The Nationals have one of the weakest bullpens in baseball -- their 5.93 ERA is second-worst, ahead of only Baltimore -- unless they don't, because since June 1, it's dropped by a full run to 4.86 ERA, 18th-best, and hey, what's a "bullpen," anyway?

You can see why this gets complicated. It's true that the Nationals' bullpen started off bad, but outside of Sean Doolittle, this has been a revolving door of players. That 5.93 tells you something about how poor the bullpen has been, but it also includes the fact that a group including Trevor Rosenthal (now with the Tigers), Dan Jennings (Yankees' Minors), Javy Guerra (DFA'd), Justin Miller (DFA'd), Kyle Barraclough (Double-A), Austen Williams (injured) and Jonny Venters (injured) combined for a 7.64 ERA in 98 games.

It doesn't include the fact that most of those players will never appear in a game for Washington again. It doesn't account for the fact that July 31 trade acquisitions Roenis Elias, Hunter Strickland and Daniel Hudson will be expected to carry much more of the load over the final two months. As you'll see, this doesn't change Washington's situation from a "bad" bullpen into a "great" bullpen, but we're also at the point in the season where what long-departed arms did months go simply doesn't matter.

So: How do we account for that? There's no perfect way, obviously. There's no crystal ball, and as we know, a single bad outing or two can inflate a reliever's ERA for months. But if we're looking for a combination of the players you have now and reasonable expectations for their performance, there's only one way to go: projections.

For that, we can look to FanGraphs, because it combines the latest projections with up-to-date playing-time expectations following the craziness of the Trade Deadline. So for Washington, for example, the four relievers expected to pitch the most are Doolittle and the three new relievers, along with Wander Suero and Fernando Rodney. There's no Rosenthal, no Jennings.

So, for the 15 teams that still have playoff odds of at least 10 percent -- that's a list that includes the Mets but not the D-backs, Giants or Reds -- which bullpens look to be in the best shape over the next two months?

You could do this by projected WAR, or FIP, or something even fancier. Let's keep it simple, with ERA. The answers may surprise you.

1. Rays (3.57 projected ERA)

The Rays have had one of baseball's better bullpens for much of the year, though it's a very different group than the one that seemed like it revolved around Jose Alvarado and Diego Castillo earlier in the season. This is a reflection in part of how strong Emilio Pagan has been both by traditional (2.00 ERA) and Statcast measures, but it also includes new acquisition Nick Anderson, who came from Miami at the Deadline and has a Top 10 strikeout rate among relievers. Also worth noting: Ryne Stanek, their primary opener, had a 2.09 ERA as a starter but a 7.82 mark in relief. He won't be doing that anymore, now that he's off to the Marlins as part of the Anderson deal.

2. Astros (3.68)

While the focus in Houston is understandably on the rotation and Zack Greinke, this is also a very good bullpen, too. Assuming relief ace Ryan Pressly's sore knee is as minor as it sounds, Houston's Big Three of Pressly, Roberto Osuna and Will Harris (combined ERA: 2.15) is as strong as any in the game. You know a lot of the names behind them -- Chris Devenski, Joe Smith, Hector Rondon and Collin McHugh -- and they've added Joe Biagini from Toronto as well. If they can fix what ails Aaron Sanchez and he ends up in the October bullpen, too, he could suddenly be a lot more interesting than you'd have expected.

3. Yankees (3.69)

You may have heard that the Yankees didn't make any additions to a tattered starting rotation at the Deadline. That's a problem, but they do at least have what's been arguably baseball's best bullpen to lean back on. (They lead in relief Wins Above Replacement, despite being sixth in ERA.) The continued excellence of such names as Aroldis Chapman, Adam Ottavino and friends is the obvious strength here, but the projections also think a lot more of Chad Green (projected 3.18 ERA) than his 5.06 ERA would have you believe. That makes sense; he's posted a 2.57 mark since returning from the Minors in May. And don't forget: Dellin Betances may yet be added to this group.

4. A's (3.73)

Oakland's run to the Wild Card Game last year came on the strength of some elite relief performances, and this year's group is once again good, though in very different ways. Last year, Blake Treinen and Lou Trivino were the Big Two and combined for a 1.82 ERA; this year that's all the way up to 4.47. So how are the A's still performing this well? Because Liam Hendriks has been excellent, and is expected to continue to be. That's because Yusmeiro Petit has been fantastic, even if the projections don't fully buy into it, and because new addition Jake Diekman is as good at missing bats as Hendriks is. This is a good group; it could be a lot better than that if they can get Treinen or Trivino pitching the way they did in 2018.

5. Dodgers (3.74)

The Dodgers didn't go out and get the big-ticket reliever to pair with Kenley Jansen that fans were hoping for, and now they'll have to prove that it wasn't a mistake. They've been fine so far -- eighth-best bullpen ERA -- but really, the rest of the regular season is irrelevant here. When they get to October, they might have starters Kenta Maeda, Rich Hill and Ross Stripling here, in addition to possibly rookie Dustin May, and don't forget that after a horrendous start, Joe Kelly has looked much better after making some changes to his pitch repertoire.

