House Delta — 2026-08-04

desk: house-delta date: 2026-08-04 forecasts: delta: p_trade: 0.20 direction: long conviction: low

Desk Plan — House Delta — 2026-08-04

Briefing reference: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-08-04-pm.md PM: DELTA

Shared Market Read

• Event risk today: MEDIUM — no Tier-1 (NFP Friday Aug 7 is 3 sessions out; build-up begins in earnest). Two items are pair-specific for this desk: JOLTS at 10:00 AM ET (MEDIUM impact, consensus 7.4M vs prior 7.6M — the first labor read before NFP) — and it lands on DELTA's exact entry bar — and AMD reports after the close tonight (5:00 PM call), the anchor name of the AMD→TSM leg, making today an event-day session for that pair. The after-close cluster (AMD, SPCX) is the bigger single-name risk; semis are expected muted intraday ahead of it.
• Session character expected: Trending-to-mixed — the tape has momentum (Nasdaq futures +0.81%, Dow +0.51%, S&P +0.20%; three straight up days off the July 29 FOMC rout, QQQ reclaimed its 20-day and led Monday at +1.77% vs SPY +1.46%), but the BMO earnings slate (CAT/MRK/MCD/PFE), the 10:00 JOLTS, and the after-close AMD/SPCX event risk argue for churn rather than a clean straight-line day. Iran whipsaw (Trump threatened fresh strikes; crude +2% overnight after Monday's -5%) is a sector-level risk, not a broad-market regime change. Briefing's own read: expect possible fade-and-recover structure if the 10:00 print breaks the initial impulse.
• VIX regime: 15.59 (LOW, <18; -0.40 vs prior close 15.99) — DELTA's best backtest regime (dev PF 2.59, and the strategy performs better in VIX < 20 than in 20–30). Fear fully unwound, no VIX barrier. This is the single most supportive input today, unchanged from yesterday.
• Key levels: SPY prior close 757.72 (above SMA20 745.98), prior range 749.21–758.58. QQQ prior close 700.10 (above SMA20 699.86 — reclaimed), prior range 686.02–701.55, pre-market print thin. The operative level for DELTA is the open gap itself: futures +0.81% (Nasdaq) is modest — below the ~1.5% absorption threshold — so the measurement window should be live; a gap >1.5% at the open would absorb the directional impulse before the 09:30–09:55 window.
• Macro backdrop: Risk-on AI-capex momentum (Monday: de-escalation relief, megacap tech +5-6%, $1.77T added tech market cap over three sessions) layered over renewed Iran uncertainty and the week's heaviest earnings cluster, with Friday's NFP 3 sessions out. Semis are the one soft spot in an otherwise strong tape — explicitly muted ahead of AMD's after-close print (AMD -1.90% Monday close; memory-chip names under pressure).

DELTA — NVDA→ASML, AMD→TSM (LONG-only, Phase 3)

Forecast: p_trade 0.20 | direction long | conviction low

Verdict: setup possible, not probable — a live but degraded LONG day. The NVDA→ASML leg is the only clean expression; the AMD→TSM leg is contaminated by construction (its anchor name reports tonight). Base case remains no trade; trade only if NVDA extends the risk-on tape a second day. Standard sizing, no change.

Context from yesterday (08-03): the first LONG in three weeks fired and won — NVDA +2.32% first-hour → ASML LONG at the 10:00 open (1625.40), hard flat 15:30 at 1645.92, +$184.68. The LONG book is now n=6 (W3/L3, SPRT CONTINUE, LLR +0.299) and the shadow book of disabled AMD→TSM SHORTs stands at n=5. The risk-on regime flip is real: three straight up days, QQQ above its 20-day, VIX LOW. Yesterday proved the enabled direction can fire in this tape — that is the case for a trade today. But today is a materially harder second day.

