Dashboard › Probes › Realness Loop › Iter 30 · #14 GROUNDED

iteration 30 · 2026-07-24 · incrementality axis · terminal · operator ruling (a) applied · laptop-drives-bigblack

📐 #14 Spanning intercept GROUNDED

Operator ruling (a) applied (Nasim directive: complete the campaign). The 6 core gates are robust across 4 seeds; the single marginal gate — power@α=5bps ~0.78 — is grounded on the published power/N-contour envelope: power reaches ≥.8 at α≈6bps (N_min≈150k), the 5bps shortfall being the documented intrinsic long-memory-effective-N floor. Supersedes the iter-29 operator-ruling checkpoint.

6 + 1

core gates + enveloped power

α≈6bps

admitted operating point

4 seeds

robust

16 / 20

terminal (15 G · 1 R)

What grounding (a) means. The spanning intercept (does a feature-as-strategy-return earn α the shipped basis doesn't already earn?) is admitted with the operating point at α≈6bps and the α-power / N-contour envelope published. A feature must clear t(α)≥3 with the frozen NW-lag set + block-SE + a PREREG factor basis. The 5bps shortfall is not a defect of the instrument — it is the long-memory floor (power flat in N; HAC-t scale-invariant; block-bootstrap more conservative; conservative t≥3 admit bar).

The gate battery (final)

Gate (§7 row 14)	Result (4 seeds)
Coverage & null-t	coverage 0.93–0.97	PASS
Null-FPR (spanned feature)	0.0	PASS
Null-FPR across Hurst/t(3)/block	HAC & block-bootstrap both ≤.05	PASS
Substitution — B spanned collapses	\\|t\\| <1	PASS
Frozen-NW-lag stability	spread ≤0.23 (no lag cherry-picking)	PASS
Omitted-premium Harden	complete-\\|t\\|<3 · omit-premium t≥7.8 spurious	PASS
Power @ α=5bps → envelope	population ~0.78 · reaches 0.8 at α≈6bps	ENVELOPE

Precedent for an enveloped-operating-point graduation: #0 (PBO descope), #4 (lag-tension resolved by a pattern), #12 (rejected). The instrument is sound (6/7 robust incl. the Harden); per recommendation (a) it graduates with the published envelope rather than being held at a bar the long-memory data cannot reach.