6. Mets (3.76)

The Mets still have the third-worst bullpen ERA in baseball, 5.23, a reflection of some truly dreadful early-season breakdowns. So that's bad, but since July 1, they've cut that all the way down to 3.72, the ninth-best mark in baseball, as they've cycled out such names as Wilmer Font and Hector Santiago. That's more or less what the projections expect, though you'll probably take the under on Edwin Díaz pitching well or Jeurys Familia turning it around. If the Mets are going to live up to this, they're going to need Seth Lugo (12 scoreless outings in July) to keep pitching like one of the best relievers in the game.

7. Brewers (3.88)

Josh Hader is probably the most dominant reliever in the game right now, and the projections reflect that. There's a big gap between him and everyone else here, however, because Jeremy Jeffress, Junior Guerra, Matt Albers, Alex Claudio and new addition Drew Pomeranz straddle either side of "average-ish." That's not a bad thing, but it also sort of reflects their current No. 16 ranking in bullpen ERA.

8. Cardinals (3.92)

Last year's "Luke Voit for two relievers" trade looked like it was going to be one of the larger missteps in St. Louis history for a while, but that undersells just how fantastic Giovanny Gallegos has been for the Cardinals; in 43 games, he's got a 72/10 K/BB with a 2.13 ERA. He's been a life saver for a bullpen that lost Jordan Hicks to Tommy John surgery, especially since the "Carlos Martínez to relief" experiment has been mixed. While John Brebbia and John Gant have been solid, the best news here for St. Louis fans is that Andrew Miller may be figuring it out. In 17 innings since June 1, he's got a 27/7 K/BB, having allowed only five earned runs.

9. Braves (3.95)

The Braves went in big on relief at the Deadline, adding Shane Greene from Detroit, Mark Melancon from San Francisco and Chris Martin from Texas. Right away, that helps, if only because a group that includes Jesse Biddle, A.J. Minter, Chad Sobotka, Kyle Wright, Dan Winkler and Touki Toussaint -- none currently on the active roster and some no longer in the organization -- combined for a 6.27 ERA, and they won't be eating as many innings. Now, is Greene going to keep pitching to a clearly unsustainable 1.18 ERA? Of course not. But these three are upgrades, and they push Luke Jackson to lower-leverage work. It's not a great bullpen. It's just a better bullpen than they had a week ago.

10. Cubs (3.97)

Chicago has made a few obvious changes over the season, namely importing Craig Kimbrel, Derek Holland and David Phelps in July, while shipping Mike Montgomery and Carl Edwards Jr. elsewhere. A lot of this improved projection -- 3.97 is better than their actual 4.21 -- depends on Kimbrel, however. His 6.17 ERA is a little misleading, because 10 of his 13 outings have been scoreless, but his wildness and declining velocity are more than a little concerning. Unfortunately, injured veteran Brandon Morrow and rookie Adbert Alzolay can't be counted on to contribute, so Kimbrel's performance is hugely important here.

11. Indians (4.02)

We feel like this might be the most controversial ranking here. After all, Cleveland has baseball's best relief ERA (3.33), and its 1.81 relief ERA in July was the third-best month by any team in the last three seasons. So why the low ranking? Partially because these things can turn quickly -- last May's 8.01 relief ERA was the second-worst of any team in the last three seasons -- and partially because the underlying performance was more "good" than "great." (In July, Cleveland's bullpen had the 14th-best strikeout rate and third-lowest walk rate, but had a .232 BABIP that was by far baseball's lowest.)

This is a good group, and Brad Hand is elite both in performance and projections. It might just be that less-accomplished relievers Adam Cimber, Nick Goody and Nick Wittgren aren't quite as good as their ERA would indicate.

12. Red Sox (4.11)

Did the Red Sox do enough at the Deadline? Most Boston fans would say "probably not," and Friday's news that Heath Hembree is headed to the injured list won't help that. The good news, anyway, is that the projections still view Matt Barnes as being near-elite, and after a very rough June, he was outstanding in July, throwing 10 scoreless outings with a .111/.200/.148 line against. If Nathan Eovaldi can successfully convert to the bullpen, and if young Darwinzon Hernandez can continue his impressive debut, and if Ryan Brasier can turn things around in Triple-A, they may have something here. That's a lot of "ifs," though.

13. Twins (4.15)

Minnesota has one truly elite reliever in Taylor Rogers, and, despite his rough Twins debut, Sam Dyson has long been a solid arm, posting a 2.60 ERA in 2018-19 for the Giants. Sergio Romo is a nice, if low-impact, add. Now: Is that enough? Ryne Harper's rookie season comes with a shiny 2.93 ERA, but that's a 1.61 ERA in April/May, and a 4.35 since June 1 (hence his 4.04 projection). Trevor May is projected to pitch about the same as he has been. Tyler Duffey is fine. Is it enough? It might not be enough.

14. Nationals (4.16)

We said the Nationals would be better than their season-long ERA suggested. We didn't say they'd be good. Hudson, Elias and Strickland are solid veteran adds, but none are difference-makers, and Doolittle remains the only clearly above-average member of this group.

15. Phillies (4.24)

Sometimes, past performance does project future production. The Phillies have the eighth-worst bullpen ERA (4.73), and they have the weakest projection among contenders here. The knowledge that David Robertson won't be back and that Adam Morgan is injured doesn't help, and scrap-heap pickups Blake Parker and Mike Morin are unlikely to change the narrative here by themselves. Héctor Neris, to be fair, is one of the better bat-missing relievers in the game, but he doesn't have enough help here.