Joe Girardi's quick thinking, hook pay off

On Tuesday, in the American League wild-card game, New York Yankees manager Joe Girardi did what Buck Showalter didn't do 364 days earlier: He managed an extraordinary game with extraordinary urgency.

But urgency is a spectrum, not an either/or, and just because a manager is more urgent than Showalter doesn't mean he can't be too urgent. Did Girardi make the right move? It worked, but could we ever know what the right move is? Consider what Girardi had to consider.

1. Severino looked bad. He threw 29 pitches, and not one of them drew a swinging strike -- this from a pitcher who typically gets a swinging strike every eight pitches. He allowed two homers to the first four batters, after not allowing two homers in a start in all of September. Two-thirds of Severino's sliders are typically in the lower third of the strike zone or lower; of the seven he threw Tuesday, four were in the middle third or above, including the one Eddie Rosario hit out to give the Minnesota Twins a 3-0 lead. Severino did not look in control.

So ... what does that mean, exactly? On July 8 of this year, Severino allowed three runs in the first inning to the Brewers, including a home run to Domingo Santana that made the score 3-0. But Severino would end up throwing seven innings, allowing no more runs, striking out 10 batters, and the Yankees would win 5-3. Sometimes pitchers need a minute to settle in. And sometimes six batters aren't predictive.

In fact, we rarely (if ever) consider six plate appearances predictive in any other context. You can see how misleading six plate appearances can be: Severino was throwing as hard as he ever does, and the movement on all three of his pitches was right in line with his seasonal averages. He was missing location, but guys who throw 99.5 mph up in the zone with movement get away with missed location in every at-bat, every game. The leadoff home run to Brian Dozier on Tuesday was 99 mph at the top of the zone -- Severino's pitch, a swing-and-miss pitch he has been throwing successfully all year -- and a great hitter simply beat it, by a few feet at most. The home run to Rosario was against a bad pitch, to be sure, but it traveled all of 336 feet, one of the 20 shortest home runs hit in all of baseball this year.

The margins between getting out of an inning and getting knocked out of an inning are often microscopic. That's what Girardi had to evaluate, in real time, with access to no granular data, while considering whether to make one of the most drastic moves of his career.

2. The relievers Girardi had available were likely to pitch better than Severino -- better even than the good version of Severino. In fact, FanGraphs' Travis Sawchik argued a few days ago that the Yankees might be better off going to the bullpen from the first pitch, continuing an annual tradition of some smart writer arguing (hopelessly) that some smart team ought to take advantage of the one-game format and ride a dominant bullpen to a one-game victory.

For the Yankees, with one of the deepest, most dominant bullpens in history, the reliever options would look especially attractive. As Sawchik wrote:

If ever there were a team ideally suited to bullpen a Wild Card play-in game, it's this Yankees team. Yankees relievers have combined to post the highest collective strikeout rate of all time among bullpens. The club has five individual relievers who have struck out better than 30 percent of batters faced: Chad Green, Dellin Betances, David Robertson, Aroldis Chapman and Tommy Kahnle. All five also have K-BB% marks of 21 points or better and FIPs of 3.20 or lower. Chasen Shreve represents another high-strikeout arm and a second left-handed option.

Toss out Shreve from the group. The five headliners allowed a combined 2.55 runs per nine innings this season. Severino allowed 3.40, which is more. Of course, a manager can't buy into this simple math on a daily basis, because starters eating innings (at slightly lower quality) are necessary to get through a long season. But the Yankees had Monday and Wednesday off. They didn't need to run this game like it was one game out of 162. It was just one out of one.

But it would be disingenuous to say that this makes the decision easy for Girardi. If he'd wanted a bullpen game, he'd have planned a bullpen game. He didn't. As he put it, "There are some things you might do a little different once you get into it, but if you start doing crazy things -- things guys aren't used to -- then I'm just not comfortable doing it. You want to keep it as normal as possible."

Going to the bullpen immediately increased the chances of multiple mid-inning pitching changes, or relievers not knowing how many outs they'd be asked to get, or having to warm up without knowing exactly when they'd be called in -- all things relievers say make them less effective, at least when they're not accustomed to being used that way. (As the Yankees' bullpen showed in a dominant group outing Tuesday, one might wonder whether those relievers are worrying more than they need to.)

If those five couldn't make it through 8⅔ innings, then Girardi would need to go to somebody -- Shreve, or some starting pitcher -- who either isn't as good as the dominant five, or who would be asked to do something awkward in a huge spot. And Girardi has been using Betances mostly in low-leverage situations this month, so while you and I might see him as one of the best, most dominant relievers in the game, Girardi clearly has some doubts.

Even in games like this one, there is still some scarcity involved. Managers do have to get to the finish line. If they knew that their five best relievers would get 27 outs, the decision would be easy: Start with the best one, then go to the second best, then the third. But some nights they can get there, and some nights they can't. Getting at least a few innings out of the starter is mostly about reducing the risk of "they can't." When Girardi walked to the mound, he absorbed all that risk.

3. By pulling Severino after just 29 pitches, he'd quite possibly be able to bring his ace back to start Game 2 of the ALDS, if the Yankees managed to get there. That might not end up mattering, but there are scenarios where it could: If Sonny Gray is brought back on short rest for Game 4, for instance, Severino could pitch Game 5 on full rest. If they don't go that route, then Severino would be available out of the bullpen if necessary in Game 5, or perhaps even Game 4. That's not only a nice, tangible benefit, but it also gives Girardi a way of spinning the early hook into something face-saving for Severino: The decision was as much about getting the most out of Severino this postseason as it was about Severino's failure.

Of course, there's a cost. Robertson threw 52 pitches on Tuesday. There's a pretty good chance he'll be unavailable in Game 1 on Thursday. Green might be, too, after throwing 41 pitches. And regardless of how Girardi explains the move, Severino will spend the next three days carrying a career postseason ERA of 81.00, while New York fans and tabs undoubtedly talk about whether he can handle the pressure of the postseason.

4. By making a dramatic move early, Girardi changed the momentum of a game that was threatening to get away from him. The Yankees were down three runs. But they average more than 5½ per game at home. They scored four or more runs in nearly 70 percent of their home games this season, and they scored six or more in almost half. Green gave the Yankees' fans something to cheer for; the two strikeouts he immediately got gave them -- and the players -- a sense that although they'd lost the top of the first inning, they won the final five minutes of it.

On the other hand, one could build the opposite narrative: a manager who was panicking, a team whose ace had just been felled, a team with no margin left for the rest of the game. That's not the narrative we'll tell, because Green did his job. But even Chad Green, even Chad Green, normally doesn't get out of that situation without allowing another run to score.

Girardi had to weigh all of this, every fact in support of an early hook easily interpreted as counterfeit. The key detail might well have been, ironically, paradoxically, that Max Kepler "only" doubled against Severino. If he'd homered, the Yankees would have trailed by five, but the immediate threat also would have been quieted. Severino might have been left out to steady himself, to eat at least a couple innings -- as he'd been left to steady himself after the Domingo Santana homer on July 8. He might even have done it.

But when Kepler doubled, setting up second and third with one out, setting up a situation where a single might end it and two strikeouts could save it, the urgency was undeniable. It was forced upon Girardi. That's what made the decision -- not easy, probably, but a lot clearer.