Cracking the Kodai to Senga’s Success

A couple of weeks ago, Hannah Keyser of The Bandwagon asked Kodai Senga about his reverse splits when it comes to times through the order. His .347 wOBA allowed the first time through the lineup drops to .275 the second time through it, and then sinks to .205 the third time through. However, his total batters faced sits at just 90 for each of the first two times through the order and 48 for the third time through. That’s a small enough sample that it’s difficult to say how much of the credit goes to Senga and how much is naturally occurring variation.

Either way it doesn’t matter much to Senga, who cares more about process than results, “Quite honestly, I don’t feel like I’m very good at any point during the game this year,” he told Keyser through his interpreter. Harsh words for a pitcher with an ERA- of 37. He does acknowledge that he makes adjustments throughout the game to get a better feel for his pitches and tweaks his strategy based on how each pitch is looking in a given start, which offers some explanation for the improved outcomes as the game goes on. But according to Senga, all of his pitches feel off this year, as he remains in the process of re-establishing his mechanics following the shoulder and calf injuries that sidelined him for nearly all of last year.

Senga’s self-evaluation reminds us that folks performing at the elite level of their fields don’t get there by letting good enough be good enough, but rather by insisting that nothing is ever good enough. To the layperson, it’s baffling to hear that Senga feels he’s pitching worse, despite posting better results than in 2023, his first MLB season, and the last in which he was fully healthy. Add on that he somehow manages to gain effectiveness throughout his starts despite a certain level of displeasure with all of his pitches, and it couldn’t be any clearer that regardless of what you read in Us Weekly, the stars are not actually just like us.

But in refusing to leave himself a five-star review, Senga sets up at least two brain teasers for us laypeople to puzzle through. In describing the need to get back to his pre-injury mechanics, or at least that level of confidence in his mechanics, he told Keyser, “[I]f you dig a little bit deeper, on the analytics of my pitches and the pitching itself, there’s a lot that I can improve on. And I expect myself to be a lot better.” What detectable differences are there between Senga in 2023 and Senga this season that would make him feel he still has room for improvement, even though hitters have been less successful against him this year? And then the second mystery is identifying the in-game adaptations that allow Senga to level-up his outcomes as an outing progresses, but more on that later.

Since Senga clearly feels there’s something off with his mechanics, let’s start there. His arm angle is a few degrees lower, dropping from 48 degrees to 45. Perhaps this developed in response to the shoulder injury, either as compensation or to shield against future injury in some small way. It’s not a large difference, but it could impact Senga’s wrist position when releasing the ball. Several of Senga’s offerings derive some of their movement from seam-shifted wake, which happens when the ball’s seams are oriented in such a way that creates a “wake” of additional force as the ball moves through the air, impacting its trajectory. When a pitch is influenced by seam-shifted wake, its observed spin axis differs from what we’d normally expect from a pitch with the same velocity and movement. Of Senga’s offerings, the largest benefactors of seam-shifted wake are his forkball (or ghost fork, as it’s popularly known), cutter, sweeper, and slider. What’s new in 2025 is the amount of seam-shifted wake. The observed spin axis is deviating farther from expectation and subtly altering the movement profile of these pitches. The nature of the change suggests that rather than primarily generating backspin by staying largely behind the ball, Senga is instead getting slightly inside the ball and generating more arm-side movement. Whether or not this is a side effect of a slightly lower slot, it does seem that his leverage on the baseball as he releases it is distributed differently.

But why would extra movement be a bad thing? Isn’t more movement typically the goal? Especially the less conventional movement generated by seam-shifted wake? More often than not, the answer would be yes to all of the above, but it’s ultimately context dependent. Senga’s primary breaking pitch is his forkball. Though forkballs use the same grip as a splitter and produce a similar downward-diving shape, the rising popularity of splitters and splinkers has not brought with it a wave of new forkballers, likely because the forkball requires mastery of a distinctive wrist action. Dave Stewart, known for reviving his career by tuning up the forkball taught to him by Sandy Koufax years prior, says that to throw a good forkball requires a loose wrist and deception. When asked about Senga’s forkball back in January 2023 as the Japanese pitcher prepared to make the transition from NPB, Stewart told The New York Times, “Everything has to look the same. The fastball has to look the same, and his does. He’s got great deception. And then he’s got tremendous downward movement, which is what makes the pitch even more successful.” So if the goal is to make the forkball look just like the fastball, ideally right up until the hitter has to make a swing decision, then added movement, particularly movement that deviates from the fastball too early in pitch’s trajectory, could hurt the effectiveness of both pitches. Senga throws a rising four-seamer that averages 8.0 inches of arm-side run. In 2023, Senga’s forkball averaged 7.2 inches of arm-side movement, but this year it’s up to 9.9 inches, which might be making it more identifiable to hitters.

