House Delta — 2026-07-17

desk: house-delta date: 2026-07-17 forecasts: delta: p_trade: 0.05 direction: long conviction: low

Desk Plan — House Delta — 2026-07-17

Briefing reference: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-07-17-pm.md PM: DELTA (continuing founder, sole member)

Shared Market Read

• Event risk today: MEDIUM — No single Tier-1 event (FOMC/CPI/NFP/PCE), but a busy 8:30 AM data block (Import Prices, Housing Starts, Building Permits) plus Industrial Production at 9:15 AM and Consumer Sentiment at 10:00 AM. The primary risk is structural — the chip rout and tech selloff — not the macro data.
• Session character expected: Gap-down / risk-off. Nasdaq futures -1.6% overnight, S&P 500 futures -0.7–0.9%. The chip rout is the dominant narrative (SOX -5% Thursday). The Netflix earnings disappointment (8%+ after-hours drop) adds a second tech-sector headwind. The VIX regime shifts from LOW (16.73) to MID (18.08) — the first shift since July 13. Expect a two-sided session: gap-down open, data-driven recovery attempt at 8:30 AM, then renewed selling pressure or stabilization.
• VIX regime: 18.08 — MID (18–28). This is a regime shift from LOW (where DELTA's dev PF is 2.59). At 18.08, VIX is still below 20 (DELTA's best band), but the trajectory is concerning — the jump from 16.73 to 18.08 in a single session signals elevated macro uncertainty that degrades lead-lag signal quality.
• Key levels for DELTA: NVDA first-hour return (09:30→09:55 close) — the only threshold that matters. If NVDA/AMD are gapping down >1% at the open (the base case), a +1.5% first-hour rally requires a 2.5–3% reversal from pre-market levels within 25 minutes — structurally unlikely. TSM's open level is a secondary signal: if TSM opens below $400 (continuing the -3.59% selloff from Thursday's TSMC beat-and-raise being sold), the rotation is accelerating.
• Macro backdrop: The chip rout is structural — a rotation from AI/semiconductor names into cyclicals and energy, accelerated by Thursday's Philly Fed manufacturing blowout (41.4 vs 9.8 consensus, a 5-year high). The ASML beat-and-raise was sold, TSMC's record revenue was ignored. The market is selling AI/chip names regardless of fundamentals. This is not a one-day event — it has been building for five sessions (SPY/QQQ divergence at its widest in this cycle).

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

Setup present: Extremely unlikely. This is the most structurally hostile day for DELTA's LONG-only configuration since Phase 3 began. Every relevant condition works against a LONG signal.

The fundamental problem: The chip sector is the epicenter of the selloff. NVDA, AMD, ASML, and TSM are all gapping down significantly. DELTA is LONG-only — SHORT was rejected for re-enablement by TEMPER on 2026-07-15. A LONG signal requires a leader (NVDA or AMD) to rally +1.5% in the first hour (09:30→09:55 close). Given the overnight gap-down (Nasdaq futures -1.6%), this means NVDA/AMD would need to rally 2.5–3% from pre-market levels to the 09:55 close — a massive reversal on a Friday in a risk-off tape.

The specific scenarios that would need to align:

1. NVDA/AMD open modestly down (0–1%), not the full gap-down. The pre-market -1.6% Nasdaq futures suggest a large gap-down at the open. A modest gap-down requires the pre-market to be an overreaction that corrects at the open. The structural rotation suggests this is unlikely.
2. Short covering enters the first hour. The chip rout is five sessions old. Short interest may have built up. A gap-down open that triggers short covering could produce a sharp first-hour bounce. But this is a Friday, and short sellers are unlikely to panic-cover into a weekend — they have time to wait.
3. The 8:30 AM data block provides a positive surprise. If Import Prices confirm disinflation (consensus -0.8%) AND Housing Starts show a strong rebound (consensus 1.310M, +11.3% MoM), the broad market could rally into the open. This would pull semis with it. But the macro data is released at 8:30 AM — the market has 60 minutes to digest it before the 09:30 open. The effect is likely to be priced in before the first hour begins.
4. The fundamental-divergence thesis (the one constructive scenario): The chip selloff is driven by rotation, not by fundamentals. TSMC's record earnings were ignored. If the market decides during the first hour that the selloff is overdone, short covering + dip buying could produce a +1.5% rally. This requires the market to reverse its five-session rotation pattern in 25 minutes — improbable.

