Explaining My NL Cy Young Ballot

Congratulations to Jake Arrieta, who was just named the National League Cy Young Award winner, edging out Clayton Kershaw and Zack Greinke in one of the best Cy Young races in baseball history. Gerrit Cole and Max Scherzer rounded out the top five in the voting, just released over at BBWAA.com.

Back in mid-August, I was informed that I had been chosen to vote on the Cy Young Award in the National League this year; it would be my first time getting to vote on the pitcher awards after being an MVP and Manager of the Year voter last year. As a voter, I feel an obligation to the players to try and acquire as much information as I can in order to make the most educated decision possible, and so I spent the next six weeks working my way through various ideas and philosophies about how to evaluate a pitcher’s performance.

About a week after receiving the ballot, I wrote a piece exploring the idea of using a pitcher’s hitting performance as a variable in deciding the Cy Young Award, and I asked a lot of people in the game whom I respect about their views on that issue. A month later, I worked through some of my thoughts on how to best separate out a catcher’s value from his pitchers’ results, and then the next day, I looked at some other factors mentioned by the commenters, including the potential value of consistency.

In writing those pieces, I wanted to lay some groundwork for how I was attempting to put this puzzle together, because I knew that when the time came to actually show my ballot, no one was going to read a 5,000 word treatise on the complexity of evaluating pitchers with the tools we currently have available. But given the abundance of worthy candidates this year, even that kind of article probably wouldn’t provide enough room to lay out the case for each of the pitchers who had deserving years. In my view, there were three guys who all had seasons that would be easy choices for a Cy Young Award in a normal year, and picking between them was basically an impossible task.

While I eventually had to settle on a 1-5 order, a more realistic assessment of my view is that I turned in ballot with #1A, #1B, #1C, #4, and #5. As I write this, I don’t actually know who is going to win, and I’m not going to feel the wrong guy won no matter how the top three end up being ordered. You can make a strong case for any of them, and the gaps between them are so slight that I don’t think we should be arguing with significant conviction that there is a right answer. There are three deserving Cy Young winners in the National League this year, and I only wish there was some mechanism in place where I could have voted for a tie.

But, at the end of the day, we were asked to split hairs and determine an order. So, now that the voting has been revealed, here is a look at how I filled out my ballot, with some brief overviews of the reasoning behind those decisions. And by brief overview, I mean there are another 2,500 words to this post, but I left out some discussion of things I looked at that ended up not having a huge impact on my decision, such as quality of opposition, park factor adjustments beyond what is already calculated, and which umpires they pitched to, among other things. I’ve tried to lay out my thoughts as coherently as possible while also keeping this post on the shorter side of War and Peace. With that said, on to the ballot itself.

One of the best young pitchers in baseball, Cole turned into a dominating #1 starter this year, and when I began the process, I assumed I’d have him somewhere on my ballot. In the end, though, he fell just short, coming in at the sixth spot on a ballot with only five openings. While his numbers were excellent, a few small flaws in his resume conspired to push him just outside the top five.

1. He allowed 11 runs that were categorized as unearned

While Cole’s 2.60 ERA is certainly sparkly, rating fifth overall in the NL, his 3.07 runs allowed per nine innings (RA9) actually ranks ninth. In my view, the distinction of an earned run versus an unearned run is not meaningful, and wiping some runs allowed off of his record in an attempt to account for defensive performance behind him while ignoring all of the plays that defenders have an impact on is not a worthwhile endeavor. I think we should indeed attempt to untangle the pitcher/defender responsibility balance in looking at run prevention, but the earned runs/unearned runs dichotomy just isn’t helpful.

2. He was a poor batter, hitting .150/.177/.150 in 71 plate appearances.

In the end, I decided that this award was position-specific, not skill-specific, and that the intent was to honor the best pitcher of the year, not the guy who pitched the best. It might seem like semantics, but a pitcher’s job is more than just throwing the ball to the plate. Everyone agrees that a pitcher’s defensive skills are also part of his performance, since it contributes directly to preventing the opponents from scoring, so at the very minimum, we’re really evaluating a pitcher’s pitching and fielding.

But in the National League, pitchers also have to hit; it’s part of their job. It’s not a big part of the job, and most of them aren’t very good at it, but it is something they are required to do when they take the mound, and I chose to include offensive production as a minor variable in my decision making. Relative to the other candidates for the five spots on my ballot, Cole performed poorly at the plate, and gave his team fewer chances to win while he was on the mound because of it.

These are minor nitpicks in an otherwise excellent season, and if the ballot ran six names deep, I would have happily placed Cole in that final spot. But given how good the field was this year, these flaws were just enough to keep him out of my top five.

Reiterating that last point a bit, if I had voted solely based on pitching performance, Bumgarner probably wouldn’t have made the cut. He was seventh in the NL among qualified starters in FIP-based WAR, RA9-based WAR, and a 50/50 mix of the two, putting him in the close-but-not-quite group in terms of run prevention. But NL pitchers don’t just pitch and field, they also hit, and no NL pitcher hits quite like Madison Bumgarner.

