House Delta — 2026-08-14

desk: house-delta date: 2026-08-14 forecasts: delta: p_trade: 0.30 direction: long conviction: low

Desk Plan — House Delta — 2026-08-14

Briefing reference: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-08-14-pm.md PM: DELTA

Shared Market Read

• Event risk today: MEDIUM (briefing) — July Retail Sales 8:30 AM ET + Michigan Sentiment Prelim 10:00 AM ET. Neither is in this desk's hard-skip set (FOMC/CPI/NFP/PCE only; VIX ≥ 30). Hard-skip check CLEAR, firm-wide GO — this desk CAN trade. Retail Sales prints before DELTA's 09:30–09:55 leader-measurement window and 10:00 entry, so the primary macro fork is absorbed into the open, not carried as a live-position risk. Note the Michigan print lands exactly at DELTA's 10:00 entry bar — second-tier, not a skip, but a minor live-print nuance to watch (see member section).
• Session character expected: Trending-up bias, event-conditioned / two-sided. The briefing's explicit calibration (from 08-13 EOD, after the PPI morning): on a benign-macro morning, weight the macro leg — a benign Retail Sales print (consensus +0.1%) keeps the disinflation bid intact and the S&P extends its record-high drift; the AMAT sell-the-news gap-down (~-5% premarket) is a conditional risk leg, NOT the index base case. If Retail Sales surprises hot OR the AMAT fade broadens into the semis complex, expect a two-sided-to-weak session.
• VIX regime: 14.53 — LOW (<18), DELTA's best backtest regime (dev PF 2.59 in VIX < 20 vs. 1.30 in 20–30); the ≥ 30 hard-skip barrier is far away. This is the single most supportive input and it licenses trading today.
• Key levels (DELTA): briefing carries no NVDA/ASML/AMD/TSM prices (not in its coverage). Last cached closes are 08-10 (NVDA 217.49, AMD 469.71, ASML 1,733.90, TSM 418.59) — stale (08-13 trade shows TSM ≈ 434). Indicative spread triggers from the 08-10 cache (NOT verified-current): NVDA +1.5% ≈ 220.75 → clean NVDA→ASML LONG; NVDA -1.5% ≈ 214.23 → disabled SHORT. AMD +1.5% ≈ 476.76 → degraded AMD→TSM LONG; AMD -1.5% ≈ 462.67 → disabled SHORT. The runner computes its own triggers from the actual 08-13 close.
• Macro backdrop: S&P at a record (27th all-time high this year) on back-to-back benign inflation (in-line CPI Tue, soft PPI Thu); benign-expected Retail Sales argues for risk-on continuation. Two live legs: a broad-macro bid vs. a marquee semi-capex sell-the-news (AMAT) and an oil/Hormuz escalation (Brent ~$88, +5% week) that is an energy trade, not a semis trade — the semis-specific tension is the AMAT gap and the memory-strength (SNDK +13.7%, MU +4.2%) vs. AI-networking-bleed (CSCO -8%, CBRS -14%) divergence.

PRICE-VERIFICATION LIMIT (stated per dispatch): the pm briefing is Alpaca/IEX-verified but covers none of DELTA's four instruments. The 08-10 cache is stale (no verified 08-11/08-13 closes for NVDA/AMD/ASML; TSM moved to ~434 on 08-13). I anchor spread triggers from the cache with that caveat and treat any 08-14 premarket print for NVDA/ASML/AMD/TSM as unverified — the runner measures the actual 09:30–09:55 first-hour returns, which are the only authoritative signal inputs.

