Who can be trusted on the mound in big October moments?
After a catastrophic end to the month of July, the Yankees seem to be righting the ship. After rattling off nine consecutive victories from trade deadline day onwards, the team is now 10-2 in the wake of yesterday's win against the Blue Jays.
It's easy to forget just how dire things were looking on July 31, when Cashman stood pat despite the abysmal performance of the starting rotation. Yankee World was apoplectic over the seeming indifference of the front office to the obvious need for reinforcements to the pitching staff. But following a couple weeks of strong performances, now's a good time to take a step back and analyze the team's options objectively as we hurtle toward October.
A few weeks back, I wrote a review of the past five World Series winners, trying to discern what actually makes a championship-caliber pitching staff. I urge you to read the article for the full scoop. But the key takeaway for this year's Yankees was that while a strong bullpen can compensate for a weak rotation, it must be complemented by starters that can keep teams competitive into the middle innings of games. While this isn't groundbreaking, it does dispel the misconception that a super bullpen can simply erase the need for a competent rotation.
So with the final cast for this year's playoff dance all but set, let's break down how Aaron Boone can optimize the arms he has at his disposal to carry this team to championship number 28. In order to do so, we're going to hone in on two key elements of a pitcher's makeup that can help dispassionately assess their utility in October.
The first is simply their performance on the field, and what the underlying metrics tell us about their true talent level this season. For this aspect of the evaluation, we will use metrics like xFIP, SIERA, and yes, even ERA to analyze who has been effective enough to deserve our confidence in the most important games of the year.
But unlike some nerds, I still find some value in the abstract concept of wins, holds, and saves as a measure of how well pitchers manage to keep their teams in the game. Though there are too many external factors to make these stats a valid evaluation tool, it can still tell us something about the consistency and competitiveness of a pitcher's performance. But the old-fashioned construction of these stats is frankly illogical. Can't win a game unless you pitch 5 innings? Arbitrary lead in the 9th means you get a save? Surely there is a better way to get at what these stats claim to tell us.
And indeed, I would argue there is. A few weeks back, Andrew mentioned the notion of WPA Wins and Losses on the podcast. For nerds like me who want the full Fangraphs primer, the article on this stat is here. For those who don't care to read it in depth, here's the main idea.
Every play in a baseball game increases or decreases a team's probability of winning. A home run late in a tie game has a huge impact on that probability. A two-out walk with no runners on in a blowout has almost no impact. Fangraphs tracks the change in win probability for each play, and credits the pitcher and batter with equal and opposite WPA that corresponds to that change in probability. For example, when Brett Gardner hit a go-ahead double in yesterday's game, he was given 0.123 WPA (12.3% win probability added), while the pitcher Thornton received -0.123 WPA. The WPA winner or loser is simply the pitcher on the winning team with the highest WPA, or the one on the losing team with the lowest WPA.
This makes much more sense than the classic construction of wins and losses. If a starter allows six runs but his team is up 7-6 when he leaves the game, why should the reliever who shuts the game down with a tight lead not get the win? And the same with losses -- why should the starter who goes seven scoreless get the loss if a reliever allows an inherited runner on first to score in a 0-0 game?
If you agree with this notion, you'll like Shutdowns and Meltdowns too. This is the WPA equivalent of holds, saves, and blown saves, and it applies to relievers. If as a reliever, you add or subtract 6% win probability in your outing, you are credited with a Shutdown or a Meltdown. This captures the fact that a reliever getting only one out with the bases loaded in a one-run game is actually playing a more important role than a closer who gets a save up 3 in the 9th. Ultimately, both WPA Win/Loss and Shutdowns /Meltdowns paint a valuable picture of each pitcher's performance in context and their ability to help their team win games.
Whew, that was a long intro. But hopefully, if you've stuck with me, you've bought into the idea that both a pitcher's true talent and their performance in context are key to deciding who is trustworthy in the playoffs. So with this in mind, let's analyze the Yankees starters and relievers to build the perfect playoff blueprint for Boone and the nerds.
The Starters
There are two things that are obvious about the rotation, one good and one bad. And both will come as absolutely no surprise to fans who have been paying attention this year. The first is that Domingo German has been the Yankees' best starter by a long shot, and should absolutely be the team's #1 starter despite his inexperience. The second is that JA Happ should not set foot on a mound in October barring disaster.