And then there’s Senga’s cutter, which starts out looking like his four-seamer and then … stays looking like his four-seamer. Which is not exactly the desired result. There’s just five mph of separation in velocity between the two fastballs, and both pitches start with a similar spin axis, but then ideally, the cutter would veer off and forge its own path with an assist from seam-shifted wake. But while the 2023 version of the cutter managed 0.6 inches of glove-side movement, this year the pitch is leaking arm side and muddying its identity compared to the four-seamer. And while some amount of similarity enhances deception, too much similarity turns an otherwise useful cutter into a secondary fastball that’s easier to hit because it’s slower and moves less. In 2023, Senga’s cutter was his best pitch by run value and worked as a pitch he could throw in the zone to generate whiffs or weak contact, but this year, hitters are swinging and missing less at it and doing more damage on contact. With neither the forkball, nor the cutter playing off the primary fastball as intended, Senga’s four-seamer is also less effective, allowing a .444 wOBA this season, compared to .338 in 2023.

But if his pitches no longer work and play well together, how is Senga managing such stellar results? One reason is that even though Senga’s pitches are not in perfect harmony with one another, they’re all still pretty good pitches on their own merits. Especially the ghost fork. It has its own special name for a reason. He’s also improved his control of said good pitches. While stuff remains the biggest mover and shaker in the out-getting formula for most pitchers, knowing where it’s going can paper over a certain amount of diminished stuff. And with the relationship between his four-seamer and ghost fork on the rocks, Senga has introduced a mediator to act as a go between. Which is to say, he throws a sinker now. The sinker gives hitters another offering that spins similarly to both his four-seamer and forkball out of his hand, but by the time it reaches the plate, it splits the difference between the two in terms of vertical drop and darts even farther toward the third base side of the plate than any of his other offerings. It’s one more chaotic little wrinkle for a batter to consider in the split-second allotted for deciding whether or not to swing.

That brings us to Senga’s reverse times-through-the-order splits, because getting hitters to take bad swings at misjudged pitches is a critical part of his strategy. In terms of pitch usage, his approach leans fairly conventional — fastball heavy in the early going, then working in more and more bendy, offspeed stuff as the game progresses. But with Senga, it’s not just about the shock factor of a new offering, it’s also about adding complexity, introducing new variables for the hitter to consider between when the pitch leaves his hand and the moment when a swing should start. Y’know, that big sprawling expanse of time, in which hitters get to carefully consider the clear and complete facts of the situation and make a well-reasoned plan of action regarding a potential swing. Oh wait, I got that backward. Hitters have perhaps the least optimal circumstances under which to make a decision. They rely on some combination of instinct and pattern recognition, but with Senga, pattern recognition is difficult because he refuses to provide anything resembling a pattern.

So yes, the first time a batter faces Senga they’ll see a lot of four-seamers, thrown to all quadrants of the zone. But after that, the four-seamer becomes primarily a tool to change the hitter’s eye level. As his other offerings dive downward, he has to keep the memory of a fastball at the top of the zone alive as a threat to keep hitters off those other pitches. And as he lays off four-seamers aimed at the bottom half of the zone, he also throws fewer of them in 0-0 counts, refusing to continue his own pattern from prior plate appearances and refusing to follow the established patterns around pitch sequencing. Instead, he opts for more 0-0 cutters and sweepers. After that, hitters can expect a higher rate of forkballs, and while there is such a thing as a fastball count, forkball counts don’t exist, and even if they did, Senga would probably pay them no mind.

We can see how all of this works in practice by walking through Shohei Ohtani’s at-bats against Senga last weekend. Ohtani leads off the game by taking an outside cutter for a ball, then sending a four-seamer to the upper deck in right field. Imagine Senga is attempting a joke that requires an elaborate setup, but just as he’s getting started, Ohtani jumps in and spoils the punchline. Before Senga can show him a ghost fork or slider or plant even a single seed of deception, Ohtani has decided he’s seen enough. When he comes to the plate in the third inning, he takes a similar approach and unleashes his home run swing again, clearly hoping for another four-seamer, but this time Senga is able to land the joke as a cutter on the inner half.

