House Mneme — 2026-07-16

desk: house-mneme date: 2026-07-16 forecasts: mneme: p_trade: 0.40 direction: long conviction: med trace: p_trade: 0.30 direction: long conviction: low

Desk Plan — House Mneme — 2026-07-16

Briefing reference: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-07-16-pm.md PM: MNEME

Shared Market Read

• Event risk today: MEDIUM — Retail Sales (June) at 8:30 AM ET concludes the week's macro trifecta; TSMC earnings before open; Netflix after close. No Tier-1 hard skip (FOMC/CPI/NFP/PCE) — both stacks trade through MEDIUM event risk.
• Session character expected: News-driven / potentially trending — the 8:30 AM data block (Retail Sales + Philly Fed + Jobless Claims + Import Prices) drives first-hour volatility. By 10:00 ET (Window A entry) the data is digested and a clean trend may be established. Post-data direction likely sets the tone for the rest of the session.
• VIX regime: 16.00 (LOW regime). Both stacks are profitable in LOW VIX (MNEME VIX < 20: PF 2.39, 197 trades; TRACE holdout at VIX < 20 not yet disaggregated but DIA's lower-beta structure suggests vol resilience). LOW VIX is MNEME's weakest regime by PF (2.39 vs. 3.54 in VIX 20-30) but still profitable with meaningful sample size.
• Key levels: SPY prior close 754.77, SMA20 at 744.85 (+10pts above). DIA has no direct level in the briefing — infer Dow 30-stock structure is less impacted by the tech rotation. The SPY/QQQ divergence (SPY above SMA20, QQQ below) is the defining structural feature — SPY is the constructive side of that divergence.
• Macro backdrop: Disinflation narrative confirmed by CPI and PPI earlier this week. Retail Sales is the demand-side test — a soft print (confirming the consumer is moderating) extends the soft-landing trade; a hot print introduces oil-driven inflation concerns. SPY is positioned to benefit from the former; DIA's industrial/cyclical tilt is a secondary beneficiary.

MNEME — SPY — KNN Similarity-Based (Long Only)

Setup present: Likely. SPY is structurally constructive (above SMA20, +0.19% futures), and the 8:30 AM data block resolves before Window A's 10:00 ET entry — meaning the KNN similarity engine queries a clean post-data observation window, not a noise-distorted one. Low VIX (16.00) is MNEME's weakest regime but still profitable (PF 2.39 on 197 trades). The SPY/QQQ divergence is structurally favorable for SPY-long strategies.

Key considerations for today:

• Window A (9:30-10:00 observation, entry 10:00) — the 8:30 AM data prints are fully absorbed by the opening auction. The first 30 minutes will reflect the market's settled read on Retail Sales + TSMC earnings. Clean post-data observation bars are a positive for the KNN engine, which compares against historical patterns that mostly lack an 8:30 AM catalyst.
• Window B (10:00-12:00 observation, entry 12:00) — the richest observation window and MNEME's anchor (holdout PF 3.76). The full morning session post-data is the most information-rich feature space. Today's macro backdrop (disinflation narrative + SPY above SMA20) aligns with the constructive bias the window historically predicts well.
• Window C (12:00-14:00 observation, entry 14:00) — flagged (holdout PF 1.25). No reason to increase or decrease its allocation today; the standard monitoring rule applies.
• Sizing: Standard 0.75% per trade ($187.50 on $25K). No adjustment for MEDIUM event risk — the data resolves before entry windows.
• Invalidation: If Retail Sales surprises significantly (1.0%+ MoM vs. 0.2% consensus), SPY could gap-and-sell-off on hawkish consumer-demand read-through. The opening range would be volatile, potentially distorting the first 30-minute observation window. In that scenario, Window A patterns may not match historical norms — the KNN gate (max_sim ≥ 0.50, vote ≥ 62%) is the correct circuit-breaker. If the TSMC earnings selloff (-3.59% pre-market) extends to SPY and QQQ futures gap down >1%, the first hour could be dominated by forced selling rather than pattern structure — monitor for this before Window A's 10:00 entry.

TRACE — DIA — KNN Similarity-Based (Long Only)

Setup present: Possible, with lower conviction. DIA's 30-stock blue-chip composition is less exposed to the tech/chip rotation that is pressuring QQQ — the Dow's industrial, financial, and healthcare names provide a buffer. However, DIA's structural fire rate is lower than MNEME's (59 holdout trades vs. 98), and the tighter max_sim gate (0.60 vs. 0.50) means fewer sessions qualify.

