The 2019 AL Cy Young Voting Guide

With just over a week to go in the regular season, Justin Verlander and Gerrit Cole are running neck and neck as favorites for the American League Cy Young award. Verlander leads the league in innings (212) and ERA (2.50). Cole has the lead in strikeouts with 302 with Verlander 19 behind. Even after accounting for Verlander’s 34 homers, his 3.28 FIP is still one of the best marks in the league. On Cole’s side are the strikeouts, a league-leading 2.78 FIP and a 6.7 WAR half a win clear of Verlander. Several other pitchers, like Charlie Morton and Lance Lynn, boast strong resumes, and with five slots on voters’ ballots, many pitchers will receive down-ballot consideration worthy of discussion.

While awards voting is a mostly objective process, when trying to differentiate between a group of very good pitchers, personal preferences are likely to play into the selections. When voters rely on particular stats, be it FIP, ERA, or some other metric, they are making decisions about the importance of defense, park, opponent, and how much talent a big league pitcher is expected to exhibit when it comes to contact quality. Before we get to all of those issues, let’s identify the candidates. There’s a fairly clear top seven among AL starting pitchers (Liam Hendriks might deserve some consideration as well) with Eduardo Rodriguez also included due to his rank based on Baseball-Reference’s WAR.

Here are the eight pitchers under consideration, with some traditional and more advanced statistics:

A look above shows Gerrit Cole leading in the more advanced statistics, with Verlander gaining the nod from traditional metrics, and Lance Lynn and Charlie Morton sort of splitting the difference between the two Astros. Shane Bieber and Lucas Giolito are a bit behind, with Giolito unable to add anything to his file after being shut down for the season. Mike Minor’s case is made by his low ERA combined with his difficult park, as his strikeouts and walks lag behind the other candidates. Eduardo Rodriguez is the poor man’s version of Minor.

If we looked at FanGraphs WAR, we’d see Cole as the leader due to his incredible strikeout rate and ability to limit homers, at least somewhat. Though he has a 20-inning deficit compared to Verlander, the strikeouts and homers make enough of a difference for Cole to take the day. Verlander and Lynn are in a dead heat when it comes to WAR, with the huge difference in home runs balancing Verlander’s lead in strikeouts and walks and Lynn’s more difficult park in which to keep balls in the field of play. Comparing Lynn to Morton, we see Morton with the homer advantage, but the innings deficit, combined with Tampa Bay being a hard park to homer in, gives Lynn the edge.

If you believe in the importance of getting strikeouts, inducing pop-ups, and preventing walks and homers, and that the park and league one plays in should be the factors when considering how good a pitcher has been, then you don’t really need to look further than the WAR totals above, which have Cole in front, Lynn and Verlander a tossup, and Morton bringing up the rear. This view ignores what happens on non-pop-up batted balls that stay in the field play, however. There are a few reasons for taking this route; it’s difficult to separate pitching talent from a player’s defense, and to confidently ascribe the results of those batted balls to pitcher talent rather than mere luck. Still, there are other ways of evaluating pitchers.

We will get to contact quality in a moment, but first let’s take a look at the different versions of WAR. Below are the versions of WAR found at FanGraphs (our FIP-based version along with RA9-WAR), plus the versions from Baseball-Reference and Baseball Prospectus:

In the average above, RA9-WAR and BRef WAR are similar enough that they were counted as one for the result. If you trust the general framework of WAR, but don’t want to put too much emphasis on a FIP-model used here, the runs model used at Baseball-Reference and in RA9-WAR, or the more complicated DRA-based version of Baseball Prospectus, a simple average puts Verlander in the clear lead with Cole a clear second and Lynn an easy third choice, with the next four contenders all in a bunch. If that approach isn’t for you, we need to dig down to show exactly what each metric emphasizes.

Baseball-Reference’s version of WAR takes more information into account, including defense and opponents, which is how Mike Minor fares so well. The Rangers play in a difficult park and face difficult opponents and don’t have a great defense. Minor’s success in run-prevention despite those difficult circumstances causes him to lead Baseball-Reference’s version of WAR, though he falls behind in the other versions. This was a major issue in Aaron Nola’s Cy Young case last season, as the adjustment for a poor defense might not have been warranted. Baseball Prospectus’ WARP attempts to isolate “every pitching event” and credit pitchers for their role in those events.

