It’s Time for the Astros to Trust Their Relievers
The World Series kicks off tonight with a battle of elite left-handed starters, Clayton Kershaw and Dallas Keuchel. The game will also feature the two best offenses in MLB this year. And yet, for all the talent on the field to begin the game, the series may very well hinge on whether A.J. Hinch is willing to once again trust his bullpen.
During the regular season, the Astros’ relief corps was better than their reputation suggests. As a group, they posted the second-highest strikeout rate of any bullpen, and while their 101 ERA- was a bit below average, their 84 xFIP- was second only to the Indians.
And while the results didn’t match the peripherals, most of the guys who drove those differences —James Hoyt (4.38 ERA, 3.11 xFIP), Michael Feliz (5.63 ERA, 3.58 xFIP), Francis Martes (6.63 ERA, 4.59 xFIP), and Tony Sipp (5.79 ERA, 4.34 xFIP) — aren’t on the World Series roster. The seven relievers the team is carrying were mostly excellent, by both results and peripheral stats.
Using the active roster filter on our leaderboards, which includes just players currently eligible to play in tonight’s game, the regular season data doesn’t show a significant difference in bullpens in this match-up.
| Team | BB% | K% | K-BB% | HR/9 | BABIP | LOB% | ERA- | FIP- | xFIP- |
|---|---|---|---|---|---|---|---|---|---|
| Astros | 8.3 % | 29.0 % | 20.7 % | 1.1 | 0.269 | 81.4 % | 70 | 79 | 80 |
| Dodgers | 6.4 % | 27.5 % | 21.1 % | 1.2 | 0.275 | 80.8 % | 76 | 83 | 81 |
Of course, the groups have not performed similarly in the postseason, with the Dodgers bullpen absolutely dominating so far, while the Astros relievers have mostly struggled when allowed on the mound. And because of those struggles, A.J. Hinch has done whatever he can to avoid using them.
Between the ALDS and ALCS, the Astros have played 11 games so far this postseason. Here are the number of batters the regular relievers — not counting starters who have pitched out of the bullpen — have faced in those 11 games.
| Pitcher | Batters Faced | Outs | Pitches |
|---|---|---|---|
| Ken Giles | 29 | 18 | 123 |
| Chris Devenski | 16 | 9 | 62 |
| Collin McHugh | 13 | 12 | 53 |
| Will Harris | 11 | 6 | 54 |
| Joe Musgrove | 11 | 8 | 44 |
| Luke Gregerson | 10 | 8 | 46 |
| Brad Peacock | 10 | 7 | 40 |
| Francisco Liriano | 8 | 5 | 30 |
There are no sugar coating most of those numbers; the Astros regular relievers have been ineffective in the postseason so far. Take away McHugh’s good outing in his one appearance, a low-leverage outing at that, and you’re looking at the trio of Giles, Devenski, and Harris only retiring 33 batters of 54 faced. That’s a .389 OBP allowed, and it’s not all bad luck; they’ve struggled to throw strikes and have gotten hit hard when their opponents made contact.
But while the results have been undeniably poor, the obvious reality is that we’re talking about the smallest of samples here. And the Astros have already benefited from giving early-playoff-strugglers a second chance this postseason.
In his one division series outing in relief, Lance McCullers allowed a .385 OBP to the 13 batters he faced, and the Astros are probably pretty happy they didn’t stop trusting him after that outing. In his first two postseason starts, Charlie Morton faced 39 batters and allowed a .382/.462/.618 batting line, but the Astros believed in his stuff and track record, and in his Game 7 start, he retired 15 of the 18 batters he faced over five shutout innings.
Just as Morton and McCullers were allowed to redeem themselves, Hinch should once again give his best relievers a chance to show that they can still get outs too. Because the Astros path to the World Series title looks pretty difficult if they have to rely solely on their starters for high-leverage innings.
For one, the Dodgers look like a tough matchup for Dallas Keuchel. As Jeff noted yesterday, the Dodgers hitters don’t chase balls out of the zone, and no starter throws fewer strikes than Dallas Keuchel. The Dodgers offense is geared to taking advantage of guys who don’t throw strikes, and Keuchel against a patient line-up of guys who won’t chase his sinker could be a problem for the Astros.