German has pitched to a miraculous 12-1 WPA record, in line with his true 15-2 W-L. He has been without question the most consistent Yankees pitcher, averaging nearly 6 innings per start and pitching to a 4.12 ERA. He has managed to improve his main weakness, reducing his BB% to 5.5% from previous seasons in which he posted numbers of 14.5% and 8.8%. While plagued by the home run like many Yankees starters, he still posts solid peripherals with a 4.07 xFIP and 3.99 SIERA. If the Yankees actually intend to win in October, they will ignore innings limits and allow German to anchor the rotation in October much as he has all year.
Happ has no business pitching in anything resembling an important game, let alone a playoff game. His 4.91 xFIP and 4.84 SIERA are legitimately offensive, and untimely home runs have pushed his actual ERA even higher to 5.48. Though he has salvaged a 5-6 WPA record, there are few redeeming qualities to Happ's profile right now. He should be benched in October, and used only in blowout situations (and maybe not even then).
Next, we come to two guys the Yankees will have no choice but to use: James Paxton and Masahiro Tanaka will be in the playoff rotation out of necessity, like it or not. I would argue that fans should actually be excited for the potential role Paxton might play in October. As my colleague Sean constantly points out, Paxton is a very good pitcher who has been plagued by bad luck and inconsistency this year. His 4.02 xFIP and 3.94 SIERA are even better than German's, and his 29.4 K% puts him in elite company when it comes to missing bats. But he has been unreliable, as his mediocre 6-5 WPA record demonstrates. At his best, Paxton has truly ace-like ability that can provide tremendous value to this club. Boone should plug him in at #2 in the rotation to bet on this upside.
That's especially true because Tanaka comes with all of the same risk and inconsistency with less of an upside. Yes, I know he's been nails in the postseason in the past and pitched lights out yesterday. But the reality is that Tanaka is just not a very good pitcher. His 4.39 xFIP and 4.53 SIERA are mediocre at best, and his propensity to melt down and allow home runs in the middle innings does not inspire confidence in the postseason. He struggles to strike out batters and has an xwOBA in the bottom half of the league, a terrible combination that means more baserunners and an inability to escape jams. The silver lining is that, at 9-5, he has shown an ability to keep this Yankees team competitive and pitch with length. Boone will have to bank on that track record and a little playoff magic when he plugs Tanaka in at #3 in the playoffs.
So what about the fourth starter? You have to begin with CC, whose illustrious Yankee career will be coming to a close after this year. Much as I love him, his 4.96 xFIP and 4.75 SIERA simply do not play in must-win games. Not only is he not the ace from 2009, but he's no longer the reinvented, crafty lefty of recent years. At this point, CC is a replacement-level starter who can add experience and fire from the dugout. But he sadly should not be out on the mound in October.
That leaves Boone with two choices, but that choice will be made for him in due time. I of course am referring to the health of staff ace Luis Severino, who we are told is progressing well in his long road back from a Spring Training shoulder injury. At his best, Severino is a true superstar, with a career 3.28 xFIP and 3.44 SIERA that puts him up there with the very best in the game. Everyone remembers his transcendent first half last year, when he was on a Cy Young trajectory and practically guaranteed a Yankees win when he took the mound.
But should his health fail (don't worry, I'm knocking on every piece of wood with crossed fingers), the Yankees will likely turn to an opener in Chad Green. Since his return from AAA and assimilation into the role, Green has a 1.98 ERA, 3.27 xFIP, and 2.99 SIERA. He's been downright filthy in this position, and is part of why the Yankees only lost their first opener game on Saturday. That being said, it's hard to imagine Boone going to Cortes or Tarpley in a must-win playoff game (he also shouldn't). The reality is he will turn to the main arms in his bullpen, which may increase the chance of victory but would likely be unsustainable throughout a full World Series run.
In fact, the best-case scenario is likely Chad Green opening for Luis Severino, who will be on a reduced pitch count even if he does return to health in time. Otherwise, should Severino not return, the Yankees will be left in a bind for their #4 starter or forced to try a 3-man rotation, a tactic that has not proven successful in the current era of baseball.
The Relievers
Of course, the Yankees will have the luxury of four of the very best relievers in baseball locking down games in their bullpen. I refer to Aroldis Chapman, Adam Ottavino, Zack Britton, and Tommy Kahnle, all of whom have been magnificent this year. Kahnle has arguably been the cream of the crop, pitching to a 2.48 xFIP and 2.60 SIERA thus far. His 20/6 Shutdown to Meltdown (SD/MD) ratio demonstrates his consistent excellence, with 5 WPA wins to boot. Kahnle, not Ottavino, should be employed in the David Robertson role of years past, inserted in the most high-leverage situations during the game. Boone has failed to do this so far this season, but would be wise to continue transitioning Kahnle into the role through the remainder of the regular season.
