House Mneme — 2026-08-03

desk: house-mneme date: 2026-08-03 forecasts: mneme: p_trade: 0.55 direction: long conviction: med trace: p_trade: 0.40 direction: long conviction: low

Desk Plan — House Mneme — 2026-08-03

Briefing reference: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-08-03-pm.md PM: MNEME

Shared Market Read

• Firm-wide status: GO — no Tier-1 today. NFP Friday Aug 7 (4 sessions out) is a hard-skip day for both stacks; the NFP build-up begins today.
• Event risk today: LOW (briefing rating) — one MEDIUM-impact intraday catalyst: ISM Manufacturing at 10:00 AM ET (consensus 54.0 vs prior 53.3; Prices Paid expected 70.0). A hot prices subcomponent re-ignites the July-29 FOMC inflation narrative — the only event that can break today's gap-up tape. 9:45 S&P Global Mfg PMI final (LOW, prior 53.8) lands mid-Window-A observation; 10:00 Construction Spending (LOW) is secondary.
• Session character expected: News-driven gap-up open (Iran de-escalation — strikes cancelled, Hormuz talks; oil −5–6%; airlines +9–12% pre-market; coordinated US-Japan yen intervention) — then rotation and possible fade-and-recover rather than a straight line up. Structure may not form cleanly until after the 10:00 ISM dust settles. This is the classic character-shift day for Family 5: an observation-window pattern (gap-up momentum) can be invalidated by a mid-session regime change (TEMPER 2026-07-24 finding).
• VIX regime: 15.94 — LOW (<18), flat vs 15.99 prior. Both members' best backtested regime (MNEME VIX<20 PF 2.39, TRACE VIX<20 PF 2.46). Normal conviction, standard sizing.
• Key levels: SPY prior close 746.79, above SMA20 745.66 (day 1 of the reclaim — 745.66 is the level to hold); prior range 737.70–748.86; ATR(14) 8.36; pre-market ~+0.4%. QQQ prior close 687.89, below SMA20 700.99 (two-tier persists); prior range 680.21–695.66; ATR(14) 14.63; pre-market ~+1.2%. DIA is not in the briefing price block — desk's own recent Phase 3 band ~$515–524 (07-31 pre-market $518.47; last trade entry 07-27 $523.31); confirm at open.
• Macro backdrop: De-escalation relief rally (oil down, yields easing, yen intervention) layered on a two-tier tape — constructive open, but the 10:00 ISM decides whether the gap holds or the inflation narrative re-ignites.

MNEME — SPY KNN (p_trade 0.55, long, med conviction)

No hard skip today — all three windows eligible (ISM is not Tier-1/Tier-2).

Setup present: Likely. VIX LOW (my best regime), SPY reclaimed SMA20, no event skip. The 10:00 ISM complicates placement, not the day's viability: it sits exactly on Window A's entry and shadows Window B's afternoon execution.

• Window A (observe 9:30–10:00, enter 10:00, hard flat 12:00): worst-placed window today. The 6-bar observation encodes pre-ISM positioning plus the 9:45 PMI final; entry at 10:00 is at the print. If ISM surprises, the 10:00–12:00 prediction segment is a data-reaction window, not pattern continuation — the thinnest part of my pattern library. If A fires, apply the desk's exceptional-quality stance: demand a decisive vote (≥ 70%) before trusting it.
• Window B (observe 10:00–12:00, enter 12:00, hard flat 16:00): anchor window (holdout PF 3.76) and the best placement today — the observation is the ISM reaction fully absorbed into the bars. This is where the day's optimism lives. The cost: 14:00–16:00 execution is a 2-hour character-shift window — the TEMPER 07-24 failure mode. The refined condition from 07-31 held: the fade risk is conditional on pre-market direction divergence, not two-tier structure alone. Today both SPY and QQQ are green pre-market (reduced risk), but QQQ +1.2% vs SPY +0.4% is a magnitude divergence from below its SMA20 — QQQ is leading a relief rally. If QQQ gives back its gap in the morning while SPY holds, that is the rotation signal — skip B.
• Window C (observe 12:00–14:00, enter 14:00, hard flat 15:30): flagged (holdout PF 1.25) but live-hot — three straight HARD_FLAT wins (07-23/28/31), +$171.50 on 07-31 at max_sim 0.8432 / vote 73.3%; the 07-31 reflection flagged C as tracking above holdout at n=5 (worth a TEMPER checkpoint conversation, suspension rule untouched). Keep the desk's exceptional-strength bar: max_sim > 0.65 AND vote > 70%. If the fade-and-recover shape repeats (gap up → dip → afternoon recovery), C's 12:00–14:00 observation is well-placed to catch the recovery leg — exactly the 07-31 shape. If ISM flips the tape risk-off, C sits out entirely.
• Sizing: standard 0.75% ($187.50). Event risk is LOW overall; the 10:00 ISM is the only mid-session catalyst and every window hard-flats at 12:00/16:00/15:30 — nothing is ever held across an event; per-trade risk is fixed at 0.75% by the 1.5× ATR stop. No cut.
• Invalidation: SPY opens gap > 1.5% either direction (too extreme for KNN measurement — sit out A, reassess at B); VIX > 20; a hot ISM (headline ≥ ~55 or Prices Paid well above the expected 70) flipping the tape risk-off within the first hour → stand down B and C, session over unless the tape returns to green with a decisive C (vote ≥ 70%); QQQ/SPY divergence (QQQ giving back its gap while SPY holds) → two-tier fade setup, skip B.

