The Three Relievers With Four Fastballs

Of all 193 relief pitchers with at least 20 innings pitched this season, exactly three have thrown four distinct types of fastballs a minimum of 20 times each: Reed Garrett, Chris Martin, and Cole Sands. They all have one non-fastball offering, but none of them throw it more than a quarter of the time. Justin Choi wrote recently about the strategic options available to pitchers with more than one fastball, but four? Four whole fastballs? These guys feel like doomsday preppers getting ready for some apocalyptic scenario where money is now worthless and fastballs are the new currency.

But anytime a new strategy pops up in baseball, it’s worth checking to see if the outliers are onto something others should attempt, or if their “one weird trick” to pitching works only for them. Shoot, maybe it doesn’t even work for them all that well. Regardless, we’re gonna get to the bottom of what’s going on with these pitchers and all the fastballs they’re hoarding.

Reed Garrett

Garrett has thrown 34.2 innings for the Mets so far this season, posting a 3.12 ERA and a 3.17 FIP. He’s struck out 37% of the batters he’s faced and walked 12%. His performance this year has earned him an ERA- of 81, firmly better than average. What the averages aren’t telling you is that Garrett started the season with a 0.57 ERA in March and April, a ridiculous run that earned him a full breakdown on his evolution from last season by Ben Clemens on April 23. But that April ERA had to buy new pants after swelling to 6.08 in May. His performance has regressed somewhat in June, settling somewhere between those extremes. The current version of Garrett is probably more representative of what the Mets should expect from him moving forward.

The table below shows a breakdown of Garrett’s pitch repertoire with the usage and a few metrics for evaluating each offering (run value per 100 pitches thrown, xwOBA, Stuff+, and Location+). The two most common fastball types (four-seamers, sinkers) that most pitchers feature at the center of their arsenals are the pitches he throws the least. But the metrics linked to Garrett’s outcomes — either actual outcomes (RV100) or expectations based on the characteristics of the outcomes (xwOBA) — agree with his decision to de-emphasizing those pitches. They like Garrett’s four-seamer the least, even though it has his highest velocity and second best Stuff+. The pitch’s Location+ score reveals its critical flaw: a lack of command. Stuff+, RV100, and xwOBA agree that his sweeper and splitter are his two best pitches. Based on usage, Garrett agrees with that assessment.

His pitches mostly hover around league average in terms of individual characteristics, but the sweeper and splitter are both a tick or two harder than average and generate a bit more spin leading to more horizontal break, which is likely why Stuff+ likes them more than the rest of Garrett’s arsenal.

He makes the most of middling pitches by playing them off one another. The sweeper and cutter mirror the spin direction of the sinker and the splitter. As a result the pitches look similar out of the hand but fork in four different directions as they approach the plate to keep the hitter guessing (see movement plot below). So even if hitters guess the horizontal direction correctly, they’ve still got two similarly spinning pitches that fan out vertically as they approach the plate.

Garrett deploys all of his pitches no matter the handedness of the hitter, but he does vary the flavor of his approach. To lefties, Garrett likes to fill the zone with his cutter and dangle the splitter down and away when looking for a chase. To righties, he keeps the hitter off balance by throwing the sweeper to a variety of locations, but then comes down and inside at varying speeds with the splitter and the sinker.

The flowchart below gives us an idea of Garrett’s sequencing habits. He tends to start hitters with a cutter or sweeper. Once ahead in the count, he’s more likely to play around on the periphery of the zone with his sweeper and splitter, whereas while behind in the count he rolls with the four-seamer and cutter as more zone-friendly options. The wOBA values for plate appearances passing through each given count indicate the approach works well in early counts and with two strikes, but not as well when the count forces him back into the zone, in part because his four-seam command limits his ability to actually hit the zone with that pitch when circumstances demand it.

Here’s a representative example of how hitters respond to Garrett’s two-strike splitter.

Looking at swing metrics by pitch type, each pitch adds a valuable tool to his kit. The splitter is Garrett’s best combo play for inducing swings (56% swing rate) without courting disaster. The pitch owns his best swinging-strike rate (30%) and second lowest hard-hit rate (20%) when batters do connect. He gets batters to swing at 74% of the sinkers he throws in the zone, he uses the sweeper to induce weak contact (17% hard-hit rate), and turns to the cutter to mix things up. The four-seamer is the weak link in the chain so long as it keeps taking the scenic route to the catcher’s mitt.