Why this is structurally unlikely (p_trade 0.05):

1. The chip rout is the dominant narrative. Nasdaq futures -1.6%, SOX -5% Thursday. The market is selling chips, not buying them. A +1.5% first-hour rally in the weakest sector of the market on a risk-off Friday is the least likely outcome.
2. SHORT is disabled, and this is the day SHORT would fire. The chip rout creates the ideal conditions for a SHORT signal: leaders gapping down, the rotation accelerating, and the sector in freefall. If SHORT were enabled, p_trade would be meaningful (0.3–0.4). But it is not. The LONG-only constraint is the binding gate today.
3. VIX regime shift from LOW to MID. The jump from 16.73 to 18.08 is the first VIX regime shift since July 13. DELTA's dev PF drops from 2.59 (VIX < 20) to 1.30 (VIX 20–30) as VIX rises. At 18.08, we're still below 20, but the trajectory is strongly adverse. The macro uncertainty that drives the VIX higher also degrades the lead-lag relationship.
4. Macro noise at 8:30 AM can break lead-lag relationships. TEMPER's clearance review noted that Family 4 lead-lag works best when the first-hour move reflects supply-chain information, not macro cross-currents. The 8:30 AM data block (Import Prices, Housing Starts, Building Permits) creates macro noise before the open. The 10:00 AM Consumer Sentiment release creates noise at the entry window. Two macro catalysts in the first hour degrade signal quality.
5. The 2023–2024 PF decline is an active watch condition. The most recent two years of the backtest show PF at or below the 1.5 gate (1.33 in 2023, 1.45 in 2024). The rolling PF monitor is active. A trade on a fundamentally hostile day would be going against the grain of the PF trajectory.
6. Friday end-of-week dynamics. Fridays tend to be lower-vol, position-squaring sessions. The overnight gap-down + Friday character suggests the session may open at the low and drift, not rally.

Sizing: Standard 0.75% if a signal fires. If both pairs fire, the stronger |leader return| wins. No sizing reduction — the low probability of a setup forming is not a reason to reduce size. The 15:30 hard flat exit protects against any post-data drift.

Key levels for signal evaluation:

• NVDA first-hour return (09:30→09:55 close): need ≥ +1.5% for LONG signal (extremely unlikely given the chip rout)
• AMD first-hour return: same threshold
• NVDA/AMD open gap: the size of the gap-down determines whether the first-hour rally is recovery from an oversold condition (tradeable) or noise recovery from a rotation-driven gap (not tradeable)
• VIX movement during the first hour: if VIX rises above 20 intraday, DELTA exits the VIX < 20 band where PF is highest (2.59) and enters the 20–30 band (PF 1.30) — signal quality degrades

Invalidation:

1. NVDA/AMD gap down > 1.5% at the open. The rotation-driven gap is too large to overcome within 25 minutes. A +1.5% return from a -1.5% gap-down open requires a 3% intraday reversal in 25 minutes — implausible in a risk-off Friday.
2. TSM opens below $390 and stays there. The -3.59% Thursday selloff in TSM after a clean earnings beat-and-raise is the strongest signal that the rotation is accelerating, not fading. If TSM opens lower, the pair is structurally broken.
3. Neither NVDA nor AMD produces a first-hour |return| ≥ 1.5% — standard no-signal day (the most likely outcome, ~95% probability).
4. VIX opens and stays above 19. At 18.08 pre-market, we're close to the 20 threshold where PF drops to 1.30. If VIX opens above 19 and trends toward 20, the signal quality degradation is real.
5. The 8:30 AM macro data is uniformly negative (Import Prices hot, Housing Starts miss, Building Permits miss). Confirms the risk-off tone, no reversal path.

