House Mneme — 2026-07-27

desk: house-mneme date: 2026-07-27 forecasts: mneme: p_trade: 0.40 direction: long conviction: low trace: p_trade: 0.25 direction: long conviction: low

Desk Plan — House Mneme — 2026-07-27

Briefing reference: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-07-27-pm.md PM: MNEME

Shared Market Read

• Event risk today: HIGH — FOMC Wednesday, PCE Thursday, but today itself has Durable Goods Orders (8:30 AM, HIGH impact). The US-Iran ceasefire and oil collapse (-9.6% Brent) dominate pre-market narrative.
• Session character expected: News-driven / recovery rally gap-up. Futures sharply higher (SPY +0.88%, NQ +1.44%, DOW +0.78%). The pre-market gap is driven by a macro regime shift (ceasefire, oil unwinding), not a clean post-data resolution. Durable Goods at 8:30 AM adds a second catalyst layer — the 8:30 print determines whether the gap holds or fades. Expect high first-hour volatility and potential fade into the afternoon as pre-FOMC positioning begins.
• VIX regime: 17.57 (LOW — <18). Both MNEME and TRACE are profitable in this regime (MNEME VIX<20 PF 2.39; TRACE VIX<20 PF 2.46). The fear premium from the July oil spike has unwound.
• Key levels: SPY prior close 738.90, pre-market gap-up implied, SMA20 746.15 (still below SMA20 — structural pullback). DIA prior close ~520 (approximate — Dow futures +0.78%). The two-tier divergence (QQQ underperforming SPY) may narrow today as semis bounce on the ceasefire relief rally, but the FOMC/hyperscaler capex overhead limits conviction.
• Macro backdrop: The US-Iran ceasefire transforms the week's macro landscape. Brent crude collapsed 9.6% to $89 — the oil-risk premium that drove the July selloff is unwinding. The Dow's industrial/transport tilt is the structural beneficiary. The FOMC decision Wednesday and mega-cap earnings (Meta, MSFT, AMZN, AAPL) in the same 72-hour window make this the most consequential week of the year.

MNEME — SPY (K=15, max_sim≥0.50, vote≥62%, long-only)

Live record (Phase 3): 9 trades, 44.4% WR (4W/5L), net P&L +$123.81. Last trade (2026-07-24): Window B, STOP at -$143.30 — the TEMPER-flagged two-tier divergence failure mode (morning pattern identified correctly, afternoon rotation reversed it). SPRT likely below the prior +1.477 after the 2026-07-24 loss.

Setup assessment: Below-expected probability. The gap-up recovery rally is a structurally challenging environment for KNN similarity:

• Gap-up opens distort the first 30 minutes (Window A's observation window) with gap-fill mechanics rather than pattern-repeating behavior. The 6-bar observation window may not capture reliable microstructure.
• Today is a narrative-driver session (macro regime shift, ceasefire, oil collapse) — MNEME's 2026-07-17 reflection and the TEMPER 2026-07-24 finding both show narrative-driver sessions underperform clean post-data sessions for KNN.
• The TEMPER 2026-07-24 finding specifically identified two-tier divergence as a Window B failure mode — the observation window captures a pattern that the execution window's rotation invalidates. Today's oil-driven rotation (energy collapsing, everything else rallying) is a new two-tier divergence that may produce the same failure mode.
• Durable Goods at 8:30 AM is a positive catalyst if the data is clean — a beat on the -0.5% consensus would reinforce the recovery narrative and could produce the cleanest pattern of the day between 10:00-12:00.