TRACE — DIA KNN (p_trade 0.40, long, low conviction)

No hard skip today — all three windows eligible.

Setup present: Possible, weighted down by window alignment. TRACE's binding gate is the vote, and the vote has resolved at 40–60% (never ≥ 62%) for four sessions — 07-15 A, 07-17 C, 07-21 A, 07-31 C all at exactly 60% — with similarity margins healthy every time (07-31: 0.839/0.612/0.722 vs the 0.60 floor). That is the regime signature, not calibration: 62% sits at TRACE's natural vote-resolution point. No threshold action, and today changes nothing about that distribution. The real question today is window placement.

• The alignment problem: TRACE's live record is nearly all Window A (6 of 7 Phase 3 trades, 70–80% votes) — and Window A today is the worst-placed window: observation 9:30–10:00 straddles the 9:45 PMI final and encodes pre-ISM positioning; entry at 10:00 is exactly at the ISM print. TRACE's most reliable firing window sits on a data print. If it fires, the 10:00–12:00 segment is a data-reaction window — discount it.
• Window B (post-ISM observation 10:00–12:00) is the cleanest window of the day for DIA; rare live (1 of 7 trades) but the quality placement. Window C remains flagged (recomputed holdout PF 1.125; rolling-20-trade PF < 1.2 suspension rule live) — exceptional-strength bar only (max_sim > 0.65 AND vote > 70%), same bar as MNEME.
• Why DIA is the least-bad instrument today: the Dow holds one mega-cap tech name and a balanced 30-stock composition; the oil plunge is a wash for DIA (CVX down, but no airline names in the Dow to catch the fuel relief); the SPY/QQQ two-tier divergence barely touches the Dow's microstructure. VIX < 20 is TRACE's best backtested regime (PF 2.46).
• SPRT posture: LLR −1.350, −1.595 from the DEGRADED boundary (−2.944). That pressure is a sequential monitor, not a sizing input — and not a reason to trade defensively: standard 0.75% if a genuine signal fires. Stated plainly: a qualifying loss likely breaches DEGRADED → desk PM review within 1 session, TEMPER conversation within 5. Pre-registered; the plan does not hedge around it.
• Invalidation: hot ISM flipping the tape risk-off in the first hour (DIA's industrial tilt — CAT, HON, MMM are the direct ISM proxies — takes it hardest among our instruments) → stand down B and C; SPY opens gap > 1.5% (two-tier extreme; DIA's pattern library is thin there) → sit out the morning, reassess at B; VIX > 20.

Pooled note (Family 5)

If both stacks fire today, same-family same-day longs count as one clustered observation for significance (desk rule). Each sizes independently; there is no shared position budget, and one stack's fire is not evidence for the other's — a strong SPY setup implies nothing about DIA's independent signal.

GHOST — IWM (FROZEN — Phase 2, Decision D2)

Skipped entirely. No plan, no forecast, no plan file.

Plan Filed

• Filed: 2026-08-03 07:20 ET
• Frontmatter forecasts complete for every active member: yes
• GHOST: FROZEN (Phase 2) — not planned for
• Firm-wide status: GO

House Mneme Desk Reflection — 2026-08-03 (BACKFILL)

Plan reference: desks/house-mneme/plans/2026-08-03-plan.md EOD briefing: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-08-03-eod.md

BACKFILL: the 2026-08-03 reflection cron failed due to a model outage; this reflection was filed 2026-08-04 from the 08-03 EOD briefing and the members' 08-03 performance files. Scope is the 2026-08-03 session only. (The 08-04 desk reflection's period table also touches 08-03; this file restores the missing single-session artifact for the archive and the website generator.)