Chris Martin

Martin has thrown 21.1 innings for the Red Sox in 2024, logging a 4.22 ERA with a 3.90 FIP. He’s struck out 28.2% of the batters he’s faced while walking just 2.4% of them. He has been on the IL since June 5 while proactively seeking help with anxiety.

Again, we’ll start with a synopsis of each pitch he throws according to the value metrics. Stuff+, RV100, and xwOBA all like his splitter best. The pitch is very similar to Garrett’s splitter from a velo/movement/spin perspective, but he doesn’t throw it nearly as much. His four-seamer is his next best pitch by RV100 and xwOBA, but fourth best by Stuff+. However, he locates it well enough to still get results. Martin’s cutter is his consensus third-best pitch, striking a balance between stuff and command to get the job done. Like Garrett, Martin’s non-fastball pitch is a sweeper, but unlike Garrett, he throws it so infrequently that it’s hardly worth discussing.

His pitch characteristics all hover around average, thrown maybe a tick or two harder, but with slightly less spin and therefore less movement. What helps overcome somewhat middling profiles is a distinct release point created by his long levers. Though his delivery is composed of a pretty standard three-quarters-ish arm slot, the arm attached to his 6’8” frame allows him to release the ball several inches higher and farther to his right than other pitchers throwing from a similar slot.

Rather than mirroring the spin on his offerings like Garrett, Martin takes a different approach to cultivating deceit. The puzzle for his hitters is more akin to spotting the difference between two nearly identical photos. All of Martin’s pitches spin in a similar direction, and his four-seamer, sinker, and cutter do so at almost the same spin rates. Where they differ is in the amount of active spin, or the amount of spin contributing to the pitch’s movement. The four-seamer, as one might expect, has the most active spin and the most rise. The sinker has a little less active spin and creates more horizontal break and more drop. The cutter drops in a comparable fashion to the sinker, but refuses to follow his fellow fastballs and break toward the third base side. Then there’s the splitter that spins at a much slower rate and with less active spin, which translates to roughly the same amount of horizontal movement as his four-seamer, but with even more drop than the sinker. Yet another carbon copy, but with a small but crucial edit.

Martin uses the same theory to guide his approach to both righties and lefties: Fill the zone with the primary fastball(s), use one of the secondary fastballs as a threat inside, and pepper the bottom of the zone with splitters. Against right-handers the four-seamer and cutter are the pitches he consistently throws to all parts of the zone and the sinker backs the hitter off the inner half of the plate. Against left-handers, Martin stays away from the sinker, so the cutter becomes the weapon he aims inside, while the four-seamer and the splitter maintain their existing roles.

The job of each fastball is further etched in stone by Martin’s sequencing, visualized below. He starts an overwhelming majority of hitters with the four-seamer or cutter and sticks to those zone-friendly pitches if he falls behind in the count. But if he gets ahead, Martin starts mixing in the splitter and sinker. His results tend to be better if he gets to those splitter/sinker counts, but it’s unclear whether that’s because of the effectiveness of those pitches or because he gets too predictable in unfavorable counts.

The swing metrics indicate Martin’s cutter is his best option for getting swings (55% swing rate) that lead to either strikes (13% swinging-strike rate) or weak contact (27% hard-hit rate). The splitter is his overall best bet for a swinging strike (19%), but when hitters do make contact, it yields the highest hard-hit rate (70%). The sinker is most effective when thrown in the zone because it has the lowest out-of-zone swing rate (18%) and in-zone contact rate (78%) compared to Martin’s other offerings. And avoiding contact is key, since the sinker has the second highest hard-hit rate (67%) of the bunch.

Cole Sands

Sands has pitched 32 innings for the Twins this season. Those innings have amounted to a 4.22 ERA and a 3.30 FIP. His strikeout rate sits at 28% and his walk rate is a measly 3%. Sands’ season trajectory mimics Garrett’s: on a rocket to the moon in April, a crash landing in May, and now back up and cruising at altitude in June. At his peak, Sands was striking out Shohei Ohtani on three pitches, and Minnesota was considering stretching him out to start while managing injuries in the rotation; now he’s settled into a multi-inning relief role.

