The story is so standard it barely registers as a claim. A rookie arrives, tears up the league for three weeks, and then — the announcer explains — the league adjusts. The book gets written. The sliders start coming. The kid has to adjust back.
We tried to film that process and failed. Across the 2023–25 debut classes, aligning every rookie to his own debut date and watching the league’s attack in 30-day bins, the pitch mix thrown at rookie hitters just… doesn’t move over their first four months. No fastball decline, no breaking-ball ramp, nothing that survives a control group of ordinary hitters measured over the same calendar.
So we asked the other question: if the book isn’t visibly being written across his first season, how early is it already good? Very early. A debutant’s first 150 major-league pitches predict the diet he’ll see a full season later at least roughly as well as an established hitter’s opening week predicts his — and whatever the league settles on in those first 150 pitches, most of it is still in place in September.
The opening window is not a blank page
Take a hitter’s first 150 major-league pitches — roughly ten days and forty plate appearances — and ask how much it tells you about the pitches he’ll see in his second season, a year later.
Left: predicting next season, 114 debutants from the 2023–25 classes against 756 veteran seasons (343 hitters) measured the same way. Right: the same question inside 2026, opening window vs the rest of the season on entirely separate pitches.
A debutant’s opening fastball share correlates with his year-two fastball share at +0.43 [+0.30, +0.56], and his opening breaking share at +0.44 [+0.30, +0.57]. Established hitters sit at +0.39 on both. We could not detect a rookie penalty — but that phrasing is doing real work, and the shaded band on the chart is why. The test would have accepted any rookie correlation from about +0.25 to +0.54 as “no penalty,” and at the bottom of that range a rookie’s opening window would carry only about 40% as much information as a veteran’s. This is a failure to find a gap, not a demonstration that none exists.
The 2026 season lets us ask the cleaner within-season version: a debutant’s opening-window fastball deviation predicts the rest of his 2026 at +0.52 [+0.27, +0.73], against +0.50 for established hitters — point estimates that land on top of each other, though with only 32 rookies clearing the 200-pitch post-window threshold (of the 51 who qualified at all) the difference could be anywhere in [−0.25, +0.26]. Read that as “no difference detectable,” not as equality.
There is also a hint that the opening plan carries information about the hitter that his major-league record hadn’t yet supplied: how much breaking stuff a debutant sees in those first 150 pitches predicts how often he swings and misses hundreds of pitches later, on a set of pitches sharing nothing with the first (r = +0.20 [+0.07, +0.33], against a shuffled placebo at p ≈ 0.012). We are deliberately not building on it. Taken one debut class at a time the estimate runs +0.21, +0.04, +0.29; dropping the 2025 class pushes it across zero; it predicts swing-and-miss but not how well he actually hits (against later wOBA, −0.10, interval spanning zero); and established hitters show the same pattern, so it isn’t rookie-specific. Interesting, unfinished, and worth another debut class before anyone leans on it.
A hitter’s welcome, not a rookie’s welcome
Which brings us to this year’s class, and the two Rookie of the Year races currently being argued about.
Top: the five ROY-race figures’ opening books, on the only two metrics a 150-pitch window can measure reliably. Faded markers include the league rate in their interval. Bottom: how much opening books vary from hitter to hitter, debutants vs established hitters.
Kevin McGonigle owns the class’s largest fastball-share deviation: 38.0% fastballs in his first 150 major-league pitches against a calendar-matched league rate of 55.6% — a gap of −17.6 percentage points [−27, −8]. And here is the sentence that has to travel with it: it is not unusual. We went looking for the largest deviation among 51 rookies, and the largest of any 51 hitters is expected to be about 18.4 points by chance alone. Five of 255 established-hitter windows beat him outright. Price in the search that found him and he sits comfortably inside the ordinary — a useful reminder about every “look what the league is doing to this kid” graphic you’ll see this September. Murakami, Okamoto, Wetherholt and Griffin all got ordinary opening books; none of their intervals excludes the league rate on either measurable dimension.
The class-level pattern is the more useful one: opening books vary enormously from hitter to hitter, and for debutants that variation is no wider than for established hitters — on breaking share it is detectably narrower, and on the other measures no difference is resolvable either way. Our reading, and we’ll flag it as a reading: that is what it looks like when there is no rookie script. The league doesn’t appear to run a standard welcome on new arrivals; debutants’ plans look like everyone else’s in shape and spread.
What we could not show
We cannot say the adjustment period doesn’t exist. Comparing rookie year-one-to-year-two attack shifts against veterans matched on where their own first season started, nothing is detectable on any measure. But with 85 rookies, only a shift larger than about half a percentage point would have registered at all. Absence of evidence, and we’re not going to sell it as the other thing. The one apparent exception we found in an earlier pass — a rookie-specific decline in strike-zone pitches — is substantially regression to the mean: rookies simply start their first season further from league average than veterans do, and anyone starting there drifts back. That accounts for nearly all of the parallel effect in heart-of-plate pitches and about half of the zone effect. The remainder is too small to distinguish from zero, which is not the same as being zero.
