Bullpens Helped Decide the World Series

Just as we all predicted in the run-up to the World Series, relief pitching had a hand in determining the outcome. Though starting pitchers contributed admirably to relief efforts (two-thirds of a scoreless inning for the Astros, six scoreless for the Nationals), real relievers had to take the stage occasionally. And after you strip out those innings by starters, a trend emerges.

The Nationals bullpen was bad. That’s no shock — they were bad all year, and they were facing one of the best hitting teams of all time. Strip out Patrick Corbin’s four innings of lights-out relief, and Joe Ross’ two scoreless innings earlier in the series, and the balance of the bullpen recorded a 5.51 ERA, with as many walks as strikeouts. They allowed four home runs in only 16 and a third innings. Fernando Rodney’s line looks like a work of comedy — two innings, no strikeouts, six walks. It was more or less what every Nationals fan feared going into the series.

But if the Nationals bullpen was bad, the Astros bullpen was a full-fledged disaster. When the Astros needed relief innings, one of the best bullpens of the regular season simply wasn’t up to the task. Over 21 and a third innings, they recorded a 5.91 ERA and a 5.37 FIP. They kept their heads above water on the non-contact front, with 24 strikeouts and only 13 walks, but also gave up four home runs. Eight Houston relievers appeared in the series, and seven of them allowed runs.

But even that grim statistical record undersells things. Houston’s bullpen also allowed three unearned runs, while Washington’s pen allowed none. That leaves the Astros with a 7.17 RA/9 out of the bullpen, a number that almost doesn’t look like a baseball statistic. The Astros bullpen put together a 3.75 ERA in the regular season, and a 4.24 FIP. As recently as the ALCS, they’d looked like a cohesive unit, with a 4.12 ERA and 4.80 FIP — not great, but enough to get by against the fearsome Yankee offense. In the World Series, it all crumbled.

Add in the relief outings by starters on both sides, and it looks even worse for the Astros. With Corbin and Ross in the fold, the Nats bullpen had a 4.03 ERA in the series. Urquidy barely moves the needle for Houston — they still check in at 5.73, and 6.95 RA/9. The Nationals bullpen performed above its defense-independent metrics, to be sure — even counting Corbin and Ross, they had a 5.63 FIP — but in raw run prevention, they lapped the Astros.

Of course, these statistics are fun, but they don’t mean very much in small samples. Not only is 20-some innings not enough to show a bullpen’s true talent, but it doesn’t even consider which relievers are pitching and when. Rodney, for example, pitched three times in the series. When he entered, the Nationals were 9.9%, 14.5%, and 98% to win those games — he never pitched in a close situation. Likewise, Chris Devenski appeared three times, with win probabilities of 0.0%, 100%, and 3.4% for the Astros. Their bad performances (and they were both awful) can’t tell us much about how the bullpens performed when the game was on the line.

To get to that, we’ll need to use Win Probability Added. It’s a toy stat, not fit for any kind of forward-looking analysis, but if we want to see what bullpens did to affect the outcome of the World Series, it’s a good starting point. The Astros bullpen had no heroes — Joe Smith did the best, with 0.10 WPA:

On the other side, the Nationals bullpen worked exactly as planned. Its two best pitchers, Corbin and Doolittle, got the highest-leverage chances and excelled:

In a nutshell, this is how the series went. The Nationals bullpen, warts and all, added half a win to the team’s chances. In a 4-3 series, that’s momentous. The Astros bullpen, on the other hand, subtracted around half a win. In a 4-3 series… well, you saw what I said above.

There aren’t really any clear remedies to this for the Astros. They used good pitchers, and their good pitchers performed poorly. Will Harris has a 2.84 ERA and 3.03 FIP over 400 career innings. He had a 1.50 ERA this year. He just got beaten, twice, in big spots — once by an Anthony Rendon home run that would only be a home run in Houston, and once by a Howie Kendrick home run that literally hit the foul pole. Baseball is hard to forecast sometimes.

On the other hand, the Astros made a few curious decisions in bullpen management. Pressly missed most of September recovering from arthroscopic surgery on his knee. In the playoffs, his fastball velocity was down 1.3 mph from the regular season, a decline mirrored in the rest of his pitches. It’s hard to tease out how much of his rough performance this postseason was injury-related — he allowed a .300 xwOBA in the playoffs, which means his expected results were better than average despite his 11.12 ERA (he even posted a sterling 2.68 FIP). But even if it wasn’t injury-related, Pressly looked a bit rust, and Hinch seemingly bumped him down the leverage ladder as the playoffs wore on.

More pressing, however, was the team’s use of Roberto Osuna. Setting aside any personal feelings on the matter (and boy, do I have personal feelings on the matter), Osuna pitched in huge spots all regular season for the Astros, as befits a team’s best reliever. In his two World Series appearances, he entered with Leverage Indices of 1.01 and 1.31. Set aside getting your best reliever into only two games — getting him into only two games with neither time being particularly important spots seems poorly conceived.

Leverage is a fickle concept in the playoffs, of course. Every inning of yesterday’s game was equally important when the inning started. And yet, there were big spots in this series, and there were big spots in yesterday’s game. Kendrick’s go-ahead home run came off of Harris when he was pitching for the second straight day, and it came with a man on base and one out. Craig Edwards wrote more extensively about the Astros’ seventh inning decision-making, but Osuna would be the natural choice to come in there — and indeed, he came in to relieve Harris after Harris gave up that home run and then a single to Asdrúbal Cabrera.

Would this have impacted the outcome of the series? Who knows. Harris is no slouch — he had a lower ERA and FIP than Osuna this regular season, and allowed fewer home runs per nine innings. But in this universe, he gave up a dinger, and that was, more or less, the series.

The Nationals, by contrast, had three clear best pitchers, and they used them accordingly. Two times this series, the outcome of the game hung in the balance when Dave Martinez was using his bullpen. In Game 1, he used Hudson and Doolittle to hold the lead after Tanner Rainey made the game exciting (and after Corbin had thrown a scoreless inning already). In Game 7, he used Corbin and Hudson, with Doolittle available in reserve, to get all four innings worth of outs.

This series doesn’t mean anything overarching about the Astros bullpen, or about A.J. Hinch’s managerial acumen. They used some pretty good pitchers, and those pretty good pitchers got beaten. When you’re facing off against Anthony Rendon, Juan Soto, and Howie Kendrick, you’re not going to win every time. That’s just life.

It’s refreshing to me, though, that one of the biggest stories coming into the series — the bullpen disparity between two teams otherwise loaded with stars — really did end up deciding the series. It just didn’t decide it the way you would expect.