keyboard_arrow_uptop
Baseball Prospectus is looking for a Public Data Services Director. Read the description here.

We talk about pitching decisions in terms of leverage, but most of what we think of as modern pitcher usage—five-man rotations, seventh-inning guys, closers, LOOGYs—is born only secondarily of leverage. The real factor is attrition, and the acknowledgment that we can’t expect the team’s best pitcher to throw every inning of every game without breaking down or getting much, much, much worse from overuse. Say we know that Craig Kimbrel is less likely than any Braves pitcher to allow a run in any situation, but he can only do it 75 times all year; the goal is to avoid wasting a single one of those 75 on a low-leverage situation, thus dooming some other high-leverage situation to an inferior pitcher. Without the attrition—if Kimbrel could pitch all day every day—there would be no decision. Kimbrel’s manager would look very smart.

There are tons of decisions, though, because it turns out that getting the right guy in the highest leverage situation is practically impossible. An extremely high-leverage situation might come in the sixth inning. Does Kimbrel come in then? Well what happens if he gets out of it, but then a higher leverage situation comes up in the eight. What are you, some kind of idiot using your best reliever in the sixth? What if the highest leverage situation comes up in the second? Now you’re without your starter and you’re without your best reliever and a bigger situation might be coming in the ninth inning and you’re eight pitchers deep by then. Or, maybe it only looked like a high-leverage situation was coming in the second inning; maybe, while you were warming your guy up, the pitcher on the mound got out of it. Unless you can warm up your relievers constantly without fear of exhausting them, you simply can’t have them ready for the biggest situations at all times. All of this uncertainty makes it impossible to use your best pitchers exactly as much as they are capable of.

So managers more or less estimate, using rules of thumb and/or rigid reliever roles that simplify the decision, that keep the players happy, that preserve the pitchers' arms, and that do a pretty good job of making sure the best pitchers get the most innings and that the next tier of the best pitchers in the game pitcher with the highest leverage. It all follows a simple, two-word poster: WASTE NOT. Thinking about your pitching decision in terms of leverage is a pretty good way of dealing with one of baseball’s most screw-you challenges.

It’s also totally wrong in a game like this, which Bumgarner should throw the first two innings of.

[Author caveat: There are other reasons that would come into play, most notably Hudson's comfort coming out in relief compared to Bumgarner's. Also, there are reasons not to use Castilla in the third, or whatever dumb thing I might suggest in a hypothetical to come. I can't actually say who should start. Nor do I care that much.]

The idea of waste is completely undone in a game like this, with a tomorrow like tomorrow. Imagine this scenario: The Giants are down 10-0 going to the bottom of the eighth. Let’s say, hypothetically, that Bochy has used all of his pitchers except for Bumgarner, Hunter Strickland, Jean Machi, and Ryan Vogelsong. Who should Bruce Bochy have pitching in that frame? It is obvious: He should use Bumgarner. Bumgarner is the Giants’ best pitcher. Using him down 10–0 in the eighth inning would be a terrible waste at any other point of the season, up to and including yesterday. But there is nothing to save him for now, and using him would probably increase the Giants’ win probability from, oh, .001 percent to .002 percent. The right move would be Bumgarner there.

Now let’s back up an inning. Say it’s 10-0 going to the bottom of the seventh. Say you’ve figured Bumgarner’s got two innings in him. Who should pitch? Bumgarner!

So you can keep this hypothetical going and just keep moving backward. Is Petit the next best pitcher? Then no matter the leverage, if he is available for the sixth, he must be the next best option. Casilla? Use him if you got him. Are Affeldt, Romo, Lincecum and Lopez, used in matchup work, next best? Pencil in two innings for them as a group, maybe three. Even if the Giants are down 10-0: It makes no sense not to use the best pitchers. There is no waste. There is nothing to save.