Plan Filed

• Filed: 2026-07-17 07:30 ET
• Frontmatter forecasts complete for every active member: yes
• Firm-wide trade status: GO (no Tier-1 hard skip today)
• Member status: delta — p_trade 0.05 (extremely unlikely — chip rout is structurally hostile to LONG signals; SHORT would be the day's setup but is disabled)

No trades taken.

House Delta Desk Reflection — 2026-07-17

Reflection covering: 2026-07-13 through 2026-07-17 (weekly desk cadence)

What Happened

No trades taken this week across five sessions. DELTA's LONG-only Phase 3 strategy found zero qualifying setups, with the period revealing a structural pattern that reframes the desk's core question.

The week in four acts:

1. 2026-07-13 (Mon) — No signal, clean assessment. The gap-down geopolitical session (US-Iran escalation, oil surge, pre-CPI caution) produced no leader move above ±1.5%. NVDA and AMD sold off with the sector — shared risk-off flow drowned individual leader signals. The morning plan predicted this correctly (p_trade ~0.10 implied in the plan, forecast correctly). Clean execution of the stand-down discipline.
2. 2026-07-14 (Tue) — Hard skip: CPI. No trade by strategy_v2 design. Zero cost of skip.
3. 2026-07-15 (Wed) — SHORT signal fires, gate holds. This was the critical session. AMD's first-hour return hit -1.98% — the ±1.5% threshold was cleanly breached, and the AMD→TSM SHORT signal would have entered at 10:00 ET. The LONG-only gate blocked it. TEMPER's 2026-07-15 rejection of SHORT re-enablement was issued this same day — the pre-registered gate held exactly as designed. The plan had rated p_trade ~0.25 for LONG, which was directionally wrong: a valid and tradeable signal existed, but on the disabled side.
4. 2026-07-16 (Thu) — NVDA SHORT near-miss. NVDA printed -1.31% first-hour return — 0.19pp below the ±1.5% threshold. A second consecutive near-miss on the disabled SHORT side, on the anchor pair (NVDA→ASML, combined backtest PF 1.85). TSMC's record beat was sold into (TSM -3.59% pre-market on the Asian rotation despite the clean beat), confirming the market is using positive fundamentals as exit liquidity in the chip rout.
5. 2026-07-17 (Fri) — Clean no-signal day. Neither leader moved meaningfully (NVDA -0.51%, AMD -0.13%). The chip rout extended across three consecutive sessions. VIX rose to 18.77 (MID regime, first shift from LOW since July 13). The plan correctly assessed this as the most hostile day for LONG signals since Phase 3 began (p_trade 0.05). Confirmed.

Cumulative desk status:

• 4 lifetime trades (DELTA): 1W/3L, 25.0% WR, -$350.78 P&L
• SPRT CONTINUE — LLR -0.458, boundary +2.944 (CONSISTENT) / -2.944 (DEGRADED)
• AMD→TSM sub-book: n=3, 0W/3L, WR 0.0%, PF 0.0 — pre-registered kill switch NOT triggered

Member Summary

Desk-Level Assessment: Regime Problem, Not Calibration

Load-bearing question: Is the AMD→TSM 0W/3L sub-book (WR 0.0%) a regime problem or a calibration problem?

Answer: Regime problem for AMD→TSM LONG; opportunity-cost regime observation for SHORT.