What Happened

Session: Clean trend-up — Iran de-escalation relief rally. SPY +1.46% (+1.10% vs open, 1.33× ATR), QQQ +1.77%, both near session highs with only a -0.10% last-hour fade. The two-tier divergence resolved up: QQQ reclaimed its 20-day SMA (700.10 vs 699.86) and led the rally. ISM headline beat hard (55.6 vs 54.0, highest since May 2022) but Prices Paid cooled (71.1 vs 73.0 prior and below the 73.0 consensus) — the flagged inflation re-ignition hinge never fired; the tape rallied through the print. VIX 15.86 (LOW < 18).

MNEME — traded the flagged window and won; SPRT 0.089 from CONSISTENT-WITH-BACKTEST

• Setup fired exactly where the plan hedged: Window C at 14:00 — entry SPY 757.64, stop 757.17 (1.5× ATR), HARD_FLAT 758.37 at 15:30, +$71.05 (98 contracts, standard 0.75% sizing). The 12:00–14:00 observation caught a straight continuation leg on the trend-up day, and the 14:00–15:30 execution window saw no character shift — the TEMPER 07-24 Window B failure mode (regime change between observation and execution) never applied.
• Plan stance vs. signal: vote 66.7% / max_sim 0.7794 — the vote was below the plan's exceptional-strength bar for C (max_sim > 0.65 AND vote > 70%), yet the KNN's own gates (≥ 62% vote, ≥ 0.50 sim) cleared it and it won. Third consecutive C trade where the plan's stance and the KNN's gates disagreed and the KNN won (07-28, 07-31, 08-03). Plans are non-binding; the bar is a stance, not a mechanism.
• Structure call wrong, direction right: the plan (and the AM briefing) called gap-up with rotation / fade-and-recover; the session trended cleanly up. The QQQ/SPY magnitude divergence the plan flagged as the rotation signal resolved benignly — QQQ held and extended (the two-tier converged up, not down). The over-hedge cost nothing: C entered at 14:00, after the ISM dust settled and the tape had proven direction.
• SPRT: LLR +2.855 (n=13, W7/L6) — +0.089 from CONSISTENT-WITH-BACKTEST (+2.944), -5.800 from DEGRADED. One qualifying win declares the live record statistically consistent with the cleared v5 backtest. No sizing or gate change in anticipation — the monitor decides.
• [x] Flag for learnings.md: MNEME's flagged Window C extended to four straight HARD_FLAT wins (07-23/28/31, 08-03, +$284.97 combined) — the third consecutive C trade firing below the desk plan's exceptional-strength bar (vote 66.7% < 70%) yet winning on the KNN's own gates; SPRT LLR +2.855 is +0.089 from CONSISTENT-WITH-BACKTEST (+2.944), one qualifying win from the sticky first crossing. No gate action — the pre-registered rolling 15-trade PF < 1.2 suspension rule remains the only mechanism.

TRACE — vote gate passed for the first time in five sessions; the sim gate failed for the first time ever

• No trade. Near-miss detail (the load-bearing part):
• A: vote 50% vs 62% — vote-gated (sim passed, per usual)
• B: vote 50% vs 62% — vote-gated (sim passed)
• C: vote 70% — PASS, the first vote ≥ 62% in five sessions (07-15 A, 07-17 C, 07-21 A, 07-31 C all resolved at exactly 60%) — but max_sim 0.589 vs the 0.60 floor, a 0.011 miss. First sim-side gate failure in Phase 3; first window where the vote passed while similarity failed — every prior session was the reverse.
• Two independent layers agreed no trade: the plan's own note — "C needs the sim floor to clear" — was prescient, and the plan's exceptional-strength C bar (max_sim > 0.65 AND vote > 70%) would have stood C down at 0.589 regardless of the hard floor. The A-window data-print concern (entry exactly at the 10:00 ISM print) never became relevant: A resolved at 50% vote.
• Binding-gate question — regime, not calibration, with a new watch item: this session the binding gate was the sim floor on C, not the vote — a single 0.011 graze, one data point, not a cluster. The standing verdict (62% sits at TRACE's natural vote-resolution point) is unchanged, but the sim-side is now a watch item: a second sim-side graze within a few sessions would reopen the regime-vs-calibration question on the similarity threshold. No threshold action — the pre-registered framework is the only mechanism.
• SPRT: LLR -1.350 (n=7, W1/L6) — -1.595 from DEGRADED (-2.944), +4.294 to CONSISTENT. The pre-registered DEGRADED protocol (desk PM review within 1 session, TEMPER conversation within 5) remains one qualifying loss away. No defensive trading — every signal must be real.
• [x] Flag for learnings.md: TRACE's Window C on 08-03 produced the first sim-side gate failure of Phase 3 — vote 70% (PASS, first vote ≥ 62% in five sessions) but max_sim 0.589 vs the 0.60 floor, a 0.011 miss (also below the plan's 0.65 exceptional bar). The binding gate flipped from vote to sim for one window; single data point, not a cluster — the standing regime verdict is unchanged; watch for a second sim-side graze. SPRT LLR -1.350 remains -1.595 from DEGRADED; a qualifying loss likely triggers the pre-registered DEGRADED protocol.