With a .247/.275/.468 slash line, Bumgarner’s performance at the plate resulted in roughly 11 more runs scored for his team than you’d expect from an average pitcher. If Bumgarner were an average-hitting pitcher but we knocked 11 runs allowed off his total, he’d have had a 2.56 RA9, good for fourth best in the NL. From my perspective, Bumgarner’s offense on days he pitches is just as relevant to the discussion as his pitching performance, since his job is to help his team win, and scoring runs helps your team win just as much as preventing runs from scoring.

Bumgarner combined a very good pitching season with best-in-class offense, and the combination of the two was enough to slot him in at at the final spot on my ballot. He was actually quite close to finishing fourth, in fact, and I’d have no quarrel with anyone who wanted to place him at the top of the second tier of pitchers in the NL this year.

It speaks to the remarkable crop of candidates this year that one of the best pitchers in baseball could have the best year of his career — including throwing two no-hitters, and nearly a third, with two of them almost ending as perfect games — and not even really be in the mix for the top of the ballot. That’s just insane.

By almost any measure you want to use, Scherzer was better this year than he was in 2013, when he won the AL Cy Young Award. That year, he threw 214 innings and allowed 73 runs; this year he threw 228 innings and allowed 74 runs. His K%-BB% was 27%, the 10th-highest single-season mark for a pitcher in baseball history. Yeah, he gave up a few more home runs, but this was peak-Scherzer, a completely dominating season from a legitimate ace.

But this year, that just gets you a nice participation trophy. Max Scherzer couldn’t have done much more, realistically, but he just picked the wrong year to have the best season of his career.

And now we get to the meaty arguments. I know a lot of people think Zack Greinke should be #1, and I expect to take some flak for having him here. I’ll reiterate what I said earlier about seeing almost no difference between any of the pitchers in the top three, so this isn’t any kind of firm stance against Greinke being a worthy candidate, but in working through the process, I also came to the conclusion that I was less comfortable with the arguments for him than I was for the other two choices.

In the end, the case for Greinke essentially comes down to this: run prevention is entirely, or almost entirely, the function of the pitcher’s performance. If you believe that statement, then Greinke is your Cy Young winner. I don’t believe that statement. I think run prevention is a function of pitching and defense, with a bulk of the work being done by the guy on the mound, but the infielders and outfielders having a legitimate impact on the results as well. The trick has always been how to divide up the credit between pitcher and fielders, and the reality is we still don’t know how to do that effectively.

But we do have some tools that at least help point us in the right direction, and we’ve identified things that a pitcher can do to induce weak contact and allow fewer hits on balls in play than others. They can pitch up in the zone, generating a lot of infield flies and balls that land in the outfielders gloves; Chris Young and Jered Weaver are primary examples of this. Or they can just allow weak contact in general, allowing few hard-hit balls, so that their defenders have more time to make plays; this is a more rare skillset, but it does exist to some degree.

In 2015, Greinke didn’t really do either of those things, though, at least as far as we could tell. He posted a 48% ground ball rate and a 9% infield fly rate, so he doesn’t fit the Young/Weaver mold at all. And while Statcast didn’t record batted ball velocity or distance for every play in 2015, on the ones for which it did record data, Greinke’s 87.8 mph average exit velocity ranked just 39th among starters; his competitors for this award ranked #1 and #2, by the way. Tony Blengino, who has been working with this kind of data for years while serving as an executive with the Mariners, broke down the batted ball metrics in more detail and found Greinke’s numbers were not consistent with the idea that he caused his .229 BABIP through dominant pitching.

It’s certainly possible that, at some point, we’ll figure out that Zack Greinke did indeed pitch in such a way that his .229 BABIP was almost entirely a result of his pitching and wasn’t also influenced by some good performances from his supporting cast. And certainly, some of that batted ball prevention was the result of Greinke’s impeccable command, throwing pitches in hard-to-hit locations; he had a better year than FIP gives him credit for. But I don’t think we can reliably conclude that a pitcher has as much control over an out via a ball in play as they do via a strikeout, and so I’m not comfortable simply using metrics that equate all kinds of outs when attempting to isolate a pitcher’s contribution to run prevention.

In the end, the number that probably most convinced me to make Greinke #1C was actually not a number from his 2015 performance, but instead the .305 BABIP he’d allowed over his career up until this season. If Greinke had a longer track record of generating weak contact or showed some recent trend towards hit prevention, I’d have been more comfortable skewing towards his run prevention numbers. However, Greinke has actually been more hittable than average throughout his career, and it’s actually the other two guys in the race who have shown that they can give up fewer hits than FIP expects. And since they also performed better in the numbers that we’re more confident are capturing pitcher performance, Greinke ended up in third place on my ballot.

He had an amazing year, and again, I don’t see a significant separation between the top three spots on this ballot. I have no problem with anyone casting a vote for Greinke at the top of their ballot, and I’ll be happy for him if he wins. For me, though, I just don’t see quite enough evidence that he should get credit for the full amount of his hit prevention to move him past two other pitchers who also had remarkable seasons. And given that he missed a start down the stretch, putting him at only 32 games pitched instead of 33, that tiny sliver of a difference is enough to slide him to third on my ballot.

