House Mneme — 2026-07-31

desk: house-mneme date: 2026-07-31 forecasts: mneme: p_trade: 0.65 direction: long conviction: med trace: p_trade: 0.55 direction: long conviction: med

Desk Plan — House Mneme — 2026-07-31

Briefing reference: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-07-31-pm.md PM: MNEME

Shared Market Read

• Firm-wide status: GO — No Tier-1 events today. Clean macro day with ECI (8:30 AM), Chicago PMI (9:45 AM), Michigan Sentiment Final (10:00 AM) as mid-tier releases. The dominant driver is the earnings afterglow from the July 30 mega-cluster (AMZN, AAPL, MA, SYK, GDDY, DXCM, FSLR).
• Event risk today: LOW — no Tier-1 events. ECI (8:30 AM) is the most consequential mid-tier release, a Fed-watched wage inflation print. The 3 FOMC dissenters who voted for a hike cited labor cost pressures, so a hot ECI >1.0% would reinforce the September hike narrative. A cool print <0.7% strengthens the dovish case from recent PCE data.
• Session character expected: Trending — mega-cap tech earnings afterglow. The AMZN +12.75% pre-market surge (following MSFT +15.5% on July 30) extends the AI capex validation narrative into a third day. The AMZN/AAPL divergence (AMZN surging, AAPL -6.33% after-hours on services miss) creates a two-tier structure that is the defining intraday feature.
• VIX regime: 16.78 (LOW, declining from 17.09 prior close). Both members' backtests show healthy PF in VIX < 20 (MNEME 2.39, TRACE 2.46). The declining VIX confirms the session is not in a stress regime.
• Key levels: SPY prior close 741.63 (below SMA20 745.56), pre-market +0.35% to ~$744.30. QQQ prior close 683.60 (below SMA20 702.22), pre-market +0.77% to ~$688.82. DIA prior close ~$515.43, pre-market +0.59% to $518.47. Both SPY and QQQ remain below their 20-day SMAs — the correction trend is not yet broken, but the MSFT/AMZN two-day surge has recovered ~60% of the post-FOMC selloff.
• Macro backdrop: The three converging forces of this week — rates (FOMC hold with 3 hawkish dissents on July 29), macro (dovish PCE + GDP miss on July 30), and earnings (MSFT, AMZN, AAPL, MA mega-cluster) — have produced extreme single-stock moves and two-tier divergences. Today is a clean-up day: the ECI print at 8:30 AM is the only macro event that could challenge the prevailing narrative. The post-earnings environment creates stock-specific dispersion that benefits KNN pattern matching.

MNEME — SPY KNN (p_trade 0.65, long, med conviction)

No hard skip today — clean macro day, no Tier-1 events. All three windows are eligible.

Setup present: Likely. The trending session character, post-earnings dispersion, and low VIX regime create a favorable environment for KNN pattern matching. The key consideration is the two-tier divergence (AMZN/AAPL) and how it interacts with the observation→execution window structure.

Window-by-window assessment:

