Correction — August 17, 2026 · the headline finding is withdrawn
What we published on June 6: out-of-zone swing rate up from 30.0% to 32.3% over a matched calendar window, a +2.3-point jump, framed as hitters chasing more under the robo-zone.
What was wrong: the two seasons were measured with two different rulers. In the 2026 Statcast feed, sz_top/sz_bot are the fixed per-batter ABS challenge box — a single pair of numbers for 593 of 635 batters all season. In 2025 and every season before it, those same columns carry the per-pitch tracked zone. Across 276 regulars the ABS box top sits about 2.6 inches lower than the hitter’s 2025 tracked mean top; 63% fell by two inches or more and none rose by two. So an as-tracked 2025-vs-2026 comparison of any zone-relative rate reclassifies pitches instead of measuring hitters.
Corrected: apply each hitter’s own 2026 ABS box to both seasons, on the 389 hitters who appear in both windows, and chase goes 32.0% → 32.4% — +0.4 points, 95% batter-clustered interval [−0.16, +1.03], which contains zero. About 84% of the number we printed was the ruler moving.
Withdrawn: the +2.3-point headline and the 30.0% baseline; the by-count gradient below; the figure caption’s claim that the increase “excludes zero”; the methodology sentence “per-hitter chase is highly stable year over year (r ≈ 0.76), so the league shift reflects a real behavioral change, not reshuffling,” which was both unsupported and a non sequitur; and the “March–June” window label, which was always March 27 to May 27. What holds: the 2026 level itself, and the arithmetic — re-running the published pipeline on the raw data returns +2.35 points against the +2.3 we printed, and reproduces the published out-of-zone count of 128,057 pitches exactly, the 2026 count to within one. The article below has been rewritten around the corrected numbers rather than left standing.
The same artifact runs through a good deal of this season’s public analysis. We found it while working on Where Cal Raleigh Broke — and Where He Didn’t, which documents it in detail and flagged this correction as coming.
The pitch to a robot umpire was supposed to be simple: with a consistent, accurate zone, a hitter can finally trust his eyes and lay off the borderline junk. Discipline gets easier; chase should fall. We went looking for that and reported the opposite. We were measuring our own ruler. Scored against the same box in both seasons, league chase went 32.0% → 32.4% — +0.4 points, with an interval that contains zero. Nothing detectable happened.
Here is the mechanic, because it is simple enough to check by hand. In the matched window, 4.3% of 2025 pitches (8,794 of them) sat inside the tracked zone but outside the 2026 ABS box — mostly at the letters, where the box is lower. Hitters swung at 61% of them. Score 2025 with the tracked zone and those pitches are strikes taken or fouled off; score it with the box and they become chases. The traffic the other way is a quarter the size (0.8%, swung at 46%). The out-of-zone denominator alone moves from 53.7% of pitches to 57.2% before a single hitter changes anything.
116,549 and 115,360 out-of-zone pitches). The gap is +0.4 points with a batter-clustered 95% interval of [−0.16, +1.03] — indistinguishable from no change. The +2.3 points we printed in June came from scoring 2025 with the tracked zone and 2026 with the box.By the count, nothing moves either
The original piece leaned hardest on the count split: the two-strike row showed the biggest jump, which looked like hitters expanding when the count was on the line. That gradient was the artifact too — the reclassified pitches are concentrated up in the zone, and hitters swing at those a lot in every count. Measured like-for-like, all three rows flatten and all three intervals span zero.
Out-of-zone swing rate by strike count, matched window, each hitter’s 2026 ABS box applied to both seasons. Differences, top to bottom: +0.2, +0.3, +0.9 points, with batter-clustered intervals [−0.52, +0.89], [−0.66, +1.20] and [−0.06, +1.84]. The two-strike row is the largest of the three and is still not separable from zero. As tracked, the same three rows read +1.7, +2.5 and +3.3 — that gradient is the ruler, not the count.
What actually changed — and it points the other way
Once both seasons are on the same box, the interesting question is where the residual sits. Sort every out-of-zone pitch by how far it missed the enforced edge, and the answer is the opposite of the story we told.