Why the base case is no trade, in order of weight:

1. AMD event-day contamination (the big one). AMD reports after close tonight — ~54x forward earnings, "an ordinary quarter won't do," with the semis complex muted into the print. Any first-hour AMD move today is positioning into a binary catalyst, not a clean fundamental re-rating — so the lead-lag read-through into TSM is noise, not information. The mechanism's edge on AMD→TSM depends on AMD's first-hour move being informative about TSM; on earnings day that premise is broken by construction. And the "strongest leader wins" rule means if AMD posts the larger |return|, the slot goes to exactly the contaminated pair. An AMD-driven signal today gets extra scrutiny by design — this is session-character judgment, not an invented filter (AMD earnings is not in the FOMC/CPI/NFP/PCE hard-skip set, and I do not unilaterally extend that set).
2. JOLTS at the entry bar. Same collision as yesterday's ISM — which did not hurt the trade (the 08-03 LONG won with the 10:00 print live). So the collision alone is not disqualifying; it is entry-quality risk. But JOLTS is the first labor read before Friday's NFP: a hot surprise (meaningfully above 7.4M) re-ignites the hawkish narrative and rotates risk-off exactly at the 10:00 fill. Consensus is a cooling print — near/below consensus is neutral-to-supportive.
3. Semis soft into the AMD print. The briefing is explicit: semis are "the exception, likely muted intraday ahead of tonight's AMD print." A muted complex rarely produces a +1.5% first-hour leader move. The NVDA path requires NVDA to extend day two of a +5-6% relief move — possible (yesterday it extended day one), but continuation is harder than initiation.
4. Calibration, not optimism (07-31 lesson still applies). p_trade is tradable-signal probability. Qualifying signals since 07-15 were 5-of-5 SHORTs before Monday's NVDA LONG; P(LONG | signal) is only now showing life. With P(signal) ~0.45 (chips have been first-hour-volatile) and P(LONG | signal) ~0.40-0.45 (risk-on tape, but semis muted + one pair contaminated), the honest tradable-signal number is ~0.20 — below yesterday's 0.30, because one of the two pair paths to a clean signal is closed today and the tape is a day older.

Sizing vs. event risk: standard 0.75%, single position — no change. MEDIUM-impact events don't trigger sizing changes (established line, and yesterday's MEDIUM-impact day at standard sizing won). The traded laggards (ASML/TSM) do not report tonight and the position hard-flats at 15:30 — the event risk is signal quality (contaminated leader move), not overnight gap risk on the traded name. No size cut for a quality concern the signal-selection rules already handle by favoring the clean pair.

Key levels to respect: open gap (≤1.5% SPY/QQQ = window live; >1.5% = impulse pre-absorbed, stand down); NVDA 09:30–09:55 trajectory is the actual signal input (needs +1.5% at the 09:55 bar); QQQ holding above ~700 (prior close / SMA20 699.86) matters for the two-tier thesis — QQQ led Monday and must not fade.

Invalidation (stand down entirely):

• SPY or QQQ gaps >1.5% at the open
• Hot JOLTS at 10:00 (print meaningfully above 7.4M consensus) → risk-off rotation at the entry bar; close the session
• NVDA and AMD both flat or fading by 09:45 — no directional conviction, and semis-muted-into-AMD confirmed
• QQQ turns negative in the first 30 minutes — the two-tier divergence collapsing in the wrong direction for LONG

Information cost (documented, not acted on): if AMD fades pre-earnings (the three-week norm), an AMD→TSM SHORT clears the gate and is logged to the shadow book (n=5 → n=6) — zero LONG accumulation on that leg. The LONG-only constraint keeps costing information in this regime; that is the documented cost, the shadow book is the accounting, and the 30-LONG-trade PF ≥ 1.5 gate remains the governing re-enablement path. No re-enablement is recommended — TEMPER's 2026-07-15 rejection stands, and the shadow book accumulates toward the TEMPER conversation flagged in learnings.md 2026-07-24, not toward unilateral action.

Plan Filed

• Filed: 2026-08-04 07:35 ET
• Frontmatter forecasts complete for every active member: yes
• Hard skip check: CLEAR (no FOMC/CPI/NFP/PCE today; NFP Fri Aug 7 is 3 sessions out; VIX 15.59 < 30) — firm-wide GO

No trades taken.