• Window A (opens 9:30, observes 9:30-10:00, enters 10:00, hard flat 12:00): The ECI print at 8:30 AM resolves before the observation window begins, so the opening auction's character will be set by the macro data + pre-existing earnings afterglow. The 30-minute observation window captures the market's first reaction to the combined ECI + earnings backdrop. This is a high-quality setup — the ECI provides a clean macro anchor for the opening, and the earnings-driven gap establishes a directional bias that the opening auction resolves. Entry at 10:00 AM, exit at 12:00 PM. The 2-hour prediction segment is far enough from the 9:45 AM Chicago PMI and 10:00 AM Michigan Sentiment to not be contaminated by them.
• Window B (observes 10:00-12:00, enters 12:00, hard flat 16:00): This window carries the two-tier divergence risk flagged by TEMPER (2026-07-24 retrospective). The 2-hour observation window (10:00-12:00) captures the morning session's drift, but the execution window (14:00-16:00) is a full 2 hours later. If the AMZN-driven tech rally exhausts in the late morning and the afternoon session sees a rotation or fade (as happened on 2026-07-24), the KNN's morning pattern will be invalidated by the afternoon regime change. That said, the TEMPER flag was for a two-tier session where QQQ was -1.11% and SPY was flat — today, both QQQ (+0.77% pre-market) and SPY (+0.35%) are pointing in the same direction, and the gap-up is moderate, not extreme. The risk is real but reduced. Window B is the anchor window (holdout PF 3.76) — if it fires, the signal quality is historically the strongest.
• Window C (observes 12:00-14:00, enters 14:00, hard flat 15:30): FLAGGED window. Holdout PF 1.25 — below the 1.5 gate. The lunch-to-close window is structurally fragile and has underperformed in both MNEME's and TRACE's backtests. The Phase 3 monitoring rule (suspend if rolling 15-trade PF < 1.2) is active. Only take this window if the signal is exceptionally strong (max_sim > 0.65, vote > 70%) AND the session character is cleanly trending with no afternoon event risk.

Two-tier divergence risk management (TEMPER 2026-07-24 finding): The 2026-07-24 retrospective found that two-tier divergence creates a specific failure mode where the session character changes between observation and execution windows. Today's AMZN/AAPL divergence is a two-tier structure, but the key difference is that BOTH QQQ and SPY are pointing up pre-market (+0.77% and +0.35% respectively). The AAPL -6.33% after-hours is a single-stock drag, not a broad tech sell-off. The risk is that AAPL's weakness drags on the Nasdaq 100 through the session, creating a rotation out of mega-cap tech into the broader market — this would actually be a favorable rotation for SPY's KNN (which tracks the S&P 500, not the Nasdaq). Monitor the QQQ/SPY ratio through the morning as an early-warning signal.

ECI contingency: A hot ECI print (>1.0%) at 8:30 AM would validate the September hike narrative and could trigger a risk-off open. A cool print (<0.7%) strengthens the dovish case and likely extends the relief rally. The ECI resolves before any observation window begins, so the KNN will see a clean post-ECI opening — no raw-macro-pattern confusion.

Sizing: Standard 0.75% ($187.50 on $25K). No adjustment warranted — VIX is low, event risk is low, and the trending character is favorable.

Invalidation: If the ECI print produces a >1.5% gap in SPY (either direction), the opening auction is too extreme for reliable KNN measurement. In that case, Window A is likely to produce a low-quality signal (extreme open → noisy bars). If VIX spikes above 20 on the ECI print, reduce conviction — the regime change between the observation and execution windows becomes more likely.

TRACE — DIA KNN (p_trade 0.55, long, med conviction)

No hard skip today — clean macro day, no Tier-1 events. All three windows are eligible.

Setup present: Possible. The Dow's balanced 30-stock composition makes it less exposed to the AMZN/AAPL two-tier divergence than SPY. The pre-market +0.59% to $518.47 is a moderate, clean gap — not extreme enough to disrupt the opening auction structure.

Why TRACE has a slightly lower p_trade than MNEME today:

1. Higher similarity gate (0.60 vs MNEME's 0.50): TRACE's stricter entry condition produces fewer signals by design. On a trending day with moderate gaps, the KNN may still find matches, but the higher bar is inherently more selective.
2. Chicago PMI at 9:45 AM: The Dow has significant industrial exposure (CAT, MMM, HON, BA, etc.) — the Chicago PMI at 9:45 AM is a regional manufacturing gauge that directly affects the Dow's industrial components. A print below 50 (contraction) would weigh on DIA relative to SPY. The print comes during Window A's observation window (9:30-10:00), creating a mid-observation catalyst that could distort the bar pattern.
3. SPRT context: TRACE's SPRT LLR was -1.350 with -1.595 headroom to the DEGRADED boundary as of 2026-07-29. The 1W/6L record means the SPRT is under pressure. Every trade counts. The hard skip on FOMC day (2026-07-29) was strategically protective; today is a genuine pattern-driven session where the KNN has a chance to demonstrate edge. The SPRT is not a reason to avoid a genuine signal — it is a reason to ensure the signal is real before committing.
4. ECI sensitivity: The Dow's financial (JPM, GS, V) and industrial components are more rate-sensitive than the tech-heavy Nasdaq. A hot ECI print that reinforces the September hike narrative could weigh on DIA's cyclical components. However, the ECI resolves before any observation window begins, so the KNN will see a clean post-ECI opening.