Digging into Sands’ repertoire via the pitch evaluation metrics, his cutter, curveball, and splitter all clock in right around average according to Stuff+, but RV100 favors the four-seamer and hates the curve and split. Comparing the curveball’s xwOBA (.305) to its wOBA (.372) suggests the pitch’s actual outcomes have been a bit unlucky compared to what’s expected based on the batted ball characteristics, which in turn is likely deflating its RV100. Meanwhile the four-seamer and sinker both have better wOBAs when compared to their xwOBAs, suggesting some good luck has swung their way and their RV100s might be a little full of themselves. Luck doesn’t explain the metrics’ diverging opinions on the splitter, suggesting something is amiss with Sands’ execution. Hopefully, this contradiction will untangle itself as we proceed.

In terms of the movement profile broken down in the table below, Sands, like Garrett, mirrors the spin of his breaking ball relative to the four-seamer, sinker, and splitter in an attempt to disguise their true identities until it’s too late for the hitter to react. And concealing those identities is necessary because, as with the other two pitchers, Sands’ pitch characteristics are far more average than overpowering. The furthest he deviates from average is with his extension, but unfortunately he deviates in the wrong direction. His 5.8-foot extension puts Sands roughly six to eight inches below league average. Releasing the ball farther from home plate gives the hitter more of a chance to identify the pitch’s trajectory, which likely explains the lower Stuff+ scores relative to what Garrett and Martin receive for comparable pitches. And while we’re talking pitch trajectory, the extra couple inches of drop on his splitter relative to an average right-handed offering of the pitch might be too much of a good thing; at times it dives too far, too quickly to really tempt hitters.

How the pitches move relative to one another is basically a hybrid of what we’ve seen so far from Garrett and Martin. The fastballs land on the movement plot in roughly the same orientation as the other two, aside from being stretched more vertically. Sands’ curveball operates similarly to Garrett’s sweeper, just with more drop.

Like Martin, Sands doesn’t throw his sinker to lefties, but beyond that omission, Sands attacks hitters in the exact same manner regardless of handedness. He aims to fill up the zone with his four-seamer, works arm side with the cutter and sinker, and keeps the ball down and/or to the glove side with the splitter and curve.

Sands mostly sticks to the standard sequencing playbook, but he’ll reach for any of his non-splitter offerings to begin a plate appearance. If he gets ahead, expect a heavy mix of splitters and curveballs; if he falls behind, expect him to thrown mostly cutters and four-seamers. His adequate command keeps him competitive, since even after falling behind, the average outcomes remain respectable and in line with the more favorable counts.

The swing metrics suggest Sands’ cutter is his best option for inducing weak contact (51% swing rate, 32% hard-hit rate), the four-seamer has the lowest in-zone contact rate (80%) to pair with the second highest in-zone swing rate (71%), and the curveball is best for forcing swings out of the zone (35%) that lead to either a strike (14% swinging-strike rate) or weak contact (25% hard-hit rate).

***

With the four-fastball approach to relief pitching now fully dissected on the lab table before us, I can’t truly say we’ve discovered the next big thing that pitchers everywhere will be rushing to replicate. Though Garrett, Martin, and Sands are the only three relievers doing this out of almost 200, their approach is not as novel as those numbers suggest. What they’re actually doing is leaning on all of the classic pitching fundamentals: changing the hitter’s eye level, attacking the zone to get ahead in the count and then make the hitter chase, varying speeds, varying locations, keeping the hitter off balance. Most relievers execute these fundamentals using one or two overpowering pitches, or in lieu of dominant stuff, they cobble together a few crafty junk pitches. Garrett, Martin, and Sands pitch as if they were junkballers, but instead of throwing knuckleballs or Bugs Bunny changeups, they take their collection of middling fastballs and deploy them as junkballs. They mix and match movement profiles and velocities so hitters can’t sit on certain pitches or locations. They do all the same stuff every pitcher does; they just dress it up a little different. Which in and of itself is novel enough to still be impactful. After all, 10 Things I Hate About You is a singularly great movie, but it’s also a classic Shakespeare play, just dressed up a little differently.