The Dodgers also don’t hit many grounders, which is Keuchel’s bread and butter, but have swings designed to elevate low pitches. So, while Keuchel could certainly adjust and pitch differently than he has all year, the Astros probably can’t go into this series expect him to give them two deep outings.
If Keuchel is more likely to be a five inning pitcher in his starts, then Verlander is probably the only hope the Astros have of getting a lot of innings from their rotation. Neither Morton nor McCullers are going to eat innings in their starts, and the Astros will probably want to make sure McHugh and Peacock are mostly fresh for games three and four, when they might want to try the tandem-starter thing again. But while Peacock could probably pitch tonight and still be ready to go for an extended outing on Saturday, the reality is that the Astros will probably need to get some good innings from their traditional relievers today.
Giles, Devenski, and Harris have been among the game’s best relievers for years now. Their stuff is good, they mostly throw strikes, and they shouldn’t be cast aside just because they had a few poor outings earlier in the postseason. If Hinch continues to avoid all of his regular relievers, he’s going to get to the end of the series with worn down, burned out starters.
While the idea of Justin Verlander single-handedly carrying the Astros to a World Series title is a nice story, it’s also an unrealistic expectation. To beat the best team the National League has to offer, the Astros will need contributions from more than just a few trusted pitchers. The Astros are carrying 11 pitchers for a reason, and Hinch needs to be willing to deploy more than five or six of them over the next seven games.
Dave is the Managing Editor of FanGraphs.


I think a big part of the issue with faith in bullpens is that we just don’t know the health and fatigue status the way the manager does.
With an offensive team like Houston we expected them to bounce back and still be good after their anemic first 5 ALCS games. We would expect the same from the Houston bullpen IF they are in the same condition they were all year. And we have no way of knowing if AJ Hinch is being biased to recent results or knows his bullpen is fatigued, hurt or compromised and aren’t the same guys. We can see with Alex Wood that he’s just not the guy right now his season line suggests and we can identify it in his lost velocity. When the injury or fatigue manifests as lack of command or location it’s harder to diagnose with our tools.
The Astros bullpen may not be what it was in the regular season for real and we have no real way to know what guys are playing through.
Legitimately curious why Musgrove isn’t getting any love? He’s been very good since switching to the BP, better than Devo in that same span.
Or they could just piggyback their starters each game the whole series. Keuchel/Peacock, Verlander/McHugh, Morton/McCullers. Tandem starting is the new bullpenning, amirite?
(I started writing this as a joke, but now I’ve sort of convinced myself)
Don’t think either Keuchel or especially freakish Verlander would look kindly at that idea.
not sure about the other two, but for Devenski, it wasn’t just the playoffs where he struggled. His second half was pretty bad (FIP/xFIP around 5) but he got lucky with a low ERA so no one really noticed.
I agree.
One of the bigger challenges for managers and pitching coaches has to be determining when pitchers, including relievers, have truly had a change in expected underlying performance versus a rough patch that’s not predictive. We can look at velocity and movement. Quickly scanning Brooks numbers, Devenski’s fastball velocity was constant, or up maybe up half a tick in September and October, though the horizontal movement on his fastball and slider changed a bit late in the season.
So I don’t at a glance know the answer to why his performance changed so much. I don’t want to make a straight appeal to authority, but the team has access to far better health information on Devenski than we do and presumably better tracking on how he’s done at pitching to intended location. It’s easy to put forward a hypothesis that Devenski is gassed after throwing 52.2 innings before the All-Star Break, but if so it’s showing up in something other than velocity.
Out of curiosity Dave, how are you establishing the line between a “regular relievers” and “starters coming out of the bullpen”? I get that Musgrove and Liriano should be thought of as relievers now, but I mentally associate Peacock and especially McHugh with the latter group.
The Astros are carrying 12 pitchers: Keuchel, Verlander, Morton, and (likely) McCullers to start, with Giles, Gregerson, Harris, Devenski, McHugh, Peacock, Liriano, and Musgrove in relief.
Note — McHugh shouldn’t be in that table of “regular relievers.” His ALCS outing was his first in relief since 2013.
I’d add Peacock as well. Peacock was only a reliever for 13 appearances this year. While he can relieve, he isn’t regularly one.