Key conditions for a signal:

• Window A (observes 9:30-10:00, enters 10:00): Requires the gap-up to settle into a contained pattern by 10:00, not a continuing trending gap. The pre-market gap structure (futures +0.88%) means the first 6 bars may be dominated by the gap-fill. Low confidence — only 6 bars of observation in a gap-up session is insufficient for pattern quality.
• Window B (observes 10:00-12:00, enters 12:00): The anchor window (holdout PF 3.76). This is the most likely path to a trade today. The Durable Goods print at 8:30 AM will be fully absorbed by 10:00. If the 10:00-12:00 window shows a clean, contained pattern (not trending), this is the strongest candidate. The TEMPER two-tier divergence finding is the key risk: if the oil-driven rotation continues through the afternoon, the observation pattern may not hold.
• Window C (observes 12:00-14:00, enters 14:00): Flagged (holdout PF 1.25). Pre-registered suspension at 15-trade rolling PF < 1.2 — not triggered. The afternoon pre-FOMC positioning (2 days away) adds noise. Low confidence.

Sizing: Standard — 0.75% ($187.50), 1.5× ATR(14) stop, single position per day.

Invalidation: SPY gaps >1.0% at open with continuing trend through 10:00 → trending gap, not pattern-repeating. VIX spikes above 20 on Durable Goods miss → elevated volatility degrades pattern quality. A Durable Goods miss (below -0.5% consensus) that reverses the pre-market rally → stand down entirely — the session character flips from recovery to risk-off.

TRACE — DIA (K=10, max_sim≥0.60, vote≥62%, long-only)

Live record (Phase 3): 7 trades, 14.3% WR (1W/6L), net P&L -$207.53. SPRT approaching DEGRADED boundary (LLR -0.996 as of 2026-07-24, boundary at -2.944). Last trade (2026-07-23): Window A STOP at -$98.60. 2026-07-24: no_signal (no_data across all three windows — DIA bar fetch issue). The live track record is significantly below backtest expectations.

Setup assessment: Unlikely — structurally lower probability than MNEME. TRACE's tighter gates (max_sim≥0.60, K=10) produce fewer signals in any environment, and the live track record is trending toward the SPRT DEGRADED boundary. However, today's session character is MORE favorable for DIA than SPY:

• The Dow's 30-stock blue-chip composition (heavily weighted toward industrials, financials, healthcare) is the structural beneficiary of the oil collapse. Lower input costs directly benefit transports, industrials, and airlines — all Dow components. The Dow's beta (~0.70) means it's less exposed to the tech overhang that drove the two-tier divergence.
• The rotation from energy (XLE) into transports/industrials provides a cleaner directional backdrop for DIA's 30-stock microstructure than SPY's 500-stock composition.
• The caveat: TRACE's KNN signal quality does not depend on DIA's sector composition — it depends on whether the observation window's bar-by-bar fingerprint matches a historical pattern. A favorable macro backdrop does not guarantee a KNN signal.

Key conditions for a signal:

• Window A (observes 9:30-10:00, enters 10:00): DIA's pre-market level implied by Dow futures +0.78%. The 6-bar observation window is particularly vulnerable to gap-up distortion. Requires a clean post-gap-settle 30-minute pattern. Low confidence given TRACE's live record.
• Window B (observes 10:00-12:00, enters 12:00): The anchor window (holdout PF 3.00, 18 trades). The 24-bar, 72-dim observation provides the richest feature space. The post-Durable-Goods settlement (10:00-12:00) is the most likely window for a clean pattern. If DIA holds its gains through the morning without a violent reversal, this is the strongest candidate.
• Window C (observes 12:00-14:00, enters 14:00): Pre-registered for suspension at 20-trade rolling PF < 1.2 (holdout PF 1.13, 18 trades). Not triggered. Low confidence — only trades if it produces the strongest single-day vote.

Sizing: Standard — 0.75% ($187.50), 1.5× ATR(14) stop, single position per day.

Invalidation: DIA gaps >1.0% at open with unidirectional trend through 10:00. VIX spikes above 20 on Durable Goods miss. A Durable Goods miss that reverses the pre-market rally into a risk-off session. If the SPRT LLR crosses -2.944 (DEGRADED boundary) during today's session, stand down — the pre-registered monitoring rule fires.