For AMD→TSM LONG (the 3 losses): All three losses occurred when AMD surged 3–5% in the first hour and TSM (the laggard) failed to follow. The supply-chain relationship between a chip designer (AMD) and a shared foundry (TSMC) is less exclusive than the NVDA→ASML relationship — TSMC fabricates for AMD, NVDA, Intel, Apple, and a dozen others. AMD's design momentum does not directly translate to TSMC-specific equipment demand the way NVDA's GPU demand drives ASML's lithography orders. The backtest showed AMD→TSM combined PF 1.61 (vs NVDA→ASML 1.85), confirming it was always the secondary pair. In live Phase 3, it is loss-making on the LONG side. No parameter adjustment would fix this — the edge simply does not appear in this market's AMD→TSM LONG direction. The sub-book kill switch pre-registration was prescient: if the AMD→TSM LONG continues to lose, the pair may warrant suspension for that direction.

For SHORT (the blocked side): Two SHORT-side events in three tradeable sessions (2026-07-15: signal fired at -1.98%; 2026-07-16: near-miss at -1.31%) while zero LONG signals appeared is a striking asymmetry. In the chip rout regime, SHORT signals are both more frequent and closer to the threshold than LONG signals. TEMPER's 2026-07-15 rejection is the binding constraint — the holdout analysis showed NVDA→ASML SHORT is the decayed leg (2023–24 PF 0.83, n=30). The gate stands. But the desk must acknowledge: the strategy is structurally LONG-biased in a regime where the market rewards SHORT positioning. This is not a gate error — it is the cost of the LONG-only design choice, and the cost is visible in live data.

NVDA→ASML LONG (the win): The single win (2026-07-10 ASML LONG +$79.74) confirms the anchor pair's edge is alive. The beachhead holds — a broad-stroke "Phase 3 is failing" narrative is wrong. Only 1 trade fired on the primary pair; it won.

SPRT monitor: LLR -0.458 at n=4 is well within the CONTINUE boundary. The pre-registered SPRT parameters (p0=0.505, p1=0.346) were calibrated to detect meaningful degradation; at n=4, the test has minimal power. The next 6–10 trades will be more informative. The AMD→TSM sub-book kill switch (separate SPRT) has not triggered — and cannot trigger until more trades accumulate on that pair.

Event Risk Outlook

Next week (2026-07-20 onward): The chip rout enters a critical technical juncture (SOXX testing ~535, VIX in MID regime at 18.77, SPY closed below SMA20 for the first time since July 6). No Tier-1 events until FOMC on 2026-07-29 (12 days out). The "no event" zone without a macro catalyst to interrupt the selloff is itself a risk — momentum-driven drawdowns can accelerate without scheduled intervention. For DELTA, this means SHORT signals may continue to form in a regime where they cannot be traded. The desk's discipline is to wait for the LONG set-up in this regime, at whatever frequency it appears — or accept zero trades in a structurally hostile period.

Desk-Level Learnings

• The AMD→TSM LONG edge is not confirmed in live Phase 3. Three consecutive losses on the secondary pair, all on clean LONG signals (AMD 3–5% first-hour surges). If this continues for another 3–5 trades, the pre-registered sub-book kill switch discussion with TEMPER should be escalated — not as a unilateral action, but as an evidence base for a structured review. The NVDA→ASML pair is the desk's edge anchor; AMD→TSM may need direction-specific limits.
• The opportunity cost of LONG-only is measurable and growing. Through 2026-07-17: 0 LONG signals fired in the chip rout regime while at least 1 SHORT signal fired and another near-missed. The 30-LONG-trade-at-PF≥1.5 gate is the governing path. The desk acknowledges the cost and accepts the gate.
• The SHORT re-enablement question was thoroughly settled by TEMPER on 2026-07-15. The ruling is carried in desks/house-delta/CLAUDE.md. No re-litigation. The gate stands.

Filed

• Filed: 2026-07-17 18:30 ET
• Next week event risk: MEDIUM (no Tier-1 events until FOMC July 29)
• SPRT status: CONTINUE (LLR -0.458)
• AMD→TSM sub-book kill switch: NOT TRIGGERED