Defense, Contact Quality, and the NL Cy Young

This year’s National League Cy Young race invites multiple interesting questions about how best to evaluate pitching performance. Jacob deGrom, for example, is the league’s leader in ERA by a healthy amount; however, he’s also recorded only as many wins as reliever Jeremy Jeffress. Max Scherzer is having another great season, but his .255 BABIP compels one to consider whether his 2.31 ERA is the product of luck or defense (although the Nationals have recorded below-average defensive numbers both by UZR and DRS). Aaron Nola, meanwhile, has recorded a similarly low BABIP even as Philadelphia has produced NL-worst figures both by UZR and DRS. Finally, while the race has been viewed as a three-person contest for some time, it’s also possible Patrick Corbin has inserted himself into the conversation with a fantastic second half.

Sorting through the candidates is difficult. Ultimately, one’s choice for Cy Young will depend on how one weighs what a pitcher can and cannot control — and how best to quantify those effects. To start, here are some general metrics that should be familiar to FanGraphs readers.

Based on these numbers, Jacob deGrom is the pretty clear favorite for Cy honors, with Max Scherzer an equally clear runner-up. What’s less clear, however, is that the results of a vote would produce a similar outcome, as both pitcher wins and other versions of WAR are likely to influence writers — and arrive at different conclusions than the figures here. Below, I’ve included some different versions of WAR, each of which paint the field in a different light.

Here we see a version of reality that suggests greater parity in the race. Averaging the numbers above, we’d still put deGrom first, Scherzer second, and Nola third, but Scherzer actually places ahead of deGrom in two of the four metrics, while Nola and Scherzer are more closely situated. Examining how each of WAR metrics arrives at its destination can help inform how to use them. Last week, Eno Sarris took a look at some of these same issues in a discussion of how large a role luck ought to play in Cy Young voting. There is also the question of what defines “luck” in the context of pitching, what sort of control a pitcher exerts over certain outcomes, and what role a a pitcher’s park ought to play in our evaluations of him.

The metrics above all feature different inputs which, naturally, lead to different results. In the version of WAR used at FanGraphs, those inputs are innings, strikeouts, infield flies, walks, and home runs — along with factors for league and park. DeGrom leads by this particular measure because his strikeout, walk, and homer numbers are all great. Scherzer has good walk and strikeout numbers but a closer-to-average home-run rate. Nola features slightly inferior (although still excellent) strikeout and walk numbers — plus a good home-run rate — but he falls behind Corbin, who has good numbers in all three.

The next metric, RA9, is another version of WAR carried at FanGraphs — one which, in this case, simply considers the number of runs a pitcher allows while also factoring for league and park. That’s how Nola, with the very good ERA, jumps up near Scherzer, though still short of deGrom. RA9 includes runs that were scored or not scored due to defense and sequencing, but does not try to make any adjustments for those factors.

Baseball-Reference begins with something like FanGraphs’ RA9 calculation but makes further adjustments for opponent and team defense, which is a significant factor in this year’s race. Nola tops the Baseball-Reference WAR leaderboard because of how well he’s prevented runs despite Philadelphia’s poor defense. Generally the effects of these defensive adjustments are muted, but because Nola appears to be headed for one of the 10 best bWAR seasons of the last 50 years, this case invites some scrutiny. Patrick Corbin suffers from the opposite scenario: Arizona has recorded strong defensive numbers, meaning he receives a “penalty” of sorts for his contribution to run-prevention.

Here are the overall team defense numbers by DRS, which Baseball-Reference uses, and UZR, which is included in WAR for position players but not pitchers here at FanGraphs.

There is obviously a much larger spread with the DRS figures, as defensive adjustments alone mean a difference of 24 runs between Nola and Corbin, which is about four times as much as the difference by UZR.

Over at Baseball Prospectus, their Deserved Run Average (DRA) metric accounts for as many aspects of a pitcher’s game as possible and attempts to factor for everything including park, opponent, catcher, umpire, and pitch effectiveness to determine how many runs a pitcher should have allowed with all those variables rendered neutral. By their methods, Scherzer leads over deGrom, with Nola and Corbin a ways behind.

There’s certainly an argument to be made for considering the strength of a defense behind a pitcher, and reason dictates that a defense can help or hurt a pitcher’s run-prevention numbers. Defense alone, however, isn’t going to fully explain the difference between a pitcher’s FIP and ERA. Luck is involved, as well. We can use Statcast information to determine just how much defense and luck might be involved, though it won’t do a good job separating those two factors. For starters, here are the xwOBA and wOBA figures for each of the pitchers above.

In terms of what a pitcher has deserved to concede based on quality of contact, strikeouts, and walks, Scherzer has gotten just about what we might expect, while deGrom and Nola aren’t far off expectations. Corbin is the outlier here, and there is a case to be made that Arizona’s defense is partially responsible for his good fortune. What’s interesting, though, is that Corbin’s ERA is actually much higher than his FIP. This could mean that Corbin has been rather fortunate this year on home runs or that the contact he’s conceded on balls in play has been of higher quality than the sort conceded by other pitchers.

We can remove the most skill-based aspects from above by taking out strikeouts and walks and looking at xwOBA on just batted balls.

Here we see almost no effect on Scherzer’s outcomes, with a slight benefit for deGrom and Nola, and then a big help for Corbin. You’ll note that the league-wide numbers are off by 15 points from each other, likely due to a potentially dead baseball, as the estimates on launch angle and exit velocity are based on previous seasons, when the ball was perhaps a bit more lively. As we are looking at numbers between pitchers in this season alone, the comparisons still provide value. What happens when we remove home runs and look solely at batted balls? See below.

In theory, these numbers factor in both defense and luck on batted balls this season. As we can see, it appears that, whatever poor defense has victimized Nola has likely been evened out by good fortune. The same is true for deGrom. Scherzer, meanwhile, appears to have received a slight benefit, with Corbin being the recipient of some good defense in Arizona. This probably doesn’t leave the reader with any definite conclusions. We have a better idea about the quality of contact and how defense might have affected run totals — which is to say not much — but the extent to which a pitcher exerts control over that contact is also a matter of debate.

If you believe that a pitcher controls very little of opponent contact — or, alternatively, are unsure of the level of control — the version of WAR hosted here at FanGraphs is your main resource. If you believe that a pitcher is wholly responsible for the quality of contact he concedes and also that defensive quality doesn’t move the needle much in one direction or another, RA9/WAR makes some sense for you. If you believe further adjustment needs to be made for defense, bWAR can provide some help. If you want a more granular look at individual pitches, DRA provides guidance. If you just want something based entirely on xwOBA, a crude attempt is made below.

While the question of value is somewhat objective, there is some subjectivity involved, but if making a decision on the Cy Young, it’s important to have as much information as possible to determine why one pitcher might be better than the other. It isn’t enough to simply prefer one stat over another and blindly rely on it because you generally agree with the methodology. Look at how the results are reached to make the best possible decision.

*****

As promised in the final paragraph above, here’s a rough approximation of WAR based on xwOBA: