On bullpen roles, analytics, and Aroldis Chapman’s new usage plan

Earlier this week, Joel Sherman reported that the Yankees plan to use Aroldis Chapman in a slightly different role at times this season. Whereas Chapman has historically been used nearly exclusively in ninth inning save situations, Aaron Boone has indicated that, from time to time, he will use Chapman in the eighth inning of a close game rather than saving him for save situations or getting him work at the end of blowouts to keep his arm fresh. It likely won’t happen particularly often, but it’s certainly a welcome approach to ensuring Chapman remains dialed in throughout the season.

This change in usage is likely possible, at least in part, because of Zack Britton’s absence at the back of the bullpen this year. Historically, teams and fans alike typically knew exactly what was coming their way when the Yankees entered the end of the game with a slim lead: they’d see Britton try to hold the lead in the eighth inning and Chapman try to nail down the save in the ninth, with very little deviation from the script. Now that Britton is on the shelf for the year, the eighth inning role — traditionally reserved for the “setup man” — is wide open.

It would be an understatement to say that this news got me thinking about bullpen roles, the Yankees, and how they differ between analytics and the players. When I pitched this article, Josh, Andrew, and I had a discussion about how, despite the fact that the infusion of analytics are so heavily impacting seemingly every area of the game, bullpen roles remain relatively untouched. For the most part, you know which pitcher you’re going to see on the mound depending on how the game script plays out.

While the Yankees might be frustratingly set in these roles, they certainly aren’t unique in seeing value in traditional bullpen roles. Last winter, the Chicago White Sox gave fiery closer Liam Hendriks a three-year, $54 million contract with an escalating base salary to exclusively pitch the ninth inning and even traded for Craig Kimbrel to become his setup man at last year’s trade deadline. This season, the Atlanta Braves will be paying Kenley Jansen a handsome $16 million for his ninth-inning services. On the flip side, we’ve seen what a poorly constructed bullpen can lead to, as the Blue Jays had one of the best run differentials in the league last year but missed out on the playoffs, and the Mets and Phillies have been, well ... the Mets and Phillies.

My dream bullpen, as a numbers guy, is one that is completely role-less. In theory, saving your best relief arm for the ninth inning while leading simply doesn’t make sense to me, since the most crucial moments in a game aren’t always reserved for the ninth inning. I mean, this is obviously a simplified version of this issue, but what’s a more pivotal spot in a game: down one with the bases loaded and nobody out in the fifth inning, or up three with nobody on base in the ninth? Clearly the moment earlier in the game is the higher leverage situation, so naturally you’d want your best arm on the mound to get you out of that jam, wouldn’t you? Think of how Mariano Rivera was used in 1996, or how Jonathan Loáisiga was used last season. That’s my dream bullpen. To my analytics-focused mind, I want the best arm available up in the biggest spots, regardless of when those moments happen.

But, as we know, it’s really not as simple as that, is it? It’s no secret that pitchers are creatures of habit, and a number of them rely so heavily on routine, whether it’s throwing at the exact same time pre-game or doing the exact same workout post-game, that they believe their performance has the potential to suffer if that routine is broken. Hell, even when I pitched at the lowest level of baseball possible when I was younger, I needed to run through the exact same pre-appearance rituals every time for fear of blowing my appearance. Because of this added human element, I don’t think it’s particularly outlandish to suggest that, despite a perceived negative effect on overall value (at least as value exists in a vacuum) from an analytics-based perspective, bullpen roles are still such an integral part of the game precisely because they are preferred by the players themselves.

Now, returning to the issue of Chapman occasionally pitching the eighth in close games to avoid rust, it will be fascinating to see how this all plays out. To understand why the team would be making this change, here are some stats to consider:

It’s apparent that he’s a guy who, historically, has excelled with less rest. How will he respond, however, when the opportunities for more appearances come in non-traditional spots that he may not used to be pitching in? Chapman’s career numbers are quite good in non-save situations so I imagine that he’ll be fine, but it’s fairly unique to see a team even broach the possibility of throwing their big money closer in non-save, non-ninth inning spots throughout the regular season.

We often hear about how much of an effect the ever-spooky ‘analytics’ have had on the game in recent years, but to me, at least, it’s fascinating to see how little of an effect they’ve had on traditional bullpen roles and usage. Whether or not he’s occasionally pitching in the eighth inning, Chapman will still be in there to shut the door in nearly every save situation, Clay Holmes will likely slide into more of a setup man role thanks to the absence of Britton, and Jonathan Loáisiga will be the fireman. The more things change, the more they stay the same, I suppose.