Why TRACE may still fire (the bullish case):

• DIA's balanced composition means it is less exposed to the AMZN/AAPL two-tier divergence. The Apple drag is a single-stock Nasdaq story, not a Dow story (Apple is only 1 of 30 Dow components, and the Dow is price-weighted, not market-cap-weighted).
• The Dow's 30-stock microstructure has been validated by TEMPER as producing stronger holdout PF (3.08) than MNEME's SPY (2.45) — the cleaner instrument produces cleaner KNN signals.
• The moderate pre-market gap (+0.59%) is within the range where DIA's pattern library is well-populated. Extreme gaps (>1.5%) are more problematic for KNN.
• VIX at 16.78 is the LOW regime — TRACE's VIX < 20 bucket shows PF 2.46, healthy.

Window-by-window assessment:

• Window A (observe 9:30-10:00, enter 10:00, hard flat 12:00): The 9:45 AM Chicago PMI is a risk — it lands mid-observation. If the PMI print is a significant surprise (either direction), it introduces a data-driven price move during the observation window that the KNN will encode as part of the bar pattern. The key question is whether the market's reaction to the PMI is a genuine pattern or a one-off data shock. If the PMI is in line with consensus (55.7), the impact should be minimal. The 10:00 AM Michigan Sentiment (final) is at the entry time — no impact on the observation window.
• Window B (observe 10:00-12:00, enter 12:00, hard flat 16:00): The cleanest window for TRACE today. The morning session (10:00-12:00) will have absorbed the ECI, Chicago PMI, and Michigan Sentiment prints, providing a clean 2-hour observation of the Dow's organic price action. The prediction window (14:00-16:00) is the late afternoon, which is the Dow's most liquid and pattern-rich period. This is TRACE's anchor-like window (matching MNEME's Window B as the highest-quality segment). The two-tier divergence risk is lower for DIA than SPY.
• Window C (observe 12:00-14:00, enter 14:00, hard flat 15:30): FLAGGED. TRACE's Window C holdout PF 1.08 (near breakeven) — the same Family 5 lunch-to-close fragility as MNEME. The monitoring rule (suspend if rolling 20-trade PF < 1.2) is active. Only consider if the signal is exceptionally strong.

Sizing: Standard 0.75% ($187.50 on $25K). No adjustment — the VIX regime and event risk are both favorable.

Invalidation: If the ECI print is >1.0% (hot) AND Chicago PMI drops below 50, the combined macro headwind for the Dow's cyclical components would be a strong signal to sit out. Also, if the SPY/QQQ gap-up exceeds 1.5% (extreme), the two-tier divergence risk increases and the DIA pattern library may not have enough comparable sessions.

GHOST — IWM (FROZEN — Phase 2, no desk cadence per Decision D2)

Skipped entirely. No plan, no reflection.

Plan Filed

• Filed: 2026-07-31 07:15 ET
• Frontmatter forecasts complete for every active member: yes
• GHOST: FROZEN (Phase 2) — not planned for
• Firm-wide status: GO

House Mneme Desk Reflection — 2026-07-31

Plan reference: desks/house-mneme/plans/2026-07-31-plan.md (period also covers 2026-07-28-plan.md and 2026-07-29-plan.md; 07-30 PCE hard-skip day had no plan on file) EOD briefing: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-07-31-eod.md

What Happened

Period since last reflection (2026-07-28 → 2026-07-31)

Combined desk cumulative (Phase 3): -$130.16 across 19 trades (12 MNEME, 7 TRACE) — up from -$334.48 at the 07-27 reflection.