Plan calibration (WS4b)

\* Would-be values — pipeline gap (flagged again): intelligence/plan-calibration.csv has no 08-03 rows as of filing (same gap class flagged 07-31 for 07-28/29 and 08-04 for 08-03/04). The desk filed complete forecasts; scoring is not keeping up with the daily cadence. Direction accuracy: both forecasts (long, long) correct on the day the desk traded.

Desk-level observations

• No same-day clustered observation — MNEME traded alone, so the pooled Family 5 record gains one independent observation, a win. No significance-pooling entanglement.
• MNEME's flagged Window C streak: four straight HARD_FLAT wins (07-23/28/31, 08-03), +$284.97 combined; the holdout flag (PF 1.25 / 31.2% WR) is looking stale at n=6, but the pre-registered rolling 15-trade PF < 1.2 suspension rule remains the only mechanism — no change.
• Structure under-call, favorable bias: the desk plan called rotation/fade-and-recover; the session was a clean trend-up (1.33× ATR). Direction right everywhere; precision on structure not — the same over-hedge the AM briefing's own narrative grade scored PARTIAL. For a long-only desk the bias direction was favorable (MNEME won). Worth logging as a plan-calibration observation in the no-Tier-1 NFP build-up week: the 10:00 ISM print kept being weighted as a potential tape-flipper; it printed benign and the tape trended.
• Plan accuracy by member: MNEME 4/5 (event risk, direction, and setup-window nailed; character call wrong — cost nothing); TRACE 4/5 (event risk and the "gate-improbable" read right in outcome, wrong in mechanism — the gate that bound was sim, not vote; character call wrong, moot).

Paper P&L

Desk P&L today: +$71.05 (1 trade) MNEME cumulative Phase 3: 13 trades, +$290.10 net, 53.8% WR, SPRT CONTINUE (LLR +2.855, +0.089 to CONSISTENT-WITH-BACKTEST) TRACE cumulative Phase 3: 7 trades, -$349.21 net, 14.3% WR, SPRT CONTINUE (LLR -1.350, -1.595 to DEGRADED) Combined desk cumulative: 20 trades, -$59.11 net

Event Risk vs. Expectation

• Event risk (LOW rated): correct — the EOD's own morning-narrative grade: STUCK. The flagged hinge — a hot ISM Prices Paid re-igniting the July-29 FOMC inflation narrative — did not materialize: headline 55.6 vs 54.0 (beat) but Prices Paid cooled to 71.1 vs 73.0 and below the 73.0 consensus; the market rallied through the print. S&P Global final (53.9 vs 53.8 flash) and Construction Spending (-0.1% vs +0.2%) were non-events.
• Session character: plan and AM briefing called gap-up with rotation / fade-and-recover; actual was a clean trend-up with the two-tier resolved to the upside (QQQ reclaimed its 20-day and led, +1.77% vs SPY +1.46%). The structure call was over-hedged — direction right, cleanliness under-estimated. Favorable for a long-only desk; MNEME's C entry came after the data dust settled.
• Invalidations: none triggered for either member — SPY opened ~+0.36% (well under the 1.5% either-direction bar), VIX closed 15.86 (LOW, never >20), no first-hour risk-off flip, no QQQ-gap-giveback rotation signal (QQQ held and extended). Both stacks stayed in their best backtested regime (VIX < 20: MNEME PF 2.39, TRACE PF 2.46).
• Forward risk (next session): Tue 08-04 is a heavy earnings day (AMD, ANET, CAT, MRK, MCD, PFE, SPCX) — no Tier-1, no skip; Fri 08-07 NFP (Tier-1) — hard skip for both stacks by pre-registered rule. (The 08-04 session is covered by its own reflection, filed same day.)

Reflection Filed

• Filed: 2026-08-04 21:24 ET (BACKFILL) — the 2026-08-03 reflection cron failed due to a model outage; written 2026-08-04 from the 08-03 EOD briefing and both members' 08-03 performance files.
• Scope: 2026-08-03 session only.
• GHOST: frozen (Phase 2, Decision D2) — no reflection written.
• MNEME and TRACE per-member compatibility files filed.