And we cannot date the plan. The opening window spans up to 29 days of major-league baseball, so “the report was written in a scouting room last winter” and “it came together over his first ten days in the league” both fit this data equally well. Shrinking the predictor window produces a non-monotone mess with overlapping intervals. We have a preference; we don’t have a finding.
And we cannot see everything we’d like to. A 150-pitch window measures pitch type well (split-half reliability around 0.59) and pitch location badly (0.22 for zone share, 0.21 for the heart of the plate, intervals reaching zero). So no debut-window or named-player claim about location appears anywhere in this piece — the location results above come from full-season samples, which are large enough to carry them. Every year-two result also conditions on the rookie still being in the majors in year two, which only 55% of qualifying debutants were, against 67% of veteran seasons.
What survives all that is still worth knowing. The adjustment everyone narrates is not visibly happening across a rookie’s first months, and the plan being run against him is already stable and already informative by his 150th pitch. So when your team’s rookie starts seeing a strange diet early on, the months-long adjustment the broadcast is about to describe is not something we can find in the data.
Methodology
Data, windows, the adversarial process, and what got withdrawn
Data. Pitch-level Statcast, regular season, 2023 through 30 July 2026. Debut cohorts come from the MLB Stats API (position players whose first major-league appearance falls in that season): 109 / 108 / 105 / 86 for 2023–2026. Batter identities are resolved by id, since Statcast’s player_name is the pitcher. All attack measures are calendar-adjusted: weekly league baselines are computed from non-cohort batters and subtracted, so league-wide drift cannot manufacture persistence. Hitters flagged intl_veteran (debut age 26+, the NPB/KBO imports) are carried separately in every headline claim.
Windows. The debut window is a hitter’s first 150 major-league pitches within 29 days of his debut; the veteran comparison window is the first 150 pitches of a season within 29 days of that hitter’s first appearance, so the two arms are matched on window size and attenuation. For the swing-and-miss association mentioned in passing, the outcome window starts at career pitch 301 — the intervening 150 pitches are discarded as a buffer — and the two windows were verified to share zero pitches.
Process. Three analysis modules were built and then attacked by independent reviewers instructed to recompute headline numbers from the raw parquets rather than trust the summaries, followed by a correction round, an integrated audit, and an independent review by a different model family that re-derived every load-bearing number with its own code. Numbers in this article are generated from the sealed artifacts rather than transcribed.
What got withdrawn, for the record. The review process killed several claims we had written down: that the opening plan is aimed at spin (the spin-specific differential is a flat null); that what it tracks is general hitter quality (it predicts later swing-and-miss, but not later wOBA — −0.10, interval spanning zero); that the swing-and-miss association is unique to rookies (established hitters show it at +0.11, difference indistinguishable); that it is carried entirely by true rookies (the 20 imports’ interval contains the true-rookie estimate); that the signal is absent in the first 50 pitches (under-powered, not absent); that the plan “sharpens” as evidence accumulates (an artifact of a reliability estimator); that the rookie zone-share exception was a real adjustment (about half of it is regression to the mean, the rest undetectable); that the opening window predicts the rest of the season “just as strongly” for debutants as veterans (point estimates match, but the difference interval is [−0.25, +0.26]); and that Kevin McGonigle’s opening book was unusual by league standards (it is the maximum of a search over 51 hitters, and milder than the typical such maximum). An earlier draft quoted a lower shuffled-placebo p-value taken from a 2,000-draw estimate; at 200,000 draws it converges to 0.0119, and that larger figure is what we publish.
Limitations. Survivorship bounds every year-two result: 55% of window-qualified debutants reached 500 year-two pitches, against 67% of veteran seasons — the veteran arm is selected too, just somewhat less harshly. No equivalence margin was pre-specified for either indistinguishability test; the margins reported here are post-hoc descriptions of what the design could have detected. The swing-and-miss association pools three debut classes and one leave-one-out interval crosses zero, which is why this piece does not build on it. Split-half reliabilities are upper bounds, since both halves share games, parks and opponents. The 2026 season is partial (through 30 July) and its debut class has no year-two outcome at all — that layer illustrates, it does not fit. Everything here is an observational association about what pitchers throw; no causal mechanism is demonstrated.
Cite this analysis
CalledThird. "The Book Is Good by Pitch 150. We Can't Tell You When It Was Written." CalledThird.com, August 10, 2026. https://calledthird.com/analysis/the-book-is-good-by-pitch-150
All CalledThird analysis is original research. If you reference our findings, data, or charts in your work, please link back to the original article. For data inquiries: [email protected]