So now maybe we’ve got an imaginary hypothetical of the game that goes like this, regardless of score:

  • Innings 1-2: Hudson
  • Innings 3-4: Matchups
  • Innings 5-6: Petit, Casilla
  • Innings 7-8: Bumgarner

Not bad. Could produce a winner (in which case we’d have to fill the ninth inning, but we’ve been conservative with these pitchers; if a win looks likely it’ll be easy enough to adjust). But even if the Giants fall behind, it gets the best pitchers in the game to keep it close, at the expense of literally nothing.

But the hypothetical I offer is pretty weird, right? Why would Bochy used his pitchers in almost the exact reverse order of how good they are? He wouldn’t, right?

So now: Why not flip it?

  • Innings 1-2: Bumgarner
  • Innings 3-4: Petit, Casilla
  • Innings 5-6: Matchups
  • Innings 7-8: Hudson

Has anything changed? In a general sense, no. Nothing has changed. You’re still getting the best pitchers in the game at the maximum amount possible. Could produce a winner, but even if the Giants fall behind, it gets the best pitchers in the game to keep it close, at the expense of literally nothing.

But in small ways, the difference is crucial. This way guarantees that you get your best pitchers in the game, and that they will be limited only by how long they can go, rather than how long you've let Hudson go before them.

Further, in the first scenario—the Bumgarner at the end scenario—there is too much uncertainty to think a manager could stick to the plan. For instance: What if Bumgarner comes out for those final two innings and is terrible? If you’ve saved him until the end, you’re screwed; you’ve got nothing left, and you’re kicking yourself for pulling Casilla or Petit or Lincecum after just one inning, or Hudson after two. In that scenario, now we’ve suddenly got Hunter Strickland trying to get outs. If you’ve drawn up a plan for how to use your pitchers and Hunter Strickland is on it at the expense of more Casilla, your plan sucks. So, to avoid the Strickland outcome, Bochy has to save a better pitcher for that possibility. Maybe he saves Casilla. And now we've got a world where Casilla might not pitch in the game at all—even though we've decided he's the second or third best pitcher available. Ugh. There are just 500 ways to find the suboptimal pitcher usage if you try going in reverse order.

Now, you might argue that Bumgarner in the seventh or eighth is more valuable than in the first or second, because the leverage index is higher. That’s just an optical illusion. Ignore that.

You might also argue, as Russell Carleton did, that Bumgarner later in the game could be used in a truly high-leverage situation—e.g., to get out of a bases-loaded, Hosmer-batting jam. He could. If that’s Bochy’s plan, there’s some logic to it. I’d be very surprised if that’s Bochy’s plan. Bumgarner is a starter who takes longer to warm up. The high-leverage jam doesn’t always give a lot of advance notice, and what Bochy definitely won’t want to do is have Bumgarner warm up and not get in the game. It’s not impossible, but I’d be very surprised if Bumgarner doesn’t start a clean inning.

So what’s the best way to use Bumgarner if he’s pitching a clean inning? Well, preferably he’d be pitching against the Royals’ best hitters. Maybe the guys batting first through sixth. There’s a way to make sure of that, you know.

Look, here’s the basic point I’m trying to make: Managers make lots of reasonable decisions that create a little waste (not using Kimbrel quiiiite as much in each game as he perhaps could, or in as high of leverage as he perhaps could) to prevent a lot of waste. That’s what’s probably going to happen with Hudson. As I put it to Russell: "Hudson will have to pitch, but the question is how much he'll have to pitch, and the preference is 'as little as possible.' So feels weird to start him, in which role and uncertainty about the rest of the game will lead to 'as much as possible.'" Bochy, unsure of how many innings he can truly count on from his bullpen, will try to get him a little deeper than is best. Or, the opposite: Bochy will have him on an extremely short leash, but maybe later in the game we realize that, whoops, that leash was too short and here comes Strickland. In a game like this, the decisions should be simple: You need 27 outs. You are going to require a bunch of different guys to get them. You want those 27 outs to be as heavily tilted toward your best pitchers as possible. So start with your best pitcher. Do the one thing you can do to take the uncertainty out of it.

Thank you for reading

This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.

Subscribe now
You need to be logged in to comment. Login or Subscribe