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

Forecast: p_trade 0.30 | direction long | conviction low

Verdict: a trade is genuinely live today — this is a risk-on-data morning for a LONG-only desk, and I am applying yesterday's calibration lessons (weight the macro leg; price the union of leaders, not one cell). p_trade 0.30, up from 0.15 on the identical structural setup that actually fired a LONG on 08-13. Yesterday the plan built its base case on a down-AI tape (CSCO/CBRS gap-down) and got the wrong side of the tape: the broad market rallied to a record on soft PPI while the single names bled, and AMD cleared +1.5% → the mechanical runner took the degraded AMD→TSM LONG. Today is the mirror setup with the calibration already baked in: benign Retail Sales expected, LOW VIX, record-high drift, and an AMAT sell-the-news gap as the conditional risk leg — not the base case. The disciplined read:

• Base case (risk-on continuation): Retail Sales benign → the macro leg releases the bid → NVDA and/or AMD clears +1.5% in the first hour → a LONG fires. This is now the operative branch, not a tail. Yesterday proved the runner will take whichever leader clears, and on a risk-on day the degraded AMD→TSM cell is the more likely to fire (AMD has been the more active leader; its -1.5% disabled SHORT has fired 6× live while NVDA's has never). Price the union.
• Second branch (no_signal): the AMAT sell-the-news drags the semis complex into a quiet, sub-±1.5% first hour — the recent 07-17/07-20/07-23 pattern. On a two-sided day this is entirely plausible and would be a no-trade.
• Third branch (short_disabled): if AMAT's fade broadens and the semis bleed, AMD clears -1.5% → a disabled SHORT, logged to the shadow book (AMD_TSM shadow n=6 → n=7), no trade. NVDA clearing -1.5% would be a new disabled-SHORT datapoint (first for that pair). This is the second most likely impulse outcome after a benign-print bid, but it is structurally untradeable.

Calibration math (day-conditional): P(signal) ≈ 0.55 — risk-on data morning, LOW VIX, two leaders with real first-hour impulse potential (up materially from the dead-flat no_signal days, where P(signal) runs ~0.15–0.25). P(LONG | signal) ≈ 0.55 — the risk-on base case favors the up path (and the down path is disabled anyway), but AMAT's direct semi-capex drag is a live two-sided risk, so the LONG branch is a lean, not a lock. Net tradable-signal probability ≈ 0.30. Direction long by construction (LONG-only book). Conviction low: a trade is more likely than a quiet no_signal day, but the single highest-probability outcome is still a two-sided day resolving sub-threshold, and the LONG path requires NVDA or AMD to clear +1.5% into an AMAT-gap morning.

Sizing vs. event risk: standard 0.75%, single position — no change. Retail Sales (8:30) is absorbed into the open before measurement; the Michigan print (10:00) lands at the entry bar — second-tier, not a skip, handled as a nuance: if a signal fires and Michigan is live at the 10:00 entry, the entry is unchanged (the leader return was already measured 09:30–09:55), but I watch for a Michigan-spike whipsaw on the laggard in the first post-entry minutes. The hard flat at 15:30 removes any overnight. The real risk is signal quality on a two-sided AMAT morning — a leader gapping with its laggard compresses the lead-lag spread → late entry — handled by scrutiny and invalidation, not a size cut. No scale-down warranted; a possible trade is not a risk to pre-cut.

Key levels (indicative from 08-10 cache, NOT verified-current): NVDA +1.5% ≈ 220.75 → clean NVDA→ASML LONG (the high-quality cell: 2W/0L live, dev combined PF 1.85); NVDA -1.5% ≈ 214.23 → disabled SHORT (new shadow datapoint if it fires). AMD +1.5% ≈ 476.76 → degraded AMD→TSM LONG (sub-book n=7, WR 28.6%, PF 0.226 — the chronically-degraded cell; taken mechanically if AMD clears, scrutinized, not celebrated); AMD -1.5% ≈ 462.67 → disabled SHORT (shadow n=6 → n=7). Laggard cash-open gaps are the second read: ASML/TSM gapping with their leader closes the spread (late entry); a laggard not gapping is the only genuine lead-lag shape left. Index tone to watch: whether the S&P record-high drift holds through a benign Retail Sales and whether AMAT's fade stays name-contained or broadens into the complex — that decides the risk-on-LONG vs. short_disabled/no_signal branch.

