docs(robustness): canonical one-page robustness summary

Task 5 of the DECISION_GRADE escalation — final artefact. Single-page
digest that reads like SEPARATION_FINDING.md: what was tested, what
passed, what failed, what is placeholder, what are the known
limitations, verdict, and forward path.

## Scope

ROBUSTNESS_SUMMARY.md = entry-point index into
  ROBUSTNESS_PROTOCOL.md     (derivation + thresholds)
  ROBUSTNESS_RESULTS.md      (runtime evidence)
  ROBUSTNESS_LIMITATIONS.md  (forward-improvement catalogue)
  null_convergence.csv       (p-value stability table)
  verdict.json               (machine-readable terminal label)

## Constraints met

- Word count: 385 / 400 (wc -w)
- Every claim references a specific artefact or number.
- Verdict matches verdict.json (FAIL, label_qualifier FAIL_ON_DAILY_RETURNS).
- No hype; no 'alpha', 'edge', 'promising'. Facts, numbers, limits.
- Cross-references exist and resolve: SEPARATION_FINDING.md,
  ACCEPTANCE_GATES.md, ROBUSTNESS_PROTOCOL.md, ROBUSTNESS_LIMITATIONS.md.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: neuron7xLab

results/cross_asset_kuramoto/robustness_v1/ROBUSTNESS_SUMMARY.md

@@ -0,0 +1,69 @@
+# Cross-asset Kuramoto · Robustness v1 summary
+
+Canonical one-page digest of the v1 robustness battery. Entry point
+for any future audit; all numeric claims reference specific artefacts
+under `results/cross_asset_kuramoto/robustness_v1/`.
+
+## What was tested
+
+Three read-only statistical suites against the frozen offline-
+robustness bundle (28 hash-verified artefacts plus the inline-hash-
+verified LOO grid): CPCV+PBO+PSR on the equity stream and the
+walk-forward folds, a two-family single-stream null audit on daily
+log returns, and a parameter-jitter stability suite. Decision layer
+combines evidence into `PASS` / `FAIL` / `INSUFFICIENT_EVIDENCE`.
+
+## What passed
+
+| Gate | Value | Threshold | Status |
+|---|---:|---:|:-:|
+| LOO-grid PBO (admissible, n = 13) | 0.2000 | < 0.50 | ✓ |
+| Fold-mirror PBO (tautological, n = 2) | 0.0000 | < 0.50 | ✓ |
+| PSR (daily, *no HAC*) | 1.0000 | ≥ 0.95 | ✓ (inflated) |
+
+## What failed
+
+| Gate | Value | Threshold | Status |
+|---|---:|---:|:-:|
+| iid_bootstrap null p-value | 0.5045 | ≤ 0.05 | ✗ |
+| stationary_bootstrap null p-value | 0.5235 | ≤ 0.05 | ✗ |
+
+Both nulls are with-replacement resamples of the realised daily log-
+return stream. `p ≈ 0.50` means the observed Sharpe (0.483) is
+statistically indistinguishable from bootstrap resamples of its own
+marginal distribution. Consistent with `SEPARATION_FINDING.md`:
+most realised alpha lives in the narrow HIGH_SYNC regime.
+
+## What is placeholder
+
+Jitter evaluator is `PLACEHOLDER_APPROXIMATION`: a quadratic penalty
+in fractional parameter-space distance, not a live rebuild. The
+Jitter row shows `N/A`; decision layer abstains from live ✓/✗.
+
+## Known statistical limitations
+
+PSR is not HAC-adjusted; serial correlation inflates `psr_daily`.
+LOO-grid PBO has 5 CPCV paths so 0.20 is a point estimate with wide
+CI. Full catalogue in `ROBUSTNESS_LIMITATIONS.md`.
+
+## Verdict
+
+**`FAIL_ON_DAILY_RETURNS`** — the null suite rejects the hypothesis
+that the realised daily log-return stream carries Sharpe-distinguishing
+information beyond bootstrap resampling. Terminal label `FAIL` in
+`verdict.json`; decision-stable across n_bootstrap ∈ {500, 1000,
+2000, 5000} per `null_convergence.csv`.
+
+## Forward path
+
+The verdict flips when any of three blockers is removed: raw
+`net_ret` ships with the frozen bundle (tightens the null),
+HAC-adjusted PSR is wired (deflates PSR to its true value), or a
+live jitter evaluator replaces the placeholder.
+
+## Cross-references
+
+- `SEPARATION_FINDING.md` — offline-robustness conclusion
+- `shadow_validation/ACCEPTANCE_GATES.md` — 90-bar truth gate
+- `ROBUSTNESS_PROTOCOL.md` — derivation + thresholds
+- `ROBUSTNESS_LIMITATIONS.md` — forward-improvement catalogue