Risk Regime Scorecard

leg_score -0.75

Vote = 200DMA rules (Liq 40% / Cred 35% / Vol 25%). Weighted leg_score = 0.40·liq + 0.35·cred + 0.25·vol (raw on ≥+0.40 / off ≤−0.40; hard VIX veto → risk_off). 1Y/3Y %ile columns are display-only context.

UUP/HYG vs 100DMA and core level 1Y/3Y %iles are display context only — none of these columns vote. Percentiles rank the current UUP/HYG/VIX level over trailing 1Y/3Y observations using the same logic as Yield Curve / Global Volatility. DXY (under UUP Current) and HY OAS / LF98OAS + CCC OAS / BCAUOAS (bps under HYG Current; 1Y/3Y %iles stacked under those columns) are BBG display-only. LQD/MOVE/10Y remain context.

* Volatility has no 200DMA. VIX level subscore: <15 → +1 (bullish), 15–20 → 0 (neutral), ≥20 → −1 (bearish). Combined with the 21d ROC subscore (≤−10% → +1, ≥+15% → −1) to form the Call.

US 2Y / 10Y / 30Y · 2s10s · as-of 2026-10-02

US 2Y / 10Y / 30Y are Bloomberg MASTER (2) PX_LAST (display only — do not vote). Read on yields: Rising if 1M Δ ≥ +15 bp, Falling if 1M Δ ≤ −15 bp, else Stable; 3M is display only. Read (2s10s): Steepening if 1M Δ >0, Flattening if 1M Δ <0, Flat if 1M Δ =0; the Current level is shown separately and does not override the Read. 1Y/3Y %ile = current level percentiles over trailing 1Y/3Y; yield/2s10s Δ in bps. Bond vol (MOVE) is on Global Volatility, not this panel. Source: Bloomberg MASTER (2) USGG2Y/USGG10Y; FRED DGS30 (US 30Y interim); FRED DGS2/DGS10 (2s10s). US 30Y is FRED DGS30 interim until USGG30Y is pasted onto MASTER (2).

VIX · NDX vol · RUT vol · MOVE · CVIX (FX) · as-of 2026-10-02

Context only — does not vote and does not change the Core Volatility leg. 1W / 1M / 3M are percent changes over 5 / 21 / 63 sessions; 1W and 3M are display only. 1Y/3Y %ile are current level percentiles over trailing 1Y/3Y. Global Volatility Reads use the signed 1M change only: >0 Rising, <0 Falling, =0 Flat; there are no panel bands or 3M confirmations. The Macro Weather Volatility light is intentionally unchanged: its legacy MOVE context tint uses the separate weather_read input, while the panel displays the simple 1M MOVE Read. Up = higher vol. Source: Bloomberg MASTER (2) PX_LAST. Core Volatility vote still uses Yahoo ^VIX only.

as-of 2026-10-02

Windows: 5d (1W) / 21d (1M) / 63d (3M). Read uses 1M corr (|r| ≥ 0.15; 3M fallback if 1M missing). 1W is display only. SPY pairs keep Risk-on/off Reads; Bitcoin/Gold/Crude vs UUP are numbers-only (Read —). SPY vs US 10Y uses BBG USGG10Y daily Δ yield (not TLT). Rolling corr is noisy (esp. 1W); context only — not a regime leg.

Now: 86.8% above 200DMA

Reads: Broad / Narrow / Mixed (participation & size; context only). Changes are percentage points. Read uses level and primarily 1M Δ (conflicts = Mixed); thresholds ≥55% / ≤45%. Volume moved to its own section below. 86.8% SPX above 200DMA → Read Broad (level ≥55% or 1M Δ ≥+5 pp = Broad; level ≤45% or 1M Δ ≤−5 pp = Narrow; conflicts = Mixed); RSP/SPY 21d -4.53%, 63d -4.78%; soft-downgraded to neutral (RSP lagging) · computed from 503 S&P constituents (portfolio OHLCV cache); n_valid=38 n_missing=465

A/D Read: Uptrend if line > 50DMA and 50DMA > 200DMA; Broken uptrend if line < 50DMA while 50DMA still > 200DMA; Downtrend if line < 50DMA and 50DMA < 200DMA; Divergence if SPX within ~3% of 52-week high while A/D is below its 50DMA (and/or 1M change negative). Context only — not a vote. Source: TRADCADS Index · as-of 2026-10-02.

SPX Volume Conviction · last vol vs 20d avg 0.92× (lighter) · Overall: Moderate volume conviction · BBG · as-of 2026-10-02

Display only — does not vote. Daily BBG SPX VOLUME vs 20d avg baseline (≥1.0× = high-conviction day). Upside = advancing days; Downside = declining days. Side color by share of ≥1.0× days: ≥70% High-conviction (green), ≤30% Low-conviction (red), otherwise Moderate (yellow). Window Read uses the taxonomy below. Today’s print: both sides light (last 0.92× 20d avg). Prefer BBG over Yahoo.

Taxonomy

• High-conviction both ways (≥70% each side)
• High-conviction advances · Moderate/Low sell-offs
• Moderate/Low advances · High-conviction sell-offs
• Moderate volume conviction (30–70% both sides)
• Low-conviction both ways (≤30% each side)
• Mixed volume conviction

Overall (1M): Offensive growth · soft lean: Risk-on lean

Taxonomy: Offensive growth / Cyclical value / Defensive / Mixed (context only). SPHB vs SPLV Read uses the 1M relative return: positive = High Beta, negative = Low Beta, and |relative return| <0.5% = Mixed. The broader Style taxonomy and Overlay keep their existing Offensive growth / Defensive labels and logic.

Overall (1M): Risk-off · as-of 2026-10-02

Context only — not a voting leg. GS HF VIP (GSTHHVIP) vs Most Short Rolling (GSCBMSAL). Pair = VIP return − Most Short return. Reads from GS basket signals (absolute + spread). Overlay for crowding; combine with Positioning / Style / breadth.

as-of 2026-12-31

Context only — not a voting leg. Revision Read = mean of available 4W/3M % changes in Bloomberg Weighted Est and Next Year EPS (50/50 when both exist). NTM EPS level = Next Year estimate. Growth = NTM EPS / TTM EPS − 1. Nowcasts = Bloomberg Economics GDP nowcasts (US/China/Euro Area); always use the last non-null datapoint in each country column (may be dated past today). EA = Bloomberg Euro Area GDP nowcast (BENWEAGQ).

as-of 2026-10-02

Non-commercial / leveraged net · 3Y %ile. Read from 3Y %ile (80/20): excess long → Contrarian Short, excess short → Contrarian Long; not a vote. Asset-manager nets omitted (noisy scale).

Overall: Neutral · composite %ile 32.8 · as-of 2026-10-02

Read from 3Y %ile (80/20): excess long → Contrarian Short, excess short → Contrarian Long; not a vote.

Headline = 4W weighted raw · Legacy persist = legacy 3-week confirm — not the headline.