Framework handover — v7

Generated 2026-10-01T15:08:51+00:00 (content is versioned in the repo, not market-driven)

Framework handover — 2026-10-01. v7 is the live framework (monitor: /monitor-v5/, frozen at first live deploy). v1–v6 are archived below as historical records; their pages stay up, untouched. Track A (/expected-returns/) is unaffected — strategic posture, quarterly cadence, firewall intact (hard rule 7).

v7 methodology note

The correction. v1–v6 evaluated signals on P(day is within 252 sessions before a known peak | gate hot) — the “pre-peak window” metric. It rewards signals that fire during late-cycle rallies with +10–15% remaining upside, precisely when a defensive portfolio underperforms. v7 measures P(≥10% drawdown from signal-day close within 6 months | fire-run event): if I reduce risk today, will the market be lower 6 months from now?
Fire-run level. Each contiguous raw-fire event = one test, so the 45-session latch persistence cannot distort the statistics. A symmetric ±3-month window separates leading signals from lagging confirmation.
Evidence base. /backtest-v7/ (evaluation frame 2000-08-15 → 2026-04-07, pre-computed, rendered verbatim). 6-month random base rate: 13.8%.

v7 live triggers (frozen 2026-10-01)

Defensive Level 1 — ELEVATED: ≥2 of 4 raw signals (no latch): B OAS +50 bps / 20 sessions · SPHB/SPLV < 50-day SMA · T10Y2Y ≥ 12-month min + 0.25 · VIX > 1.5 × 20-day SMA. Expected ~3 triggers/yr, 6m drawdown rate ~35–37%.
Defensive Level 2 — MAXIMUM: B OAS level ≥ 450 bps (sustained). Expected ~1.6 triggers/yr, 6m drawdown rate ~35%.
Confirmation (2A/2B escalation only): existing B+65 velocity and CCC+200 legs (15-session latches) plus the in-drawdown window logic; 2A/2B exits, the Fed-rollover flag and TAIL_OVERRIDE are unchanged from v5.
Removed vs v1–v6: the G3 yield-exhaustion gate (22.7% 6m drawdown rate — worse than random) and the pre-peak-window metric language.
Live data basis: the credit legs read FRED ICE BofA series daily (BAMLH0A2HYB / BAMLH0A3HYC) with the Bloomberg-calibrated thresholds applied as-is; the ~8 bps (B) / ~139 bps (CCC) basis gap is disclosed on the monitor page. Level-1 leg 1 deploys at B+50 by owner decision (the evidence head-to-head recommended B+40).

Track archive — v1–v6 historical record

Track	Page	Role	Status after the v7 handover
v1	dashboard · backtest	Production early-warning track	Superseded as the decision framework. Pages still regenerate daily (immutable track) as a historical record.
v2	backtest-v2	Research backtest	Historical record.
v3	backtest-v3	Research backtest	Historical record.
v4	monitor-v4	Research monitor	Historical record.
v5	backtest-v5	Research ruleset (frozen 2026-09-21)	Superseded by v7 on the live monitor 2026-10-01. The frozen backtest page remains the ruleset record; the v5 frame still supplies the calendar, exit inputs and TAIL_OVERRIDE plumbing under the v7 layer.
v6	backtest-v6	Credit-gate scoping study (no live triggers)	Research record. Its verdict — B OAS fails the arming screen, passes confirmation decisively — is absorbed into v7 (B+65/CCC+200 are the v7 confirmation legs).

The regime-analog and 1970s stress pages (/analog/, /backtest-70s/) remain reference material, not versioned trigger tracks.