Invalidation (stand down entirely):

• NVDA and/or AMD clear -1.5% in the first hour — a disabled SHORT, not a trade; the desk is closed for the day by the LONG-only constraint (the down path is structurally untradeable).
• SPY or QQQ cash-open gap > 1.5% in either direction — impulse absorbed before measurement; a fresh mid-session >1.5% move against the LONG direction also stands me down.
• A hot Retail Sales print (consumer reacceleration, e.g. materially above +0.1%) undercuts the benign-disinflation narrative and flips the tape risk-off — the AMAT gap broadens, LONG dies. (This is the one macro fork that could kill the setup before it forms; it resolves at 8:30, before my window.)
• The AMAT sell-the-news fade broadens into NVDA/AMD after the open — confirms semis-led weakness, LONG path dead.
• QQQ turns sharply negative in the first 30 minutes while the complex follows — down-tape confirmed, no LONG.

Information cost (documented, not acted on): a down-semis day remains the desk's structural blind spot — if AMD or NVDA cascades -1.5% the disabled SHORT is logged to the shadow book (AMD_TSM n=6 → n=7; NVDA_ASML n=0 → n=1 if NVDA fires it, the pair's first live disabled-SHORT datapoint), but LONG-only keeps costing information in exactly the down-tape a broadened AMAT fade would create. The shadow book is the accounting; the 30-LONG gate (n=9/30 after today's TSM stop) is the governing SHORT re-enablement path; the 07-24 learnings observation (the LONG gate may be structurally unreachable in a sustained semis bear — shadow book as alternate evidence) stays live and is relevant if AMAT's fade broadens. The AMD→TSM sub-book is now the operative calibration conversation — n=7, PF 0.226, chronically firing on risk-on days and losing; DELTA will bring this to TEMPER (a gate question, not a unilateral tweak), per the 08-13 reflection's load-bearing finding.

Plan Filed

• Filed: 2026-08-14 04:34 PDT (pre-open, morning dispatch)
• Frontmatter forecasts complete for every active member: yes
• Hard skip check: CLEAR (Retail Sales/Michigan are NOT in DELTA's skip set; VIX 14.53 < 30) — firm-wide GO, desk CAN trade. No Tier-1 in the 2-day window; FOMC Minutes Aug 19 is not a skip
• Price-verification limit stated: briefing carries no DELTA-instrument prices; 08-10 cache anchors are stale (TSM moved to ~434 on 08-13) — runner measures actual first-hour returns
• Dominant input: benign-expected Retail Sales → risk-on base case (weight the macro leg, per 08-13 calibration); AMAT sell-the-news gap is the conditional risk leg, not the base case. A trade is live at p_trade 0.30 — up from 0.15 on the identical setup that fired a LONG yesterday (price the union)

Reflection — House Delta (DELTA) — Daily — 2026-08-14

Period Performance

Single-member desk. DELTA traded once today — Long TSM 58 @ 426.68, stopped 424.36 (11:25, STOP_LOSS), net -$134.37. Cumulative -$442.33 across 10 paper trades. SPRT sequential monitor CONTINUE (n=10, LLR -0.159, +3.103 from the CONSISTENT-WITH-BACKTEST boundary). AMD→TSM sub-book: n=8, WR 25.0%, PF 0.191 — not triggered, but now a third consecutive loss on this pair and the pattern this desk keeps bringing to TEMPER.

What Worked

• Event-risk handling stayed clean. Retail Sales (8:30, -0.6%, soft) and Michigan (10:00, 51.0) were both MEDIUM-rated and neither is in DELTA's hard-skip set; Retail printed before the 09:30–09:55 measurement window and 10:00 entry, so the macro fork was absorbed into the open with zero live-position exposure. No event missed, none overstated.
• The plan's union-pricing calibration held. p_trade rose from 0.15 → 0.30 on the explicit thesis that "the runner prices the union, not one cell" — and exactly as forecast, the degraded AMD→TSM cell (not the clean NVDA→ASML cell) was the operative path. DELTA forecast a LONG at 0.30, a LONG traded, and direction (long) hit. Brier 0.49 — materially better than yesterday's 0.7225.
• Sizing and mechanics were correct. Standard 0.75%, single position, 2.0× ATR stop, untouched. A normal stop on a normally-sized position in low-vol chop (sub-ATR 0.42–0.47×, VIX 14.25) — not an over-bet.
• The plan named the failure mode in advance. It wrote "the degraded AMD→TSM cell is the more likely to fire on a risk-on day" and priced the union accordingly. The mechanism behaved exactly as the plan feared, again.

