted ball quality and amount onto the ERA scale.

2.37 ERA // 3.01 FIP // 2.60 DRA // 2.78 xERA

4.80 ERA // 3.79 FIP // 3.40 DRA // 3.50 xERA

4.36 ERA // 3.80 FIP // 4.67 DRA // 3.63 xERA 

4.88 ERA // 3.75 FIP // 4.20 DRA // 3.68 xERA

3.96 ERA // 4.26 FIP // 4.08 DRA // 4.15 xERA

The takeaway here isn't that any one of them is "right." But every one of them believes that Nola and Arrieta's ERA was lower than their underlying performance indicates, and every one of them believes that Pivetta and Velasquez were better than their near-five ERA marks would show. (There's not such a clear outcome for Eflin.)

That being the case, it's tough to expect that better defense will do all that much for Nola in 2019. But it might be reasonable to expect more from Pivetta, who made it onto our "2019 breakout pitchers" list in part due to his outstanding strikeout rate.

After all, it was Pivetta who was on the mound when the defense behind him made plays like this ...

... and this ...

... and this:

That's not to say that the others didn't see the same, of course. Here's a runner scoring from first base on a "single" against Nola because and couldn't collect a popup. Here's Eflin falling behind in the first inning because threw a ball away. Here's Velasquez, with two outs, allowing two more runs because couldn't corral a fly ball. Here's two runs scoring on Arrieta because a hard-hit ball got right through Kingery. It happens to everyone.

But it doesn't happen equally, is the point, and it's part of the reason why it's difficult to believe the Wins Above Replacement ranking at Baseball Reference that has Nola as baseball's best pitcher in 2018, because he's getting credit for posting a 2.37 ERA in front of the same poor defense other Phillies pitchers had, even though the evidence shows that awful defense may not have existed for him.

There's plenty of reason to believe the 2019 Phillies defense will be an upgrade over the 2018 version, especially if they do end up adding Manny Machado to the left side of the infield. It's a big deal for a team hoping to make big moves forward this season. It should help their pitching, obviously. It just might not help every pitcher in the same way, especially with some evidence that the non-Nola/Arrieta trio already had to deal with worse defensive lineups behind them, particularly with the "Santana at third base" experiment.

It's hard to see improved defense helping Nola that much. If you're Pivetta, however, and you can get that inflated home run problem under control, you might have cause to be excited. Very excited.