With Ohtani off balance, Senga drops a forkball in the zone for a called strike two, then serves up the four-seamer Ohtani was hunting in the first place. But after over-committing on the cutter, and then seeing a ghost fork at 83 mph, Ohtani’s cut on the four-seamer is late, earning Senga a three-pitch strikeout.

In their third and final meeting, Senga goes straight to the forkball. Ohtani winds around and starts his swing, clearly still hoping for a first-pitch fastball but perhaps remembering what happened last time, and instead holds up and takes the pitch for a ball just off the outer edge of the plate. Comparatively, he seems far less interested in the cutter that Senga places at the bottom of the zone for a called strike. Senga then spikes a forkball for ball two, so his next pitch should probably be a four-seamer to get things back on track. Both Senga and Ohtani know this, and moreover, Senga knows that Ohtani knows. Giving Ohtani the exact pitch he’s looking for isn’t wise, but neither is falling behind 3-1. So Senga throws just his fourth sinker of the game, a pitch that looks just enough like a four-seamer to fool a hitter under the assumption that a four-seamer is coming, but with enough drop to duck under a bat. Ohtani rolls over the pitch, sending it foul up the first base line and evening the count. On the next pitch Ohtani finally gets a four-seamer in the top half of the zone, but his swing is late and he fouls it back. Senga’s ploy to disrupt his timing has worked. He goes back to the forkball on the next pitch and Ohtani’s defensive swing with two strikes is early. He gets under the ball and skies it to right field, where Tyrone Taylor makes the easy catch.

It’s fair to note that Ohtani’s day at the plate is just one example, and that even though Senga has an ERA of 1.19 and 0.00 on his second and third times through the order, respectively, his FIP and xFIP are less flattering (second time through: 2.23 FIP, 3.49 xFIP; third time through: 3.69 FIP, 5.32 xFIP). Is Senga the beneficiary of some good batted-ball luck and fortunate sequencing? Almost certainly. And since FIP strips out batted-ball luck, it’s easy to see why any FIP-based metric is going to cast a side-eye at a pitcher with a .138 BABIP, which is Senga’s number when facing batters for a third time in a game.

His xFIP numbers are even more glaring because xFIP replaces actual home runs allowed with an estimation that takes the league average home run per fly ball rate and multiplies it by the number of fly balls allowed by the pitcher in question. This season, 11.0% of the league’s fly balls have gone for home runs, while for Senga that number is 3.2% overall, but drops to 0.0% on his second and third times through the lineup. Is allowing zero home runs on fly balls a sustainable standard over a full season? Of course not. But it feels almost as absurd to say that Senga doesn’t have any influence on such an extreme outlier performance (only Tyler Mahle has a lower overall HR/FB among pitchers with a minimum of 50 IP).

But even though FIP does tend to be a better indicator of true talent than ERA overall, there are individual exceptions. Some pitchers routinely post ERAs that outpace their FIP. That was the case for Senga in 2023 and in four of his five NPB seasons. When FIP was first developed, we had to do a little hand-waving to infer why certain pitchers had more reliably good outcomes on batted balls, but now we have far more data, covering not only what happened, but how it happened. More specifically, the latest Statcast swing metrics allow us to quantify the impact of Senga’s sequencing stratagem as he attempts to wear down his opponents and induce more and more sub-optimal swings as the game progresses. Between bat speed, attack angle, and attack direction, we can get a decent feel for how well hitters are timing pitches. Tracking these metrics by pitch type and time through the order, we can see that hitters are more likely to be late on Senga’s four-seamer and early on his forkball during the second and third time they face him in a game. With the four-seamer, bat speeds trend down, attack angle shows the point of contact with the ball happening while the hitter is still bringing the bat down into the zone, and attack direction shows an increasing tendency toward sending the ball the other way. Meanwhile, the forkball swings show those numbers moving in the opposite direction — bats speed up, contact is made during the latter half of the swing as the bat slices upward, and balls are pulled more often.

With more research, we’ll likely be able to draw more precise conclusions from the interactions between swing metrics and how they relate to batted ball outcomes, but for now, they at least provide some compelling evidence that Senga has some sway over a BABIP that seems to enjoy some times-through-the-order perks as opposed to the typical penalty. And maybe with more data, Senga will be able to convince himself that he’s not pitching so poorly after all.