- Within an inch of the box edge — the genuinely borderline pitches, the ones a precise zone is supposed to help a hitter take — chase went
55.3%→53.3%— down2.0points, interval[−3.28, −0.58], clear of zero. Hitters got more disciplined exactly where the mechanism lives, and it is the most robust cell in the whole re-analysis. - More than ten inches off the box — pitches nobody has to think about — chase went
9.0%→10.1%,+1.2points, interval[+0.6, +1.8]. Whatever that is, it is not a zone-precision story; it looks like pitch mix or pitcher approach, and it needs its own work before anyone writes it up.
The bands in between move by less than a point in either direction. So the honest summary of the first season of full ABS is narrower and duller than what we ran in June: league chase did not measurably rise, and on the pitches where a fair zone was supposed to make hitters more patient, it looks like it did.
This is not a small-sample rescue, either. Re-run on the season to date (03-27 to 08-15, 499 hitters), the like-for-like gap is +0.29 points, interval [−0.19, +0.78]. Roughly 2.2× the pitches makes the effect smaller, not clearer. The as-tracked version of that same comparison reads +2.6 points — which is what the artifact looks like when it is fed more pitches.
How we got it wrong
Two failures, and only one of them is about baseball.
- We called the ruler by the wrong name. The original methodology said chase was measured “outside the rulebook zone.” It was measured against Statcast’s
sz_top/sz_bot, which is the tracked zone in 2025 and the ABS box in 2026. Those are three different things, and writing the wrong one down is what let a ruler change pass as a behavior change. - We listed the explanation and then walked past it. The June piece named “the zone moved” as one of three unproven candidate mechanisms. It was not a candidate. It was the answer, and it was checkable in an afternoon — count the distinct
sz_topvalues per batter per season.
One stability claim deserves a specific retraction, because it was the sentence doing the most work. We wrote that per-hitter chase is stable year over year (r ≈ 0.76), “so the league shift reflects a real behavioral change, not reshuffling.” Rank stability and a change in the league mean are unrelated quantities — a perfectly stable ranking is compatible with any league-wide shift, including none. The r itself is also mis-stated: 0.76–0.77 is the full-season figure, while on the window this article actually reports it is 0.69. And measured hitter by hitter, the share who chased more falls from 68% as-tracked to 53% like-for-like — a coin flip.
What we said in June that still stands: we made no causal claim about ABS, and we said the number was descriptive pending more of the season. More of the season arrived and took the number with it.
Methodology
Zone definition. Chase = swing rate on pitches outside the zone, where a pitch is in-zone if |plate_x| ≤ 0.83 ft and its height lies between the batter’s zone bottom and top. This is not the rulebook zone: it is Statcast’s sz_top/sz_bot, which reports the per-pitch tracked zone through 2025 and the fixed per-batter ABS challenge box in 2026 (one value for 593 of 635 batters). Every corrected 2025-vs-2026 number on this page — everything outside the quoted, withdrawn June figures — uses a single ruler — the median 2026 ABS box for that batter — applied to both seasons, on the 389 batters who appear in both windows. Swings are swinging_strike, foul, hit_into_play, foul_tip, swinging_strike_blocked, foul_bunt, missed_bunt and bunt_foul_tip; regular season only.
Window and intervals. Both seasons are restricted to the identical calendar window, March 27 to May 27 (03-27 to 05-27), to remove seasonality; the season-to-date check runs 03-27 to 08-15. Confidence intervals are batter-clustered bootstraps (2,000 resamples of the hitter list, both seasons of a hitter travelling together), because pitches within a hitter are not independent — treating them as independent is what makes the headline gap look like it excludes zero when it does not.
Box shift. Across 276 hitters with heavy samples in both seasons, the 2026 box top sits 2.6 inches below the 2025 tracked mean top (median 2.58), the bottom 0.24 inches above it, for a net zone height 2.85 inches shorter. 95% of hitters saw the top edge fall; 0% saw it rise by two inches or more.
Code. The correction was recomputed twice from the raw pitch data with independently written scripts — research/zone-box-audit/chase_correction/ and its verify/ re-derivation — which agree on every figure here to within band-edge rounding. The residual is descriptive; we make no causal claim about ABS in either direction.
Cite this analysis
CalledThird. "Hitters Aren’t Chasing More Under the Robo-Zone." CalledThird.com, June 6, 2026. https://calledthird.com/analysis/abs-chase-rising
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]