desk: house-delta date: 2026-08-11 forecasts: delta: p_trade: 0.20 direction: long conviction: low
Briefing reference: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-08-11-pm.md PM: DELTA
DATA-WARNING (stated per dispatch): today's scanner watchlist was generated at 10:16 ET — after the open (the 4:00 AM PT run fired late); its 'pre-market' figures reflect early-session tape. The pm briefing (Alpaca/IEX-verified) is authoritative for prices, but it carries no NVDA/ASML/AMD/TSM levels, and the ASSAY cache holds no 08-11 intraday data yet (last bars 08-10 evening). I anchored Monday closes from the cache and attempted current-session verification via web: prints conflict across sources (e.g. NVDA premarket 220.75/+1.47% per one feed vs. divergent figures elsewhere) and are not Alpaca/IEX-verified — I do not treat any 08-11 print for these four names as authoritative. The runner measures the actual 09:30–09:55 first-hour return; the forecast below stands as the day's forecast for scoring.
Late-dispatch note: filed 10:37 ET — the 09:30–09:55 measurement window and the 10:00 ET entry have elapsed at filing. The open printed small/flat (no gap invalidation). Per the established late-dispatch convention, the forecast stands for scoring with early-session tape as context, not pre-open pre-commitment.
Forecast: p_trade 0.20 | direction long | conviction low
Verdict: No setup is the base case; the only live path is a borderline NVDA→ASML fire, and even that carries compressed edge. This is the low end of the desk's tradeable range — p_trade 0.20. The CPI-eve read dominates: quiet, low-vol, no same-day catalyst, and the semis complex sagged Monday (NVDA -2.9%, AMD -2.8%). The briefing is explicit that no tech-sector leg is pre-assigned today — there is no "semis constructive" narrative to lean on, which removes the input class that powered 08-03's clean winner (constructive NVDA weekend headlines). The comparison day is 08-04 / 08-10, both quiet no-signal days (AMD +1.04% near-miss then +0.12%; NVDA -0.71% then -0.57%) — not the fired days 08-03/08-05/08-06, which ran on more directional tapes with live catalysts. On a quiet, flat, event-suppressed tape with a sagging complex, first-hour leader moves staying sub-1.5% is the honest default.
Why the clean path (NVDA→ASML) is the day's one live expression — with a compressed-edge caveat: NVDA sat at ~+1.5% premarket on one feed (220.75, right at the 220.75 threshold) — the only reason this isn't p_trade 0.10. But premarket strength is not the 09:30–09:55 cash return the strategy measures; a short-covering premarket pop that doesn't hold into the first hour is exactly the dead-cat shape that keeps the signal sub-threshold. If NVDA does clear +1.5% in the first hour, it's a clean-path LONG (NVDA_ASML is 2W/0L live, dev combined PF 1.85 — the highest-quality LONG in the book) and the mechanical runner executes. NVDA earnings Aug 26 is 2+ weeks out — no contamination. The wrinkle: if ASML also gaps up with NVDA, the lag is partially closed and the 10:00 entry is late/compressed; a laggard not gapping is the only genuine lead-lag shape left today.
The degraded path (AMD→TSM): AMD premarket ~+0.86% on one feed — well below threshold. AMD_TSM is the weakest cell in the book (n=6, WR 33.3%, PF 0.272; sub-book kill switch pre-registered and not triggered — it governs, not a plan-level gate). If AMD somehow clears +1.5%, the runner takes the AMD→TSM LONG and it is logged/scrutinized as the same shock-reversal class, not celebrated. The pre-registered kill switch governs it; I do not hand-gate it.
Calibration math (day-conditional): P(signal) ≈ 0.30 — a quiet CPI-eve day's default is sub-1.5% first-hour impulse, the complex sagged Monday, there's no semis catalyst and no pre-assigned tech leg; the only live source is NVDA sitting at the threshold premarket, which is not the measured cash return. P(LONG | signal) ≈ 0.60 — the last four signals have been UP and the up paths are the only tradeable ones (down paths are disabled SHORTs by construction), but with no constructive narrative and a sagging complex the up-skew is modest, not decisive. Net tradable-signal probability ≈ 0.20 (conservatively below 08-10's 0.30 — today is more event-suppressed and the complex is weaker). Direction long by construction (LONG-only book). Conviction low: this is a "possible, if the threshold hold" call, not a "likely" day, and I am not anchoring on the 3-of-4 hot streak from last week on a tape the briefing explicitly declines to call constructive.
Sizing vs. event risk: standard 0.75%, single position — no change. Event risk today is HIGH only by CPI-tomorrow convention, which this desk structurally cannot carry: the position hard-flats at 15:30 (never survives into Wednesday's CPI by construction), the traded laggards (ASML/TSM) don't report today, and VIX 15.59 is far from the ≥ 30 barrier. The event risk that matters is signal quality (compressed lag, degraded AMD class) — handled by scrutiny and invalidation, not a size cut. Book context: net −$182.42 on 8 trades (4W/4L, PF ≈ 0.70); SPRT CONTINUE (LLR +0.398, mid-band — +2.546 to CONSISTENT-WITH-BACKTEST, −3.343 to DEGRADED); AMD_TSM sub-book n=6 / PF 0.272 (not triggered); SHORT shadow book n=5; 30-LONG gate n=8/30 (a NVDA→ASML fire today is the highest-quality LONG available — that pair is 2/2 live).
Key levels (spread triggers, from Monday closes): NVDA +1.5% ≈ 220.75 → clean NVDA→ASML LONG (compressed-edge caveat if ASML also gaps up); NVDA -1.5% ≈ 214.23 → disabled SHORT (shadow n=6). AMD +1.5% ≈ 476.76 → degraded AMD→TSM LONG; AMD -1.5% ≈ 462.67 → disabled SHORT. Laggard gaps at the open: ASML gapping up big = lag partially closed (late entry); TSM gapping with AMD = spread zero; a laggard not gapping with its leader is the only genuine lead-lag signal shape left today. Index: QQQ holding the 720.80 prior-close area through the morning matters; cash-open gaps printed small (no absorption).
Invalidation (stand down entirely):
Information cost (documented, not acted on): if either leader cascades (-1.5%: NVDA ≈ 214.23, AMD ≈ 462.67), the disabled SHORT is logged to the shadow book (n=5 → n=6; n=7 if both). LONG-only keeps costing information in down-tape; the shadow book is the accounting, the 30-LONG gate (n=8/30) is the governing re-enablement path, and the 07-24 learnings observation (the LONG gate may be structurally unreachable in a sustained semis bear — shadow book as the alternate evidence path) stays live.
No trades taken.