This is not to disparage Ottavino. He and Britton have both been terrific this year as well. However, they fall significantly behind Kahnle in terms of context-neutral performance this year, with xFIPs and SIERAs around 4.00 that reflect their battles with command. This has not damaged their ability to lock down leads, as they have 24/5 and 19/9 SD/MD ratios respectively. But in my opinion, they should clearly fall below Kahnle in the leverage pecking order to #3 and #4.
This leaves Chapman, who in an ideal nerd world would be used as the #2 leverage guy behind Kahnle. However, I am not naive, and actually don't object to him remaining in his traditional closing role for which he is well-suited. Boone should show dexterity by using Chapman in key situations outside the 9th should the need arise (basically, don't give me Jonathan Holder over him just because it's the 7th). But with three other elite guys, the need for Chapman to be stretched outside his usual end of game position should be limited.
Who else can be trusted with relief appearances in a playoff game? Well, Domingo German and Chad Green for starters (bad baseball pun, I know). These two guys both have experience in relief as well as starting, and should be called on sparingly should the situation arise. German is more questionable, given his already alarming innings increase and importance as the #1 starter for the team. But Green should without question be employed in relief as well as opening throughout the course of the postseason. His 6/11 SD/MD ratio looks ugly, but is heavily influenced by his disastrous early season performance. In fact, his ability to reign in his SIERA to 3.61 as a reliever shows that he can and should be trusted with late innings over the course of October.
While Jonathan Holder's 3.72 SIERA is actually 5th best on the team, it is impossible to get past his 13 meltdowns this year. Boone tried to a fault to use him in the late innings of key games, and got burned time and again. There is simply no appetite for him to be anywhere near a mound during a close playoff game. Use him to soak up innings with big leads or deficits, but keep him away from important spots if at all possible.
Lastly, would you believe me if I said Luis Cessa has basically been the same pitcher as Adam Ottavino this year? Their xFIPs and SIERAs are each separated by less 0.02 points, and incidentally Cessa comes out marginally on top in both categories. No, that does not mean Luis Cessa should be pitching the 8th inning in the ALDS. What it does mean is that in any game where a starter struggles and is forced out early, Cessa should unquestionably be trusted to hold down the fort for multiple innings to get the team back into the game. He has 8 shutdowns this year, a testament to the fact that he's successfully navigated that exact role during the regular season. He can and should be relied upon as a stopgap when the Yankees fail to get length out of a starter.
Final Thoughts
The Yankees will not have the best pitching staff in the playoffs. Not even close. But you don't have to squint very hard to see four capable starters and a top notch bullpen that, when paired with this explosive offense, can make it the distance in 2019. Despite the doom and gloom predictions, the season is in fact not over, and some very fun October baseball lies ahead.
But there is no doubt that an outsize amount of risk is being taken on by the reliance on Luis Severino's return. With him, the Yankees are probably co-favorites with Houston and the Dodgers. Without him, the rotation looks dangerously thin, and a single starter injury could completely derail their World Series hopes.
Betances is less critical, but nonetheless could play a major part in a potential championship run. With or without Severino, he simply lengthens the list of elite relievers that Boone can lean on to shorten games and ease the burden on shaky starters. Getting both of these guys back as even 80% of their best selves would go a long way toward making the chase for 28 a reality.
In sum, I would line up a healthy Yankees rotation like this:
Domingo German
James Paxton
Masahiro Tanaka
Chad Green/Luis Severino
And the bullpen would look something like this, in terms of leverage situations:
Tommy Kahnle
Aroldis Chapman
(Fully healthy Dellin probably goes here)
Zack Britton
Adam Ottavino
Chad Green/Domingo German
Luis Cessa
Jonathan Holder
Writing this article makes me both incredibly excited for October and absolutely terrified for the remaining 44 games of the regular season. So much can go wrong in terms of injury or setback from now until the end of September (again, knocking on wood). But if it doesn't, and if this pitching staff goes into the playoffs at full strength, the Yankees have every chance of making this their year. Here's to hoping that everything breaks right for this team as we enter the dog days of August.