The 2017 Yankees relievers were elite, by any metric you care to use. They had the third-lowest ERA (3.34); the lowest average against (.204); the highest strikeout rate (29.1 percent); the most Wins Above Replacement (9.2). They were good. They were so, so good.
Now realize they could possibly be better in 2018. Then, think about the fact that if they were, it might put them in the conversation for "the best bullpen of all time," an extremely unofficial title that's nonetheless fun to think about.
How could those things happen? And what does "best bullpen" even mean? Let's dig in.
How to make a great bullpen even better
Part of the difficulty in evaluating a bullpen is that the members of that bullpen cycle through on a seemingly endless basis, with promotions, demotions and trips to the disabled list creating a different group nearly daily. There's not a constant group all year long; even the Yankees had 18 pitchers enter in relief at some point.
A good way to illustrate how much turnover can happen is to remind you who was actually in New York's bullpen on Opening Day 2017.
LHP -- Albertin Chapman, Chasen Shreve, Tommy Layne
RHP -- Dellin Betances, Adam Warren, Tyler Clippard, Bryan Mitchell, Jonathan Holder
Who don't you see there? You don't see Chad Green, who didn't join the bullpen full-time until May and ended up becoming one of baseball's breakout stars. You don't see Player Page for David Robertson and Tommy Kahnle, acquired from the White Sox along with Todd Frazier in July. Those three pitchers combined for 130 2/3 dominant innings with the Yanks, pitching to a 1.79 ERA along with 190 strikeouts.
Looking at Expected wOBA, the Statcast™ metric that combines quality of contact (in terms of exit velocity and launch angle) along with amount of contact (in terms of strikeouts and walks), Green (8th), Kahnle (9th) and Robertson (11th) were three of the top dozen relievers in baseball in 2017, among those who faced 200 hitters. None of these guys were in the Opening Day bullpen.

What that means is that even with only partial seasons from three of the game's most dominant relievers, the Yankees' bullpen still reached those lofty statistical heights. For example, in April, New York's relievers who pitched the most innings were Warren and Mitchell. In May, it was Holder, Warren and Clippard. By September, the four relievers to pitch 10 or more innings were Robertson, Chapman, Green and Kahnle. The decent gave way to the elite.
The new trio took innings that had otherwise gone to Mitchell (5.79 ERA, now with San Diego) or Clippard (4.95 ERA, now a free agent) or Layne (7.62 ERA, now a free agent), and that's the point. The Yanks enter 2018 with far more talent atop the depth chart. If they started the season today, the bullpen could look something like this:
LHP -- Chapman, Shreve
RHP -- Betances, Robertson, Kahnle, Green, Warren, Giovanny Gallegos
That group, combined, put up a .258 Expected wOBA, a 2.63 ERA, and a 34.2 percent strikeout rate, in over 430 innings. Put another way, that's the same performance as Clayton Kershaw (.253 Expected wOBA), Stephen Strasburg (2.52 ERA) or Corey Kluber (34.1 percent strikeout rate), just in far more innings.

Five of those eight Yankees arms were in the top 16 in strikeout rate, while Shreve "only" struck out 58 in 45 1/3 innings, limiting lefties to a mere .161/.235/.262 line. Warren, serving as a multi-inning man, pitched to a 2.35 ERA and a 3.02 FIP. He may be this team's seventh-best reliever.
The little-known Gallegos, for what it's worth, just led the entire Triple-A International League with 40.8 percent strikeout rate. Don't like him? Ben Heller, who allowed one earned run in 11 innings for the Yanks in 2017, was third, at 36.8 percent. Or what about Holder, who struck out 40 against just eight walks in 39 1/3 Major League innings?
That's not to suggest that the Yankees' arms are without risk; every team has risk. Chapman missed a few weeks with an arm injury, then slumped in August before returning to dominate in September (17 whiffs across 12 scoreless innings). Betances' late-season command issues essentially sidelined him in the postseason, and make him something of a question into 2018 -- as much as anyone who struck out 100 in 59 2/3 innings can be "a question." There's no guarantees here. Just tons of elite strikeout talent.

How do you rank a historic bullpen?
We might have undersold, at the top, how good the 2017 bullpen already was. The well-known stat FIP (Fielding Independent Pitching) takes the three things a pitcher has the most control over (strikeouts, walks and home runs) and puts it on an ERA scale. So the Yanks' relievers, for example, tied for No. 1 in FIP in 2017 with a 3.12 mark.
By itself, 3.12 doesn't rank highly in history, but you have to compare how that ranks against the league average for that season. In the same way that hitting 20 homers in 1968 was far more impressive than having done so in 2017, we have to add context for the way the game was played at the time. If we do that, to see how far above league average for that season their score was, we can see that what the Yankees just did stands out among history.
Best relief FIP compared to MLB average, 1920-2017
32 points above average -- 2003 Dodgers
32 points above average -- 1964 Reds
27 points above average -- 2017 Indians
26 points above average -- 2017 Yankees
(Yes, this could be an article about the 2017 Indians, too. But it's not -- and while the Yanks have full seasons of pitchers they added last year, Cleveland lost reliable Bryan Shaw to Colorado.)
Our group of eight in the assumed Opening Day bullpen, for what it's worth, would have been 40 points better than average in 2017 if you combine their 2017 MLB stats. If we add in Heller and Holder to get to a group that's 10 deep, it's still 38 points better than average. Even this ranking is missing an important factor, which is innings. Those 1964 Reds threw just 322 1/3 relief innings, while the 2017 Yankees threw 538 1/3. Being tied is hardly as impressive when you're throwing less than 60 percent of the number of innings. (And if you've forgotten about the 2003 Dodgers, that was the team that featured National League Cy Young Award winner Eric Gagne closing games with a 337 ERA+.)
Ultimately, with the changing way that relievers are used, there's not an ideal way to compare present-day bullpens to those of years past. The sheer number and usage of relievers barely resembles the game of decades ago. That's to say there's no "right answer" as to what the best bullpen of all time could even be, which is fine: It's more fun to argue than to know.
What we do know for sure, however, is that the Yanks' bullpen was great last year. We know that they look even better in 2018, projected to be the best group in baseball. And if everything goes right, they might just be the best group we've ever seen. As far as we know, anyway.