Devenski’s last three months before October were the worst of his 2-year MLB career
YEAR MONTH FIP
2016 april 2.77
2016 may 3.40
2016 june 1.72
2016 july 3.39
2016 august 1.91
2016 sept 0.55
2017 april 2.02
2017 may 2.53
2017 june 2.41
2017 july 6.33
2017 august 5.06
2017 sept 4.28
I guess I question the premise, that Hinch has stopped trusting his core relievers. Maybe Harris, but not Devenski and Giles.
Devenski and Giles have totaled 10 appearances and 9 innings in 11 games. As a pair, both their usage rate and their IP share are up from the regular season.
Both have had some serious failures:
After good work in the first 2 games against Boston, Devenski flopped in game 3, pitching for the 3rd time in 4 days. Hinch let him sit out the finale, then used him in the LCS opener, starting the 8th with a 2-0 lead; but Devenski committed the crime of walking Brett Gardner to bring up Judge as the tying run, so Hinch went to Giles. Hinch used him again in game 4, up 4-1 in the 7th, but a triple, lineout and walk again brought up the tying run, so out he went, and he hasn’t pitched since.
Giles really hasn’t had a clean game yet, and he only pitched 1 mopup inning after torching game 4 against New York. But overall, he’s still had the bulk of the high-leverage late work.
I’m wondering — where were the spots that it seemed Hinch might have used these two, based on regular-season performance and postseason planning, but didn’t?
Let’s just take the LCS:
— Game 1, both pitched in a tight game.
— Game 2, Verlander went all the way. To my eyes, that was perfectly reasonable.
— Game 3, Morton and Harris took the air out early, leaving an 8-0 hole thru the 4th. Maybe Morton got a slow hook, although it was written on FG that his results belied his performance. Maybe Devenski should have been first out of the pen, but (a) it was 4-0 by then, and (b) we’re already talking about Harris being underused.
— Game 4, both pitched. Badly.
— Game 5, Keuchel (who was brilliant in the opener) maybe got a slow hook. He gave a 2-out walk to Judge in the 5th and then 2 hits that doubled the deficit to 4-0. Then again, he fanned 8 of the first 16 batters coming into that inning; and the 2 who got hits in the 5th, Sanchez and Gregorius, had been putty in his hands before that.
— Game 6, Verlander went 7 scoreless; Peacock took the 8th, starting up 3-0; and Giles the no-leverage 9th. Maybe Devenski should have taken the 8th instead, but y’know, Peacock’s real good, too, and maybe Hinch wanted to keep Devenski ready for the do-or-die game.
— Game 7, Morton was brilliant through 5, handing a 4-0 lead to McCullers. News flash, he’s real good, too.
I just don’t see the issue on Devenski and Giles. To me, they’ve pitched as much as they should have; maybe a tad less than we expected, but that’s due to their poor performance and the excellence of some other pitchers. The fact is, Hinch has a lot of resources right now. When you can get 4 scoreless frames from a McHugh or McCullers, isn’t that better in some ways than using multiple guys in shorter stints?
I admit that from watching the playoffs and the games leading up to the playofs and reading articles like these I have been left with the impression that Hinch doesn’t have a ton of faith in his bullpen right now. I have to agree though. Other than perhaps the Verlander CG I can’t really think of an instance in which he made a decision that he wouldn’t have even if the relievers had his full confidence.
In fact, most complaints from Astros fans (complaints I 100% disagree with) about Hinch this postseason center around the quick hook he has given starters despite their not yet being in trouble.
75% of the pitchers on Houston’s WS roster pitched more innings as starters than relievers. I think this is by design, and, the more I think about it, I’m not at all convinced that giving more innings to Devenski, Harris, and Giles will lead to worn down, burned out starters. I think it might just be them who ends up worn down and burned out.
Game 7, top of the 5th inning was a time that a reliever (presumably Devenski or Harris) almost certainly would have come in if Hinch still had a lot of trust in those guys. Morton still pitching, Astros up 1-0. The first 3 hitters of the inning went double, strikeout, walk (with wild pitch) to put runners on 1st and 3rd with 1 out. Morton and the Astros got out of the inning with the very good Bregman to McCann play to get Bird at home followed by another groundout.
I agree with you on McCullers, because he pitched all four of his innings with a 4-0 lead and never allowed more than a single runner in an inning, and neither of the two runners advanced past first base. I think that a manager would bring in the closer to start the 9th if he’s got Kenley Jansen – to pick an obvious example – but it’s not that odd to let McCullers keep going absent such an obviously great option.