What If They Made the Entire Pitch Mix Out of Fastballs?

Bryce Miller led off last Friday’s game with a fastball to Yandy Díaz. He missed low and away, and Diaz took the pitch for ball one. Miller came back with another fastball, wildly away, then another one low to send the count to 3–0. But then things turned. He threw another fastball right down the middle for strike one, fired one past Díaz for strike two, and finally threw a perfectly-located missile that Díaz waved through. Six pitches, six fastballs, and one out: it’s exactly how Miller drew it up at home.

The rest of the first inning held more of the same. He threw Wander Franco three straight fastballs to garner another strikeout, one that Luke Raley tapped into play (he reached on a fielding error), then six straight fastballs to Randy Arozarena to duplicate the 3–0-to-strikeout path he’d walked against Diaz. All told, Miller threw 16 fastballs in the first inning and nothing else whatsoever.

To some extent, this was an unsurprising tactic. Miller boasts one of the best fastballs in baseball, period. It’s one of those mindbending pitches that seems to defy gravity by rising out of its path halfway to home plate. Whether you’re looking at it statistically or visually, it’s hard to find faults. He’d used it nearly two-thirds of the time heading into last Friday’s game.

Starters rarely throw any fastball, even a dominant one, that frequently. But there’s a second piece to Miller’s puzzle: his secondary pitches are lackluster at best. He has a gyro-style cutter/slider deal — “hard slider” for the rest of this article — that he throws in the mid-80’s, but opponents have done a great job making contact against it. He hasn’t been hurt by that contact yet — thanks, .156 BABIP against — but that feels like a fluke. Batters have generally hit the ball quite hard when they make contact, and again, they’re making a lot of contact.

That’s his good secondary. He also throws a sweeping slider, but that pitch sports an impossibly low 6.5% swinging-strike rate, and opponents have gotten as lucky against it as they’ve gotten unlucky against the cutter; they’re BABIP’ing .471 with a slugging percentage in the 600s. In other words, Miller would be well served to duck after releasing it.

He’d already been pushing the limits of how many fastballs a starter can throw these days, but last Friday’s start was something else entirely. In the second inning, Miller threw an 0–1 hard slider to Isaac Paredes to draw a swinging strike. He threw another hard slider to Christian Bethancourt in the third, and a sweeping slider that Díaz fouled off. Through three innings, he threw 41 pitches; 38 were four-seam fastballs. Even saying it out loud, I can’t quite believe it.

There are a ton of things I wanted to measure in this start. Does throwing so many fastballs create diminishing marginal returns? Is there some breaking point where the strategy becomes counterproductive? How many great fastballs in a row would you have to throw a batter before a so-so slider becomes a better option? Sadly, I don’t think we’ll get the answers to those questions today, because those three innings are about all we have to go on. Miller developed a blood blister toward the end of the third inning, and though he got treatment on it before the fourth, he was clearly feeling the effects. He lost a few ticks of velocity and also more spin than you’d expect from the velo decline. In his postgame presser, he mentioned that he was having trouble putting pressure on the ball given the location of the blister. That tracks with the pitch data; he got less movement even with slower pitches, which theoretically have more time to break. The last pitch he threw of the day was his second-slowest and least-spinny, and Arozarena hit it to Bellevue for a two-run homer. Exit Miller, and unfortunately for him, it looks like it’ll be for a while: he’s on the IL now, and manager Scott Servais hinted that he’ll be out for more than the minimum.

While I listed a ton of questions I don’t have answers to, I still wanted to poke around at the data. Times through the order as a way of measuring familiarity effect? Right out the window; Miller only faced Diaz twice before his blister issue. Number of fastballs in a single at-bat? There’s not really enough data. But here’s a question I think I can answer, or at least engage in profitable speculation about: what led Miller and catcher Cal Raleigh to lean so heavily on the pitch in this start?

One neat thing about modern baseball: teams can diagnose players on the fly. If a pitcher throws his first three fastballs and doesn’t have his usual velocity, the pitching coach will know right away. Missing some spin rate? Same deal. Release point messed up? Teams can tell right away and potentially fix it in between innings. Maybe something the Mariners saw in Miller’s first inning led them to emphasize his fastball further than his already-high usage rate.