What Didn't Work

• The AMD→TSM LONG keeps losing, and the sub-book is now unambiguously a calibration story. Third consecutive loss on this pair (08-13 -$125.54, 08-14 -$134.37). Sub-book now n=8, WR 25.0%, PF 0.191, well under the rolling 1.2 threshold and far under 1.0. Today AMD cleared +1.5% (leader +2.06%) so a LONG did fire — the setup is not merely absent (regime), the degraded cell keeps presenting itself on days when AMD prints the move and it keeps losing (calibration). This is trending directly at DELTA's pre-registered AMD→TSM sub-book kill switch.
• The clean NVDA→ASML cell is the real edge and it keeps not being the operative trade. NVDA→ASML remains 2W/0L live with the strongest backtest edge (combined PF 1.85). The book is being dragged down by the degraded cell firing more often and losing more often — the exact asymmetry DELTA flagged on 08-13.
• A loss on a LOW-VIX, sub-ATR, flat-to-slightly-down day is the weakest context to give up edge on. The whole session was compressed fade-and-hold; the TSM long was stopped in low-vol chop at 11:25 with the position underwater from the 10:00 entry. No clean trending leg to validate a momentum continuation — but the degraded cell fired anyway.

Confidence Changes

• NVDA→ASML primary edge: unchanged, 2W/0L live. This loss does not impeach the clean cell.
• AMD→TSM cell: conviction that it is the structural drag has hardened to near-kill-switch strength. PF 0.191 at n=8 with a third straight loss. The mechanism keeps surfacing it as the operative LONG on AMD-move days and it keeps losing. This desk's standing recommendation stands: bring the AMD→TSM sub-book gate to TEMPER for a calibration conversation. Not a unilateral tweak — but the evidence for a gate is now substantially stronger than at n=7.
• Union-pricing calibration improved. The 0.15 → 0.30 lift and the Brier 0.49 confirm the 08-13 lesson (price the union of leaders, not one centered cell) was worth applying.

New Hypotheses

Junior Agent Queue

None — DELTA is a single-member desk. Hypotheses are flagged for TEMPER's calibration conversation, not queued for junior-agent deployment.

Notes for Next Period

• The load-bearing gate question — regime or calibration? Calibration. The setup did not merely fail to appear (regime); the AMD→TSM sub-book (n=8, PF 0.191) keeps presenting a LONG on AMD-move days and keeps losing — a repeatedly-close, repeatedly-losing degraded cell well under threshold. Evidence worth a TEMPER conversation now, and this desk's standing recommendation is to bring the AMD→TSM gate question to TEMPER at the next review cadence. Never a unilateral change.
• The union-pricing fix is working for calibration; the remaining problem is which cell dominates the union. DELTA now prices the union correctly (Brier 0.49). But the union's operative member is the degraded AMD→TSM cell — fixing p_trade does not fix that a lossy cell is the one firing.
• Next session (08-17 Monday): no Tier-1 in the window. FOMC Minutes lands Wed 08-19. A quiet, sub-ATR tape is the near-term character — patience into next week's policy (FOMC Minutes, Jackson Hole) and retail-earnings (Target, Walmart) catalysts.

Reflection Filed

• Filed: 2026-08-14 (EOD reflection, desk cadence Friday-or-trade — today triggered on trade)
• Next session event risk (from EOD briefing): LOW/NONE — no Tier-1 in the immediate window; FOMC Minutes Wed 08-19 is the next watch.