Are Opening Pitchers Really A Good Idea?

Today at Freakonomics Blog, the idea of an “opener” position in baseball is thrown around – again. While George Will considered the idea and in fact it has been experimented with for over a hundred years in minor leagues, it’s still unusual enough to merit inclusion in a place that touts “new” or “outside the box” ideas. The basic concept of an “opening pitcher” is that if you’ve got a good pitcher who can consistently deliver a good inning or two, why not use that pitcher to start a game rather than as a closer or a middle reliever? It’s kind of a tasty-seeming alternative – you could free up some resources if you start with a really good pitcher even if that pitcher suffers from low endurance.

The big downside to this idea comes from the confluence of the sub-out rule and the fact that while all pitchers have good and bad nights, some pitchers nevertheless have more endurance than others. So if you’ve got a pitcher who can sustain 100 pitches or so, but who is having a bad night, you want to know that as early as you can so you can make substitutions while they’re still on the roster. If you use up your consistently-good-but-low-endurance middle reliever by having him open up the first two innings of the game, you can’t go back and put him in again. You still need a middle reliever.

Put yourself in the shoes of a major league manager considering this option. If everything goes as planned in a typical modern MLB game, a pitcher ideally will throw an average of about 15 pitches per inning, needing, 5 pitches to an out. (These are averages.) If things go well, your inning pattern goes like this:

Here, “A” is the starter, who throws an industry-standard 100 pitches. “B” is the middle reliever, who throws 20 pitches, and the ace of the squad, the closer, throws the last inning. This works great if you can average out 15 pitches an inning and your starter does not give up a lot of runs in the 6 ⅔ innings he’s asked to throw. (Oh, and if your bats get you ahead of the game by the close of the eighth inning.) You use three pitchers, one of which you need to rest for several days and the rest which you can draw on for tomorrow’s games because they haven’t dislocated their shoulders all that badly.

The “opener” proposal, when things go well, looks like this:

That’s a little bit less strain on “A”, who is now a “stretch pitcher” rather than a “starter,” so he recuperates a little bit faster. You’re putting only a slightly heavier workload on “B,” and getting the same job done. In fact, you’re probably at a bit of a defensive advantage, because switching the pitchers from “B” to “A” in the third inning will throw the batters off and not let them settle into a rhythm or a read of the pitchers early on in the game. They’ll get there, but only towards the end. And you can always pull “A” and put in “C” if you sense that is happening.

But things don’t always go to plan. In a regular-strategy game, let’s say that “A,” the starter, has a bad night. He’s throwing wild, or just plain gives up too many hits and runs and you have to pull him in the third inning. Now, your anticipated pitching scenario looks like this, assuming that you are able to salvage a win with your offense:

To dig yourself out of the hole “A” gets you into, you need five pitchers, and “A” is now fatigued enough that he has to wait for the regular rotation to go around before he’s ready to play again. And you need to rely on offense to do it for you, which may mean pulling pitchers for pinch-hitters if you’re in the NL. An ugly game to win, but a “W” is still a “W” even if you wind up winning with a score like 14-12. Now, what happens if we change a former relief pitcher (“D” in the above example) into the “opener,” and the guy you used to call a “starter” and now is expected to pitch for endurance, starts going wild and giving up the hits?

Now, you’ve got two fewer innings for your offense to save you, and one fewer pitcher in your bullpen to rescue you if the former starter, A, also starts giving it up like it’s prom night. You’ve cut your margin of error, in exchange for a very small marginal defensive advantage, which has already been squandered by virtue of the bad night your stretch pitcher has suffered. So the reality of it is you’re likely looking at a stop-loss spiraling out of control rather than a game in which you can salvage a win. So I think the idea of the starter being expected to last six or so innings makes sense, because you shouldn’t put players in until you need to.

On the other hand, what if we do away with the idea of pitchers being asked to throw 100 pitches a night altogether? Your typical team has twelve or thirteen pitchers, four or five of which make up the starting rotation, one of which is the “ace” who is the closer, and the rest who are relief. But the rules only say that you have twenty-five spots on your roster and once a player is substituted out, he cannot come back in. So the half-pitcher rule isn’t set in stone. You could have, say, nine pitchers, each of whom is expected to play only one or two innings every night. Perhaps this means not every pitcher faces every batter, but so what? An ideal playout here looks like this:

That looks pretty good, but we can’t judge the strategy only by how well it works out under optimal conditions. The obvious downside here is that you are increasing the chances of putting in a pitcher who has a bad night. So if things go wrong and you have a guy who has a bad night, you pull him after one inning. Let’s say, in fact, that you have two guys who have bad nights – then you might get something that looks like this:

In this example, pitchers “B” and “D” both have bad nights and it takes them twice as many pitches to get their three outs. A single pitcher makes up for two bad pitchers, which is good for long-term rotation and wear and tear; no one throws more than 30 pitches in the night so they call can recover and pitch again in a day or two. With this, you wouldn’t need a twelve-to-fourteen deep bullpen but could probably do with maybe only nine. That frees up three to five spots for offensive players. That, in turn, gives you margin for error on the defensive side of things. (Fortuitously, it also means a higher-scoring game, which the fans and TV tend to like, increasing advertising revenues.) You could even pull the pitchers who are doing badly after 20 pitches or so of their floundering around:

Sounds good, right? Sure, you use most of your bullpen each night. But each of them has only seen light work, and they’re all ready to go again the next night because you’ve chosen pitchers who are all middle relievers and have fast recuperation. If you’re losing, well, that sucks, but again, pretty much your whole squad will be good to go again the next night. Here’s the real scary risk to this low-endurance, high-substitution strategy: extra innings. If you wind up having 10 or 11 innings on a night when at least some of your pitchers are a little off, you’re going to need a deep bullpen.

Ten pitchers you go through. If you’ve shorted your bullpen to load up on offense, you are ass-out at this point and have to start having your backup catcher pitch when “I” gets worn out. So maybe you’ll stick with 12 or 13 pitchers on your roster, and just play them differently? Now you’ve given up the offensive advantage that this strategy is intended to give you – you’re supposed to have enough good batters filling those pitcher slots that you don’t often need to get to using all nine pitchers on your squad. This also takes effectively out of contention the possibility of bringing in an opposing-handed pitcher specially to face down your opponent’s best hitter in a clutch situation – a strategy used now in quite a lot of games. Because you need to eke at least one inning out of each pitcher, if the batting rotation lines up unfavorably for a particular pitcher, he might be stuck or you might have to take on an unreasonable risk to swap him out with someone better-suited to face a dangerous batter.

The protest may be, “Yes, these are foreseeable risks, but are they significantly more dangerous than the sorts of risks that a manager faces now?” Yes, I think they are. If your starting pitcher now gets into a clutch situation and the other team’s power hitter comes up to bat with the bases loaded, it’s not a huge cost to pull the starter and bring in a lefty to get a single out. And the more pitchers you have taking the mound, the more you’re spreading around the risk that each pitcher is going to perform poorly. On balance, I think it’s better to start out with a pitcher who will last six innings or so and not give up very much, and then start shifting pitchers around in the later parts of the game. The alternative – constantly-shifting pitchers and a depleted late-inning bullpen – deprives the manager of flexibility towards the end of the game, which is precisely when flexibility and options are needed most.

My verdict is that an “opening pitcher” style rotation only works well over time if you’ve got at least six pitchers who can all be counted on to deliver between 20 and 40 good pitches every night.  And if you’ve got that, well, you’ve got more options than you needed anyway.