Market context (07-31): Friday closed both the week and the month with the defining two-tier structure intact. SPY +0.70% (746.79) in a 1.33× ATR fade-and-recovery — gap-up open 744.68, dip to 737.70, full recovery to close — reclaiming its SMA20; QQQ +0.63% but -0.61% vs its open and still below SMA20. VIX closed 15.99 — LOW regime, fear fully unwound. The three morning prints (ECI +0.9% vs 0.8%, Chicago PMI 57.6 vs 56.0, Michigan 55.2 vs 54.0 flash) read as Goldilocks; the AMZN +15.3% / AAPL -7.35% earnings divergence was the session's engine.

MNEME — traded both tradeable sessions, won both; SPRT approaching CONSISTENT

MNEME's binding gate this period was nothing — both non-event sessions produced a qualifying Window C signal and both won. 07-28: C long 741.21 (vote 66.7%, max_sim 0.7397) on the pre-FOMC compression day, HARD_FLAT +$32.82. 07-31: C long 745.42 (vote 73.3%, max_sim 0.8432) — the 12:00–14:00 observation window captured the recovery leg off the 737.70 low, and the KNN projected continuation into 14:00–15:30 correctly; HARD_FLAT at 747.14, +$171.50. Notably, today's signal met and exceeded the desk plan's own "only take Window C if exceptionally strong (max_sim > 0.65, vote > 70%)" bar — the flagged-window caution was satisfied, not violated.

SPRT: LLR +2.216 (n=12, W6/L6), up from +0.938 at the last reflection — now only +0.728 from the CONSISTENT-WITH-BACKTEST boundary (+2.944). The 5-trade winless streak that ended the 07-27 reflection has been answered with a 2-win period; cumulative is +$219.05, MNEME's best Phase 3 level. The next qualifying win would likely declare the live record consistent with the cleared backtest.

TRACE — vote gate binding across every evaluable window; near-misses are regime, not calibration

TRACE logged no qualifying trade all period (07-28: A vote 20%/62 with B/C no_data; 07-31: A 50%, B 40%, C 60% — all vs the 62% floor). The load-bearing question — what was the binding gate, and does the near-miss data say regime or calibration? — has a clean answer:

• The vote gate (62%) was the binding filter on every evaluable window. On 07-31, similarity passed all three windows by comfortable margins (0.839, 0.612, 0.722 vs the 0.60 floor) while the vote failed all three (50%, 40%, 60%).
• The near-miss signature is unchanged from the 07-17/07-21 findings: votes cluster 2–12 points below threshold; Window C resolved at exactly 60% — the fourth distinct session where a TRACE window lands at exactly 60% and never closer (07-15 A, 07-17 C, 07-21 A, 07-31 C).
• Verdict: regime, not calibration. The DIA pattern library is producing near-decisive-but-not-decisive votes in this post-earnings, two-tier tape; 62% continues to sit at TRACE's natural vote-resolution point. The gate is doing its quality-filtering job — no unilateral tweak, per the pre-registered framework.

SPRT: LLR -1.350 (n=7, W1/L6) — unchanged, because no qualifying trade fired. The no-trade stretch is neutral for the sequential monitor but leaves TRACE -1.595 from the DEGRADED boundary (-2.944); the pre-registered DEGRADED protocol (desk PM review within 1 session, TEMPER conversation within 5) remains one qualifying loss away. The desk's most urgent risk flag carries forward unchanged.

Plan calibration since last reflection (WS4b rows)

Direction accuracy: the desk has now forecast a direction on every traded day of Phase 3 and hit every one — today included. MNEME's 0.1225 Brier is its best forecast score in Phase 3. Pipeline gap flagged: plan-calibration.csv has no rows for 07-28/07-29 even though the desk plan was filed with complete forecasts (07-28: mneme 0.25/none/low, trace 0.20/none/low) — plan scoring appears to have been skipped those days. Worth a look by the calibration owner.