Last year, I voted for Kershaw for National League MVP, and he had a better year this year than he did in 2014. In fact, I came very close to putting Kershaw at the top of this ballot, flip-flopping him between #1 and #2 a bunch of times before sending in my final answer on the last day of the regular season. The resume is remarkable, and while strikeout rates are at an all-time high, racking up 301 Ks in 233 innings is still amazing. Remember how we noted that Scherzer’s K%-BB% was 10th-best all-time? Kershaw’s was fourth, behind 1999 Pedro Martinez, 2000 Pedro Martinez, and 2001 Randy Johnson, three of the greatest pitching seasons in the history of the game.

And it’s not like Kershaw struggled when hitters actually did put the bat on the ball. As he does every year, Kershaw held batters to a lower-than-average .282 BABIP, which is actually a little higher than his career average. So how did a pitcher with a 1.99 FIP and a lower-than-average BABIP allow a 2.13 ERA?

Timing. With the bases empty, Kershaw allowed a .282 BABIP. With men on base, it actually dropped to .278, so he didn’t choke when pitching under pressure, or struggle out of the stretch. But with runners in scoring position, his BABIP spiked to .341, which gets us right back to the previous discussion with regard to Greinke and how much weight you put on batted ball results. It’s definitely possible that Kershaw threw some really crappy pitches to the 147 batters he faced with RISP, and that .341 BABIP is the result of him simply pitching his worst when the stakes were the highest, but again, there isn’t actually a lot of evidence for that.

His walk, strikeout, and home run numbers are actually better with RISP than with the bases empty or runners on; he posted a 1.37 FIP in RISP situations this year. He also generated his highest infield fly rate with RISP (11%) and his lowest HR/FB rate (6%), which is why the Hard Contact rates tracked by Baseball Info Solutions have his Hard% lower in RISP situations (19%) than with the bases empty (27%) or runners on (20%). From Statcast, we know that the average exit velocity off of Kershaw this year (on balls tracked by Statcast) was 84.9 mph, tied for the lowest mark in baseball. And hitters have never been able to make good contact off Kershaw, which is why he’s outperformed his FIP by 20 points in his career up through 2015.

I asked friends with teams who have access to the raw Statcast feeds if their internal measures suggested that Kershaw gave up harder contact in RISP situations; they said no. And interestingly, the majority of them said that their internal WAR measures all had Kershaw as the league’s best pitcher this year. If there’s a smoking gun out there that shows that Kershaw’s .341 BABIP in RISP situations was the result of poor pitching, I couldn’t find it, and so I’m not terribly inclined to put much weight on the fact that teams got a good amount of hits off of him in high leverage situations.

So in the end, it was basically a coin flip, and Kershaw ended up in the #2 spot. He could have easily been #1. Maybe he should have been. I have more conviction that he belongs among the top two here than I do about where he belongs within that duo, and again, I really would have liked to be able to vote for a tie. But I couldn’t, and so in the final tally, the hair splitting of minutiae put him second.

If Kershaw is the FIP candidate, and Greinke is the ERA candidate, then Arrieta is the compromise between the two. By runs allowed, he’s half a win behind Greinke. By FIP, he’s a win behind Kerhaw. If you just average the two together, Arrieta comes out on top ever so slightly, though the difference isn’t meaningful.

But I think it’s instructive to recognize that Arrieta does have a history of inducing weak contact. In nearly 800 career innings, his BABIP is .272, and he matched Kershaw’s 84.9 mph average exit velocity, supporting the idea that his .246 BABIP this year might not have been as heavily influenced by his defenders as Greinke’s .229 mark. By BIS’s hard-hit measure, Arrieta was #1 in all of MLB, and he was #3 in generating contact they labeled as Soft, again ahead of the other two viable Cy Young choices.

These measures are definitely imperfect, and we may eventually look back and realize that the absence of launch angle or vector information makes these comparisons faulty. But right now, it’s what we have, and what we have suggests that Arrieta does have some hit-prevention skill, and should be given more credit for that than just looking at FIP would indicate.

So for Arrieta, I weighted his run prevention a bit heavier than I did with Greinke, which (along with working depeer into games and throwing more innings) was enough to make him the run prevention candidate on my ballot. And so, with all of the numbers so close, I barely edged over to the guy who allowed a lower number of runners to cross the plate. With Kershaw, I can see why assigning those runs to him might not be a correct evaluation of his performance, but with Arrieta, I don’t have to make that leap: he was amazing at retiring batters on his own and he stranded runners once they got into scoring position.

There is no wrong answer here. Greinke or Kershaw are also excellent choices, as is Arrieta. I went Arrieta-Kershaw-Greinke, but had ballots with every other possible combination you could imagine, and almost turned in several of them. No matter who won, I’m comfortable that I did everything I could to try and find a separator. I couldn’t, really, and I’m not sure one exists. This was just a year where three guys deserve the trophy.