First Pitch Follies by Ben Clemens March 20, 2020 One of the joys of baseball, and sports in general, is that the narrative arc of the game isn’t preordained. You can’t know when the most important pitch of the game will be before the game starts. This isn’t a TV procedural, where nothing decisive can happen in the first 20 minutes. The visiting team might go up 3-0 in the first inning and never relinquish the lead, or they might rally furiously from down five only to lose in the bottom of the ninth. Even though the most exciting pitch of the game isn’t a given, one thing more or less is: the first pitch of a game won’t be the most exciting one. That’s partially due to the rules of baseball — no one is on base, and most at-bats take more than one pitch — but the first pitch is unique in its own way. For one, no one swings. Combining the first pitches thrown by each starter in a game, batters swing at 23% of offerings, significantly lower than the 29% overall swing rate on 0-0 counts. Secondly, it’s almost always a fastball. Sam Miller delved into the thinking behind game-opening fastballs, and pretty much everything from his piece still holds. Pitchers throw fastballs because batters don’t swing, and batters don’t swing because they already don’t swing much on 0-0, and particularly so when they haven’t seen the pitcher throw anything yet. But batters aren’t static opponents. In 2010, they swung at 25.1% of 0-0 pitches. In 2019, that number was a meaty 29.4%. Strikeouts are rising, pitchers are fastball-happy on 0-0 counts, and batters increasingly can’t afford to hang around waiting for something to hit given the decline in overall fastball usage. Pitchers have responded to batters’ newfound aggression by throwing fewer fastballs to open at-bats. They threw 68.6% fastballs to open at-bats in 2010 and only 62% in 2019. It’s a natural reaction, and part of the reason that swinging strikes on 0-0 have gone up from 5.3% to 7.9% in 10 years is that batters are coming up empty far more often when they swing on 0-0. In 2010, they whiffed on 21.2% of swings. That was up to 26.9% last year. The new whiffs have come on both fastballs and secondary pitches, but it’s not a great leap of imagination to think that some of the whiffs on fastballs are due to batters having to guess what’s coming instead of sitting dead red. It’s a complicated dance, made more complex by the fact that the incentives aren’t perfectly aligned; even if batters swung at a higher rate than they already do, falling behind 1-0 is detrimental enough in this day and age that pitchers would need to keep flinging fastballs. Intuitively, you’d expect these same trends to hold on the first pitch of the game. But they haven’t! Remember those overall 0-0 count numbers from above? In 2010, pitchers threw 68.6% fastballs to start batters off. That number spiked to 97.2% (!!!) on the first pitch of the game. By 2019, overall first-pitch fastballs fell by 6.6 percentage points. Meanwhile, first-pitch-of-the-game fastballs fell by a whopping 0.5 percentage points to 96.7%. They’re about as prevalent as they’ve ever been. That sounds weird, and it is weird. Under the hood, it gets potentially stranger. If you stop to think about it, a 96.7% fastball rate can’t be uniform across all pitchers. You’d have to make 33 starts and throw 32 fastballs to hit that mark, and given that most pitchers don’t make 33 starts, the distribution is necessarily going to be uneven. But it’s so uneven, friends. So uneven. There were 175 pitchers that made 10 or more starts last year. A good 106 of them, or 60.5%, threw only fastballs to start games. On the other end of the distribution, 16 pitchers threw at least 10% secondary pitches, led by Masahiro Tanaka, who threw only 21 fastballs in his 31 first pitches. On the low end, it was almost hard to choose a leader; Charlie Morton, Stephen Strasburg, Reynaldo López, Madison Bumgarner, and Gerrit Cole all started with 33 fastballs in 33 starts. That impressive list should tell you something: it’s not a matter of good pitchers switching it up and bad pitchers throwing all gas. In fact, if you split the data between pitchers who threw all fastballs and pitchers who threw anything else, they were nearly indistinguishable: the all-fastball group allowed a .324 wOBA to batters overall, while the group that mixed it up clocked in at .325. Even though the groups have similar results overall, the ones who made batters think a little did better on these first pitches, right? It’s only logical; a single curveball on the scouting report makes batters think twice before sitting on a fastball. There’s just one problem; the data doesn’t really bear that out. FanGraphs’ pitch weights are based on a simple idea; look at the expected run value after a pitch (as measured by wOBA), compare it to the expected run value before the pitch, and use the difference to determine the value of the pitch. We use them to value pitch types, but with a little effort we can simply value first pitches of games, regardless of type. Taking our two groups again, it’s hard to find a difference. Every 100 pitches thrown by the all-fastball group saved them 0.26 runs. That’s pretty close to zero; the pitches hardly affect the outcome. Our other cohort, the mix-and-matchers, saved 0.75 runs per 100 pitches. That’s better, but not by much. You can roughly blend it back to ERA if you think pitchers average around 6 innings a start; over 100 starts of 600 innings, the wily half of our sample would save 0.5 runs. That comes out to 0.008 points of ERA, a rounding error. Why isn’t it better? Several little factors add together. First, a single pitch can only be so impactful. There’s only one first pitch per game, and far more baseball happens after that. If you could make the whole game out of first pitches, it would be a different story. There are other issues, too: the secondary users were wilder; they threw 40.5% of their first pitches in the zone, as compared to 45% for the all-heat group. Batters weren’t passive enough on pitches in the zone to make up for it; in fact, they swung at 36.7% of the pitches secondary users threw in the strike zone, as compared to 35% for the fastball-only types. The silver lining, the factor that makes the secondary pitch users ever so slightly better, is what happens after batters swing. They whiffed on 17% of swings against pitchers who weren’t all fastballs and only 13% against the all-fastball group. When you consider how valuable it is for the offense to put a fastball into play, and how much better it is to be ahead in the count 0-1 rather than backing up first base, the whole situation gets a lot closer. Every 100 first pitches put in play adds 8.5 runs of expected value to the offense while every 100 first-pitch strikes subtracts 4.6 runs of value, so it doesn’t take many batted-ball-to-whiff transformations to balance things out. For the moment, this all still works out to nothing. There’s no real calculable edge to keeping batters guessing; the pitchers who do it give back some of their contact-avoidance gains by missing the strike zone more often, and because it’s only a single pitch, the maximum possible effect is small. If batters increased their aggression, however, the equation might flip. For now, there are still enough Gerrit Cole types — all fastballs, but a 50% whiff rate — to offset the Matthew Boyd types — all fastballs, 0% whiff rate, and two extra-base hits to lead off games. For every Tanaka, there’s a Michael Wacha — he spun his fair share of first-pitch curveballs but still got tagged for a homer and two singles. When baseball starts up again, you probably won’t see the first pitch of the game. You’ll forget that it starts at 7:08 instead of 7:10, or you’ll be grabbing a beer from the fridge. If you’re at the game, it’s even less likely that you’ll see it — those concessions aren’t going to purchase themselves, after all. But even though you don’t see it, there’s a cat and mouse game going on in that pitch. Baseball contains multitudes. Update: This article has been updated to clarify that a key reason the difference in first-pitch samples is small is due to the first pitch being a small sample of the overall game. The differences in results would be meaningful were the entire game made up of first pitches.