In many ways, the versions of WAR are all trying to do the same thing, which is to credit a pitcher for certain outcomes based on the pitcher’s work. This isn’t new. It’s why earned run average tries to strip away the unearned runs. ERA, Baseball-Reference’s WAR and RA/9 WAR both look at the number of runs and then work backwards to try to arrive at a deserved result. FIP-based WAR looks at the outcomes most controlled by the pitcher (walks and strikeouts) and then adds or subtracts credit for some batted balls via the home run and infield flies, and gives credit for all outs made. Baseball Prospectus looks at the most likely outcomes given the circumstances and assigns a value.

One piece of information we haven’t yet discussed, but which might be illuminating, is the data provided by Statcast. Using xwOBA, we can take a closer look at the results on batted balls to find out how much of a role luck or defense might have played in the results a pitcher experienced. In addition to those batted balls, xwOBA uses strikeouts and walks to come up with a single expected wOBA. It’s very similar to FIP, except instead of using home runs, it uses expected outcomes for all batted balls based on launch angle and exit velocity. Here’s where the Cy Young contenders sit this season, along with their actual wOBA and the difference between the two:

Verlander and Cole again sit head and shoulders above the rest, but Charlie Morton is a strong third. In terms of luck and defense, Lynn, Cole, and Rodriguez have experienced some bad luck or poor defense, while Shane Bieber has benefited. The above numbers factor in the result of every plate appearance, but if we want to look just at contact, we can remove walks and strikeouts:

We still see Cole, Lynn, and Rodriguez experiencing some bad luck with Bieber getting help. What’s interesting here is how bunched up most of these pitchers are on contact. Six of the eight’s xwOBAs are within 20 points of each other, with Giolito a little further away and Bieber pretty far off. If we remove home runs, we get an outlier in the group:

When we remove Verlander’s home runs, instead of looking a bit lucky (10 points better than the league differential), his results on balls in play look incredibly fortunate. Looking only at Verlander’s homers for contact might not be ideal, but considering his results on balls in play and the resulting runs isn’t going to isolate his talent, either. Lynn’s results on balls in play look very unlucky, while Cole moves from being unfortunate on all batted balls to fortunate when homers are removed. While Cole gave up a decent number of homers, a few might have been a bit unlucky, which was actually Verlander’s problem last season.

Verlander has earned his homers and been lucky on balls in play, but when combining Verlander’s batted balls with his strikeouts and walks, he and Cole stand above the rest of the league. The gap between Lynn’s very good FIP and decent ERA is easily explainable by bad luck on batted balls in play. Morton’s contact has more or less evened out. How much you rely on xwOBA depends on how much control you believe a pitcher has over his batted balls. It removes the defense, save for the extra batters a pitcher with a poor defense has to pitch to, and it should remove consideration for parks, but it is not very results-based outside of the walks and strikeouts. Still, it’s another tool to use when trying to separate pitchers who have had similar seasons.

The last few tables don’t incorporate innings, and getting more outs is better, so I created a rather crude xWAR based on xwOBA. These are the results for all AL pitches with at least 100 innings (conclusion after the table):

If contact quality is important to you, then Verlander and Cole are the clear 1-2 in the Cy Young race, with Morton, Giolito, and Lynn in a second tier. If pure results are key, look to RA9-WAR. If you want those results adjusted, Baseball-Reference can help. If you want a more dialed down version of the WAR based on expected contributions, Baseball Prospectus is there to help. If you’re skeptical that we can reasonably isolate defense or contact quality as a skill in a single season, FanGraphs’ FIP-based WAR can help by using strikeouts, walks, and only the strongest and weakest contact in the form of homers and popups. When choosing between a set of very good pitchers all of whom are having very good years, you have to decide what you believe is most important in terms of results and deserved results, and how much control a pitcher has over individual aspects of the game. We try to be as objective as possible, but a little subjectivity is bound to creep in when making these kinds of decisions.