Plan Filed

• Filed: 2026-07-27 07:30 ET
• Frontmatter forecasts complete for every active member: yes

House Mneme Desk Reflection — 2026-07-27

Plan reference: desks/house-mneme/plans/2026-07-27-plan.md EOD briefing: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-07-27-eod.md

What Happened

Family 5 took two trades today — both long, both stopped out, combined -$251.36. A perfect directional call (both plan forecasts were "long," both traded long) that produced a net loss because the open-high-fade session structure is structurally hostile to mid-session long entries.

Market context: Textbook open-high-fade / gap-up-reversal. SPY opened +0.8% above Friday's close (744.90) and closed flat at 738.85 (-0.01%). QQQ opened +1.1% (691.78) and closed -0.32% (682.13). Dow held (+0.51%) — the two-tier divergence widened, not narrowed. Range > 1.2× ATR for both indices. The catalysts that produced the gap (weekend US-Iran ceasefire, oil collapse at Brent -8.7%) were real and macro-significant, but the Durable Goods miss (+0.3% vs +1.6% consensus) and the weekend NVDA OpenAI financing story (WSJ, $250B, -4.5% NVDA) turned the session into a fading grind. VIX at 18.67 crossed back into MID regime (18-28) from Friday's LOW reading, confirming the volatility floor has lifted ahead of the Wed-Thu FOMC/PCE double-header.

Period Since Last Reflection (2026-07-24 → 2026-07-27)

Combined desk cumulative (Phase 3): -$334.48 across 17 trades (10 MNEME, 7 TRACE).

Member Analysis

MNEME (SPY, Window C — flagged window, STOP loss)

Window C (Lunch→Close, flagged holdout PF 1.25) fired at 14:00 with a KNN long signal. Entered SPY at 737.94. The afternoon continued to fade from the pre-market gap high — SPY opened at 744.90, drifted lower all day, and MNEME's stop at 736.86 was triggered at 14:35. Net P&L: -$109.08.

The plan correctly assessed MNEME at p_trade 0.40 (reasonable — the gap-up and Durable Goods miss made a setup plausible) and direction long (correct — the KNN found a long pattern). The conviction call (low) was justified. But the specific window that fired — Window C, the flagged window with holdout PF 1.25 — is the desk's most structurally vulnerable entry point. The morning plan noted: "Window C (observes 12:00-14:00, enters 14:00): Flagged (holdout PF 1.25). The afternoon pre-FOMC positioning adds noise. Low confidence." This assessment was accurate but did not prevent the trade from firing — the KNN's strongest single-day vote was Window C, overriding the desk PM's qualitative concern. This is correct by design: the desk plan is non-binding, and the KNN's signal logic is independent of the plan's posture. But it surfaces a tension: when the flagged window produces the strongest signal on a structurally hostile session, should the desk PM have a mechanism to override? The pre-registered 15-trade rolling PF < 1.2 suspension rule for Window C is that mechanism — and it has not triggered (MNEME's total Phase 3 trades are only 10, well below the 15-trade window). When Window C fires, losses like today's are the expected outcome at 31.2% holdout WR.

The loss is a statistical outcome of taking a low-WR window's signal, not a malfunction: Window C's holdout WR is 31.2% (PF 1.25) — 1 loss at N=10 does not degrade the cumulative evidence.

SPRT trajectory: LLR dropped from +1.208 (post-07-24) to +0.938 (post-07-27). Still comfortably in CONTINUE territory (+2.006 to CONSISTENT, -3.882 to DEGRADED). The loss erodes the buffer but does not change the structural health assessment.

The key finding: MNEME's most recent two trades (07-24 Window B anchor loss, 07-27 Window C flagged loss) have erased the +$267.11 cumulative peak from 07-23. The cumulative P&L is now +$14.73 across 10 trades. The 5-trade winless streak (last win: 07-21, Window B HARD_FLAT +$84.09) is the longest of Phase 3.