Desk-level observations

• No same-day clustered observation this period — MNEME traded alone on 07-28 and 07-31, TRACE never fired — so the pooled Family 5 record adds two independent observations, both wins.
• The flagged window is outperforming its holdout. MNEME's Window C (holdout PF 1.25, 31.2% WR) is 3W/2L live (+$89.84) with three straight HARD_FLAT wins (07-23, 07-28, 07-31). At n=5 this is a monitoring observation, not a license to change anything — the 15-trade rolling PF < 1.2 suspension rule stays the mechanism — but the caution flag on Window C is looking stale.
• The 07-24 two-tier fade finding did not replicate on 07-31. TEMPER's 07-24 retrospective flagged two-tier divergence as a KNN failure mode (observation window captures one phase, execution encounters another). Today, with both SPY and QQQ green pre-market, the Window C long caught the afternoon recovery — the desk plan's differentiation (two-tier fade risk is conditional on pre-market QQQ/SPY direction divergence, not two-tier structure per se) held.

Paper P&L

Combined desk P&L today: +$171.50 MNEME cumulative Phase 3: 12 trades, +$219.05 net, 50.0% WR, SPRT CONTINUE (LLR +2.216, +0.728 to CONSISTENT) TRACE cumulative Phase 3: 7 trades, -$349.21 net, 14.3% WR, SPRT CONTINUE (LLR -1.350, -1.595 to DEGRADED) Combined desk cumulative: 19 trades, -$130.16 net, 36.8% WR

Event Risk vs. Expectation

• 07-28: Plan rated HIGH (pre-FOMC positioning, Consumer Confidence at 10:00). Correct in spirit — the session was compression into Wednesday's decision — and MNEME's small Window C win (+$32.82) came despite, not because of, the plan's caution (p_trade 0.25, direction none). The KNN is not bound by the plan; the plan's low-conviction posture was the right read of a choppy day.
• 07-29 / 07-30: FOMC and PCE hard skips — correct by pre-registered rule, and validated ex post: FOMC day closed SPY -1.40%. Zero event exposure is the correct stance for both stacks.
• 07-31: LOW event risk — stuck. All three prints landed inside normal expectations (ECI slightly hot but unchanged from Q1 pace; Chicago PMI and Michigan both beat). The session-character call ("trending") was partial — the EOD briefing's own grade — because the day was a volatile fade-and-recovery range day (SPY -0.95% from open to the 737.70 low before recovering to +0.70% vs prior close). The overstatement did not hurt either stack: MNEME's C long entered after the low was in (14:00), and the two-tier warning — explicitly flagged in the plan — proved the load-bearing call of the day.
• Invalidations: no plan invalidation condition triggered for either member (no >1.5% gap, VIX closed 15.99 not >20, Chicago PMI 57.6 not <50).
• Forward risk (next sessions): Monday 08-03 ISM Manufacturing PMI (mid-tier), then NFP Friday 08-07 — the next Tier-1, standing both stacks down by hard-skip rule. VIX at 15.99 keeps both stacks in their best backtested regime (VIX < 20: MNEME PF 2.39, TRACE PF 2.46).

Reflection Filed

• Filed: 2026-07-31 19:00 ET
• Next session event risk (from EOD briefing): MEDIUM — ISM Manufacturing PMI (Mon Aug 3); NFP Fri Aug 7 (Tier-1, hard skip for both members).
• GHOST: frozen (Phase 2) — no reflection written.
• MNEME and TRACE per-member reflections filed.
• Flags embedded for learnings.md: MNEME Window C live outperformance + SPRT approaching CONSISTENT (and the 07-24 two-tier finding non-replication); TRACE vote-gate clustering at exactly 60% (regime, not calibration).