Why almost every MLB game begins with a fastball

Of the more than 4,000 first pitches this season, only 106 weren't fastballs. How can this make sense when fooling hitters is the name of the game? Tony Quinn/Icon Sportswire

On Sunday, MVP candidate Matt Carpenter dug into the box against Cy Young candidate Max Scherzer at 1:06 p.m. ET. Scherzer threw the first pitch of the game, a curveball at 77 mph.

So, to answer a question:


But it's an interesting question that turns out to have -- thanks in part to Scherzer and that hook he threw to Carpenter -- an interesting answer.

It is true that almost all baseball games begin the same way. At 12:05 p.m. ET on Aug. 1, for instance, Tommy Milone threw the first major league pitch of August, and if I told you to imagine a first pitch of a game, you'd get this one exactly right: a fastball, firm and four seams, right down the middle, taken for a strike. The next first pitch -- in the Bronx, one hour later -- was only a little different, a Sonny Gray two-seamer, low for a ball. Then in Detroit, a Mike Fiers four-seamer on the inner half for a called strike. Speed through the next five days, and the first 64 games of August all began with first-pitch fastballs, before Mike Leake opened his Aug. 5 start with a changeup.

There have been just over 2,000 games played this season, and only 59 of them -- through Friday -- began with anything other than a fastball in the top of the first. First-pitch non-fastballs to start the bottom half of the first inning turn out to be just as rare, so we'll henceforth lump both pitches together so we can get a more robust sample: More than 4,000 first pitches this season, of which only 106 were anything but fastballs. Approximately 2.5 percent. That's so few! There are more non-fastballs thrown on 3-0 -- approximately 4 percent -- than there are to start starts, which makes the first pitch of the game arguably the single most predictable moment that exists in baseball.

Clayton Kershaw has, since at least 2010, never started a game with anything but a fastball. Justin Verlander hasn't either. Bartolo Colon, Jon Lester, Madison Bumgarner: not one. Jake Arrieta hasn't since 2012, when he was still an Oriole and still terrible. Trevor Bauer never had until this July, when he dropped in a first-pitch curveball against the Reds.

Pitching is so much about keeping hitters off balance, guessing, uncertain about what comes next. That's why catchers put down signs instead of just shouting out instructions. So why, for this one moment, does such predictability exist?

Some pitchers do have more first-pitch variety than others. Since his debut, Masahiro Tanaka has begun 42 of his 128 starts with non-fastballs, mostly sliders, but occasionally curves or splitters. Leake has started 33 of his 260 starts with sliders or changeups. Tanaka and Leake have nine and eight such starts, respectively, this year, the most in baseball.

Scherzer is tied with Leake, which brings us back to Monday against Carpenter. Until last summer, Scherzer almost never threw anything but fastballs to start games. From 2010 through mid-June 2017, he started almost 250 games and threw fastballs at the commencement of all but one of them. But on June 27, he started Anthony Rizzo off with a slider (called strike), and a month later, he threw David Peralta a slider (called a ball). This year, he has begun eight games with changeups (2), curves (3) or sliders (3).

Scherzer is known to be a smart pitcher, and one who pays a lot of attention to the progression of at-bats, and this new tack could be seen as a tiny, microcosmic example of his wiliness -- particularly because of what else is going on in the game.

Beginning in 2015, the league's leadoff hitters became much more aggressive swinging at the first pitch of the game. From 2009 through 2014, they swung at the first pitch of the game just 12 percent of the time. In 2015, it spiked to 20 percent, and it has been 19 percent in the years since. This coincided with a league-wide shift toward putting more powerful hitters into the leadoff spot: Leadoff men this year have an OPS+ of 107, which is to say an OPS 7 percent higher than the rest of the lineup's hitters. That's the leadoff spot's highest OPS+ since 1914. So: Leadoff hitters swing at the first pitch of the game more aggressively than they used to, and they are scarier hitters than they used to be.

That's almost certainly what explains Bauer's first-ever non-fastball opening. It was July 10, and Scott Schebler was leading off. Schebler can really hit -- he had 30 homers last year and has hit about as well as Cody Bellinger, Matt Kemp and Kris Bryant this year -- and he is outrageously aggressive on the first pitch of the game, swinging at it a whopping 65 percent of the time during his 26 leadoff appearances. He has put seven of those 26 first pitches in play, four of them for hits, one of them for a home run. The day before Bauer faced him, he'd singled on a first-pitch fastball to lead off against Bauer's teammate Mike Clevinger.

So Bauer threw him a first-pitch curveball, to keep him off-balance. You'd think, in a world with powerful leadoff men swinging more often at first pitches, there'd be a lot more of this, that more pitchers would be doing what Scherzer is doing. But Scherzer is still the exception. This year has seen only a tiny bump in non-fastballs to start games:

  • 2010: 2.8 percent

  • 2011: 2.5

  • 2012: 2.0

  • 2013: 2.2

  • 2014: 1.6

  • 2015: 2.1

  • 2016: 2.2

  • 2017: 2.2

  • 2018: 2.7

So why would pitchers throw something so predictable? Surely a curveball in such a spot, when a batter is almost certain that a fastball is coming, would be virtually unhittable. It seems irrational, with leadoff hitters increasingly looking to swing and increasingly capable of doing damage, to throw the one pitch they're looking for.

The twist is this: The element of surprise doesn't seem to matter. Even when batters know the fastball is coming -- or think they do -- first-pitch curveballs and sliders haven't really been any more effective.

We have, since 2010, about 42,000 first-pitch fastballs, and 960 first-pitch anything-elses. Whether it was a fastball or not has had no effect on whether a batter will swing (15.1 percent either way) or whether he will get a hit when he puts it in play (.364 average against fastballs vs. .368 otherwise). But pitchers are a lot more likely to throw a first-pitch curve or changeup out of the zone, and -- because the default is still that the leadoff batter won't swing at the first pitch -- that leads to more 1-0 counts.

If we had 100 pitchers throw first-pitch fastballs to start off games and 100 throw first-pitch curves, changeups or sliders to start off games, these are what we'd end up with after the first pitch:

The curveballs avoid that extra hit, to be sure, but at the expense of four 1-0 counts. Do the math and the differences mostly cancel each other out, with the non-fastballs producing slightly more offense than fastballs. (An extra run every 700 or so games.) Which means that throwing exactly the pitch batters are expecting is perfectly rational. It might not be in any other situation -- pitchers are five times more likely to throw a non-fastball on the first pitch to the second batter of the game -- but here, in this spot, at that moment, with the pitcher and batter both just settling in, the element of surprise means nothing.

If leadoff batters start to swing a lot more -- 25 or 30 percent, as batters do on the first pitch of non-leadoff at-bats -- that could change. Pitchers will adjust, if forced. Scherzer, for example, was forced: Batters this year are swinging at almost half the fastballs he throws to start games. They've got four hits against his first pitch this year, including a triple, and all four of those hits came on fastballs.

And that explains why Scherzer threw that first-pitch curveball to Carpenter. Carpenter took it low for ball one.

Thanks to Rob McQuown and Baseball Prospectus for assistance. Pitch classifications come from Pitch Info.