TRACE (DIA, Window A — opening→morning window, STOP loss)

Window A (Opening→Morning, 6-bar observation window) fired at 10:00 with a KNN long signal. Entered DIA at 523.31. The Dow opened strong on the ceasefire/oil collapse relief and gapped up alongside the broader market. The stop at 522.22 was triggered at 10:25 — a 25-minute stay. Net P&L: -$142.28.

The plan assessed TRACE at p_trade 0.25 with conviction low — the lowest p_trade the desk has given TRACE on a trade day. The plan explicitly warned: "Window A (observes 9:30-10:00, enters 10:00): DIA's pre-market level implied by Dow futures +0.78%. The 6-bar observation window is particularly vulnerable to gap-up distortion." This was prescient. The gap-up open dominated the 6-bar observation window with gap-fill mechanics rather than pattern-repeating behavior. The KNN found a long pattern in the first 30 minutes of the Dow's post-gap-up action, entered at 10:00, and was immediately stopped out as the gap-fade continued. The 10:00-12:00 execution window saw DIA give back the gap-over-open gains.

This is the structural vulnerability of Window A in gap-up sessions that the plan correctly identified. TRACE's tighter gates (max_sim ≥ 0.60, K=10) are supposed to protect against this — the max_sim threshold being higher means only more confident patterns should fire. But on a gap-up day, even a genuine historical analog of a gap-up opening that continued into the morning would appear valid in the 6-bar window — the pattern's problem is the limited observation period, not the similarity threshold.

SPRT trajectory: LLR dropped from -0.996 to -1.350 — now only -1.595 from DEGRADED at -2.944. This is the desk's most urgent risk flag. TRACE has 7 trades in Phase 3 (1W/6L). At the current trajectory, 1-2 more losing qualifying trades could trigger the DEGRADED protocol (desk PM review within 1 session, TEMPER conversation within 5 sessions).

Paper P&L

Combined desk P&L today: -$251.36 MNEME cumulative Phase 3: 10 trades, +$14.73 net, 40.0% WR, SPRT CONTINUE (LLR +0.938) TRACE cumulative Phase 3: 7 trades, -$349.21 net, 14.3% WR, SPRT CONTINUE (LLR -1.350) Combined desk cumulative: 17 trades, -$334.48 net, 29.4% WR

Event Risk vs. Expectation

The desk plan rated today as HIGH event risk (Durable Goods Orders at 8:30 AM, plus the weekend ceasefire/oil collapse macro shift). This was correct — the Durable Goods miss (+0.3% vs +1.6% consensus) was a significant miss that contributed to the gap-up fade. The plan also noted the two-tier divergence might narrow today as semis bounced — in fact it widened, as the NVDA/OpenAI story was absent from the plan's assumptions (it dropped over the weekend via WSJ and was not in the morning briefing's narrative).

The plan's invalidation conditions were:

• MNEME: "SPY gaps >1.0% at open with continuing trend through 10:00" — not triggered (gap was +0.8%, not >1.0%). "VIX spikes above 20 on Durable Goods miss" — not triggered (VIX 18.67, stayed mid). "A Durable Goods miss that reverses the pre-market rally" — this did happen, but the plan's language was about standing down entirely, not about blocking a specific entry window. The miss contributed to the fade but didn't create a risk-off session (Dow still +0.51%).
• TRACE: Similar conditions, plus "SPRT LLR crossing -2.944" — not triggered (still at -1.350). The DEGRADED boundary was not crossed.

Plan Calibration (2026-07-27)

Both direction forecasts hit. The p_trade calibration for MNEME (0.40) was slightly underconfident on the trade outcome (Brier 0.36 — reasonable for a low-conviction forecast). TRACE's p_trade 0.25 on a trade day produces a higher Brier (0.5625) — the plan was correctly calibrated for the plan's own assessment of likelihood, but the trade fired anyway. This is the correct relationship: the plan's low probability does not bind the KNN from finding a pattern.

