Asked whether we can predict that a feature will still be independent tomorrow. Came back with a measurable forecast skill, a working error-bar fix, and the hard finding that no feature is independent regardless of market weather.
| Quantity | Value | Meaning |
|---|---|---|
| Full grid | 384 of 384 cells, 112.2 min, 0 failures, 251 MB RSS | cheap and safe to re-run |
| Feature-cell evaluations | 3,344 crypto + 24,122 forex | scale of the measurement |
| Next-regime persistence | 95.1% crypto / 93.6% forex over 19,196 transitions | features independent today mostly stay so |
| Who-flips forecast skill | conditional AUC 0.945 crypto / 0.880 forex | ranked by today's redundancy margin |
| Regime-invariant features | 0 of 91 crypto (0 of 92 on the re-shoot panel) | no feature is unconditionally stable |
| Forex coverage after the calibration fix | 89.9% / 79.9% against 90% / 80% promised (was 83.4% / 70.4%) | error bars repaired to within 0.1% |
| Firing | What it did |
|---|---|
| 1 | Tested on 20 cells whether the candidate measures discriminate at all; 2 of 6 survived |
| 2 | Launched the full 384-cell replication across every symbol and threshold |
| 9 | 384 of 384 cells done; the surviving measures replicated inside their valid scopes |
| 10 | Passed a blocking zero-leakage gate, then measured actual next-regime forecast skill |
| 13 | Re-shoot of the whole grid on complete data |
| 14 | Everything re-graded on complete data; the forex error bars now keep their promise |
Named open defects: forex ICP wiring emits 0 environments (blocks granger_echo and ICP on forex); the wold_R crypto inversion (0.278) replicates and must be understood before use as a crypto gate; forex PBO 0.20 to monitor; drop or redesign stab_freq and bootci_w. Standing rule: any 'orthogonal tomorrow' claim must be regime-conditional and paired with a live alarm.
manifest status 'E4 COMPLETE', iteration 14 titled 'FINAL: E4 verdict — skill, calibration, stamps all CONFIRMED'.
Lifecycle status is process state, not a judgement of the findings. Results are stated as numbers with their uncertainty. Generated between CLOSING-RECONCILIATION markers — regenerate rather than edit.
Can the forward matrices' output scales support declaring a feature orthogonal tomorrow, empirically?
| # | Report | Status |
|---|---|---|
| 1 | Tier-A aggregate: metric-scale verdict | CONDITIONAL GO |
| 2 | Full-grid run progress | RUNNING 255/384 |
| 3 | Heartbeat: 280/384, EURUSD complete | RUNNING 280/384 |
| 4 | Heartbeat: 296/384, GBPUSD in progress | RUNNING 296/384 |
| 5 | Heartbeat: 313/384, GBPUSD final threshold | RUNNING 313/384 |
| 6 | Heartbeat: 332/384, GBPUSD complete, XAGUSD underway | RUNNING 332/384 |
| 7 | Heartbeat: 347/384, XAGUSD@25, final stretch | RUNNING 347/384 |
| 8 | Heartbeat: 364/384, XAUUSD final symbol | RUNNING 364/384 |
| 9 | FINAL: full-grid scale verdict | CONDITIONAL |
| 10 | E1-E3: leakage-gated walk-forward — forecast skill grounded | SKILL GROUNDED |
| 11 | E4 heartbeat: 165/384, crypto re-shoot done | RUNNING 165/384 |
| 12 | E4 heartbeat: 271/384, EURUSD sweep nearly done | RUNNING 271/384 |
| 13 | E4 heartbeat: 317/384, EURUSD complete, GBP/XAG closing | RUNNING 317/384 |
| 14 | FINAL: E4 verdict — skill, calibration, stamps all CONFIRMED | CONFIRMED |