2026-06-03 — Re-read of June 2 with full metrics: the 8i has the move, the 6i doesn't

Follow-up to 2026-06-02-release-delay-irons.md. That note was written without the real bag (reference/bag.json) and without the new pull fields (low point, smash index, quality flags). Re-analyzing the same two sessions with them changes the prescription — for the 6i specifically, the "more shaft lean with the same downward strike" advice was half right.

Sessions: 6i 2026-06-02 17:01 (38 shots, 17 clean) and 8i 2026-06-02 21:03 (43 shots, 18 clean). 8i was deliberate half-/partial swings (CHS 75.6–91.2). Worked on: (re-analysis only — no new range time)

What the data showed

Clean strikes, split by attack angle — the split the old note couldn't see:

• 8i: hitting down works. AoA −3.5°, low point +3.1″ after the ball (in band), DL 29.6° on a real 40° club = 10.4° of deloft, smash index 0.96. The best strikes are the steepest (shot 27: AoA −5.1, low point +4.5″, smash 1.302). Everything in band except launch (22.5°, barely LOW).
• 6i: it inverts. Best strikes are picked — shots 13/14/22/24 at AoA 0 to +3.7° with low point behind the ball, smash 1.29–1.33. The descending attempts (shots 21, 25) kept DL at 29–32° on a 32° club (≈ zero lean): spin loft ballooned to 35–36°, spin 9,100–9,200, smash 0.86–0.93. Aggregate flags: AoA −0.7 HIGH, low point 0.7″ LOW (band 1.1–3.5″), only 5.0° delofted vs the 8i's 10.4°. Smash index 0.91 vs 0.96.
• Path, both clubs: every clean strike in-to-out — 6i +8.7° avg (max +15.1), 8i +7.5°, face 3–4° open to target. Misses: 23/38 (6i) and 27/43 (8i) finished >5 yd right; the 6i covers it with face-to-path −5.9° (push-draw/hook geometry).
• Best-case ceiling: purest strikes smash ~1.30–1.33 (6i) vs 1.40 published, with spin loft in band (~25–29°) → residual is likely off-center contact. Inference only — zero direct impact-location reads in either session (0/38, 0/43). Spin measured directly on just 4/38 and 0/43 shots; spin means are interpolated.

Mechanism

On the 6i, hitting down without the hands leading is a steep glancing blow, not compression — AoA steepens but DL doesn't drop, so SL ≈ DL − AoA explodes. On the 8i the same intent arrives with lean, so down = compression. The release move exists in the swing; it doesn't survive the longer club at full speed. The +8–9° in-to-out path is connected: swinging that far rightward is swinging up, which is what flattens the 6i AoA at 93 mph.

The trend says the speed gain caused it: 6i Feb→Jun CHS +19.5 mph (73→93), carry +39 yd, but AoA −3.4 → −0.7 and low point +3.4″ → +0.7″. The 8i kept ball-first through the same speed gain (DL −6.0°, SL −7.2°, smash +0.083).

Correction to the June 2 note: it read the 6i as needing "the same downward strike, finish ball-first." With low point visible, the 6i's downward strikes are its worst — the fix is not steeper, it's lean: keep the natural −1 to −2° AoA but move low point in front of the ball.

Highest-leverage next focus

6i: forward shaft lean at the natural shallow AoA — DL into 25–26° while still descending, low point +1″ or more. Use the 8i's hands-lead feel at ~80% speed; check ball position (too far forward would encourage the pick). If it works: 6i low point goes positive on descending strikes, SL stays ≤ 32°, spin stops spiking past 9,000, smash on downward strikes converges to the 1.29–1.33 the picked ones already get. Path (+8–9° in-to-out, both clubs) is the next item after contact — it caps start line and feeds the shallow-out pattern.

Trend

• 6i Feb→Jun: CHS +19.5, carry +38.8, AoA −3.4 → −0.7, low point 3.4″ → 0.7″ — speed up, ball-first contact lost.
• 8i Feb→Jun: DL −6.0°, SL −7.2°, smash +0.083, carry +40.0 — release work fully landed.
• Personal CHS bands were stale (calibrated on pooled Feb+Jun data, dominated by Feb at ~15–20 mph slower). Added a --since flag to calibrate.py and recalibrated 6i/7i/8i from June-only sessions (calibrate.py --all --since 2026-06-01, 2026-06-03) — CHS flags are trustworthy again. Caveat: the June 6i band now encodes the fault (AoA −1.9–+0.5, low point −0.4–+1.8″), so judge 6i AoA/low-point against the published targets, not the personal band, until the lean fix lands.