To test this, I grabbed two sets of data: all of Miller’s first-inning fastballs split by start, and his subsequent fastball usage in each start. I asked a simple question: which statistical markers were correlated with higher subsequent fastball usage?

Here’s an example of one that doesn’t work: velocity. Miller’s highest first-inning velo came on June 24, and he threw only 52.5% fastballs thereafter. His lowest average first-inning velocity was June 4, and he threw 62.5% fastballs thereafter. Last Friday featured his second-lowest first-inning velocity, and as we know, he kept the pedal to the floor the rest of the way with fastballs. In aggregate, I found a -0.10 correlation between velocity and fastball usage — essentially no relationship.

That makes sense to me; looking at a radar gun is definitely not the best way of telling whether someone’s fastball is on. What about spin-velocity ratio? That’s spin rate divided by speed, and it does a good job of accounting for the fact that the harder you throw a pitch, the more it spins. That has a strong positive correlation: 0.71. That means that half of the variation in fastball rate can be explained by variation in first-inning spin-velocity ratio.

Miller’s own words gave me another idea. In his postgame comments, he noted that Rays hitters “weren’t really close” on his pre-blister fastballs. In other words, they were swinging and missing a lot. Does that mean more fastballs? It sure does; first-inning swinging-strike rate has a 0.63 correlation to subsequent fastball usage, and first-inning whiffs per swing checks in at 0.68. In other words, you don’t need a fancy spin readout: if Miller is missing more bats early on, he throws more fastballs later.

None of these conclusions are statistically significant. How could they be? Miller has made only 11 starts this year, and this is the kind of behavior that’s hard to analyze because it’s inherently unstable. He and Raleigh are making a conscious choice about what to throw, and they’re taking the other team’s behavior into account when doing so.

In his first start back from the IL, Miller might expect opposing batters to sit fastball, so clearly, he should throw fewer fastballs than would be suggested by his first-inning performance. But wait! Batters know that Miller will expect them to sit fastball, so perhaps they’ll sit on his weaker secondaries instead. Clearly, he should throw more fastballs than you’d expect. But wait! Batters know that his secondaries are weak and can probably hit them even without looking for them, so they’ll be waiting on the fastball; maybe Miller should throw fewer fastballs than you’d expect. Aside from fighting a land war in Asia — always a mistake — figuring out what to do in these two-player confrontations is notoriously difficult. That’s the entire study of game theory, and I doubt the Mariners’ battery has time to break out their textbooks in the dugout between innings.

It’s a bummer that Miller got injured. That’s true for many reasons; it’s a bummer when any pitcher gets injured. Baseball is better when its best players are on the field. Even more than that, though, I wanted to see how this thing went. Would the Rays have adjusted? Would Miller have counter-adjusted? What did he and Raleigh have planned? Those hypotheticals are fun, and the real-life outcome was just boring: it turns out that if you get hurt, you pitch worse.

That’s the end of my Miller-related thoughts, but don’t worry, there’s plenty more rattling around in my brain. Do other fastball-dominant starters do the same thing? Does the relevant metric change from pitcher to pitcher? Miller, for example, dominates hitters with his fastball because of its spin. Do location-first pitchers change usage based on their ability to clip edges in the early frames? Do fireballers look at velocity? Does everyone look at swings and misses? Weak contact? There’s no end to the research that can be done.

That’s just for fastballs. Do pitchers with a wide array of secondaries change their usage based on either their subjective feel or some objective data? Might a pitcher feature a curveball one start and a slider the next based on how each is spinning? That’s a question for another time — probably a time in August, after the trade value series and trade deadline. But it’s one totally worth researching, and one with plenty of space to be explored, so if you have some ideas about how to look at the problem or what to look at in the first place, I’m all ears.

Thanks to Kiri Oler for a transcript of Miller’s postgame interview as well as some brainstorming about how to approach this problem without the benefit of a second time through the order.