Directional accuracy continues: 6/6 sessions on direction calls across the last 3 trading days (07-21, 07-23, 07-24, 07-27 — all long→hit). The desk plan has never forecast a direction on a traded day that missed. This is notable enough to track: the desk PM's directional instinct is running well ahead of p_trade calibration.

Desk-Level Learnings

• Both KNN stacks lost on long entries in an open-high-fade session — the session type that is KNN's worst structural mismatch. MNEME's Window C (flagged, 31.2% holdout WR) and TRACE's Window A (6-bar gap-up vulnerability) both fired on days where the morning plan explicitly warned that gap-up sessions degrade pattern quality. The plan's qualitative assessment was accurate; the KNN's signal logic overrode it for both members. The loss confirms the structural finding from TEMPER's 2026-07-24 retrospective: "The historical pattern library (2020-2024) may not have enough structurally-divergent sector-rotation sessions to correctly weight this fade risk."
• TRACE's SPRT trajectory is the desk's most urgent risk flag. LLR -1.350, only -1.595 from DEGRADED at -2.944. TRACE has 7 trades in Phase 3 with 1W/6L. The pre-registered DEGRADED protocol (desk PM review within 1 session, TEMPER conversation within 5) becomes a live possibility on the next qualifying loss. The Tuesday Consumer Confidence session (the last tradeable day before the FOMC/PCE blackout) is the next opportunity — if TRACE fires and loses, the SPRT will likely breach -2.944. [Flag for learnings.md]: TRACE's SPRT at LLR -1.350 after 7 trades (1W/6L). Current trajectory puts the DEGRADED boundary 1-2 qualifying losses away. The 14.3% live WR is below the p1=0.204 pre-registered lower bound — meaning the evidence is accumulating correctly, not malfunctioning. If DEGRADED fires, the protocol mandates a desk PM review (1 session) and TEMPER conversation (5 sessions). No pre-emptive action recommended — the pre-registered protocol is the correct response.
• MNEME's cumulative P&L is nearly back to breakeven (+$14.73 on 10 trades). The two consecutive losses (07-24 Window B -$143.30, 07-27 Window C -$109.08) have erased the +$267.11 peak from 07-23. At 40.0% WR with avg win ~$91 and avg loss ~$101, the P&L-neutral position is fragile. The SPRT remains healthy (LLR +0.938) — the sequential monitor does not see the current trajectory as evidence of edge degradation. The key question: is this run of losses random variance (consistent with a 44% backtest WR — the expectation of 3-4 consecutive losses in 10 trades is non-trivial), or is the current session regime structurally degrading KNN pattern quality? The answer at N=10 is: cannot distinguish — let the SPRT run.
• FOMC+PCE blackout on Wed-Thu (July 29-30). Both members have hard skip rules. Tuesday (Consumer Confidence at 10:00 AM) is the only tradeable session before the blackout. The Friday Michigan Sentiment crash to 49.5 suggests Consumer Confidence could be a significant catalyst. If the data continues to show consumer weakness, the two-tier divergence (Dow holding via oil/industrial rotation, everything else fading) could persist or accelerate.
• Same-day trades across MNEME and TRACE count as one clustered observation per the Family 5 pooled validation rule (TEMPER's ECHO v5 correlated-session finding). Today is the 4th same-day clustered observation (pooled total: 07-23 combined stop loss, 07-24 single trade, 07-27 combined stop loss). The clustered observation rule means the combined 17-trade desk record should be treated as ~14 independent observations for significance purposes.

Plan Accuracy

Overall plan accuracy: 3.5 / 5

Reflection Filed

• Filed: 2026-07-27 18:00 ET
• Next session event risk (from EOD briefing): MEDIUM — Consumer Confidence (Tue Jul 28, 10:00 AM). Then FOMC (Wed, hard skip) and PCE (Thu, hard skip). Tuesday is the last tradeable session before the blackout. TRACE's SPRT at -1.350 makes every qualifying trade high-stakes.
• GHOST: frozen (Phase 2) — no reflection written.
• MNEME and TRACE per-member reflections filed.