Key considerations for today:

• The 8:30 AM data block affects DIA as an index-level instrument, but the Dow's sector composition (less tech, more industrials/financials) means the chip rotation tail-risk is muted. The consumer sentiment read from Retail Sales is directly relevant to DIA's consumer cyclicals.
• DIA's lower beta (~0.70 vs SPY's 1.0) means less sensitivity to the macro narrative — this is insulating today but also means DIA may not capture the full upside of a soft-retail-sales rally.
• Window B (entry 12:00) is the anchor for TRACE as well — the full morning observation window on DIA's cleaner microstructure is the strongest signal.
• Sizing: Standard 0.75% per trade. No adjustment.
• Invalidation: If the 8:30 AM data block produces a risk-off move (hot Retail Sales + weak Philly Fed), DIA's beta exposure means it will likely decline alongside SPY, though less severely. The KNN gates are the correct filter. The >0.60 max_sim threshold is a higher bar than MNEME's — if patterns are noisy today, TRACE will naturally sit out.

Plan Filed

• Filed: 2026-07-16 07:30 ET
• Frontmatter forecasts complete for every active member: yes

House Mneme Desk Reflection — 2026-07-16

Plan reference: desks/house-mneme/plans/2026-07-16-plan.md PM Briefing: dadbrain/Analysis/briefings/2026-07-16-pm.md

What Happened

Family 5 took two trades today — one per active member — and produced a net combined P&L of $+46.33. MNEME won on a clean Window A HARD_FLAT. TRACE lost on a Window B early stop-out. The combined desk result is positive, driven entirely by MNEME's win.

The session was the third and final act of the week's macro trifecta: Retail Sales (June) at 8:30 AM ET printed +0.2% MoM, in-line with consensus, confirming the disinflation narrative after CPI and PPI earlier this week. VIX opened at 16.00 (LOW regime). TSMC earnings before open were the secondary catalyst. The session character was subdued and constructive rather than directionally committed — SPY closed -0.26%.

MNEME (SPY, Window A — opening→morning): Entered long at 10:00 @ 751.78, exited HARD_FLAT at 12:00 @ 752.75. Net P&L: $+96.03. Strong similarity score (0.8483, 66.7% vote) on a clean post-data observation window. The KNN correctly identified a constructive pattern in the morning bars and the tape delivered. This is MNEME's largest single-trade win in Phase 3 (previous best: $+203.94 on 2026-07-10, Window B). Cumulative: 4 trades, 50% WR, $+197.01 net.

TRACE (DIA, Window B — morning→afternoon): Entered long at 12:00 @ 526.94, stopped out at 12:10 @ 526.59. Net P&L: $-49.70. The signal was marginal — max_similarity 0.6013, barely above the 0.60 gate, despite a decisive 80% vote. The Dow couldn't find traction in a subdued afternoon session. This is TRACE's third loss in four Phase 3 trades (25% WR, $+22.51 net). The cumulative P&L is still positive ($+22.51) but eroding.

Desk-level observation (Pooled Validation, Family 5): This is the second session where both members traded the same day (first was 2026-07-13). Per TEMPER's ECHO v5 correlated-session finding, same-day Family 5 trades count as one clustered observation. The clustered outcome today is net +$46.33 — a positive clustered observation after the first clustered observation on 2026-07-13 (net -$53.95). Combined desk cumulative: $+219.52 net across 8 real trades, with the two same-day clusters producing +$46.33 net (today) and -$53.95 net (2026-07-13). The clustered record is 1 win / 1 loss, positive net.

The key structural divergence between members today: MNEME fired a clean, high-similarity signal (0.8483, well above gate) and profited; TRACE fired a marginal signal (0.6013, barely above gate) and lost. Both members' KNN engines are the same mechanism on different instruments, but today MTNE's screen selected a genuinely high-quality pattern while TRACE's screen accepted a barely-qualifying one. This is the first live data point suggesting that max_sim distance above the gate may carry predictive information for Family 5 — something both members' backtests found to be near-zero. Worth monitoring at the 10-trade clustered checkpoint.

Paper P&L

Combined desk P&L today: $+46.33 MNEME cumulative Phase 3: 4 trades, $+197.01 net, 50% WR, SPRT CONTINUE (LLR +0.739) TRACE cumulative Phase 3: 4 trades, $+22.51 net, 25% WR, SPRT CONTINUE (LLR -0.289) Combined desk cumulative: 8 real trades, $+219.52 net, 37.5% WR

Event Risk vs. Expectation

The desk plan rated today as MEDIUM event risk (Retail Sales). Correct. The data block resolved without surprise and both post-data windows were tradeable. The plan's call that "the data resolves before entry windows" was accurate — 8:30 AM data allowed clean 9:30 AM observation bars for both members.

The TSMC earnings before open created the expected chip-sector overhang (Asian semis sold off overnight) but DIA's blue-chip composition insulated TRACE from the worst of it — confirmed by the plan's assessment. MNEME's SPY position benefited from the broad market's constructive disinflation narrative.

VIX at 16.00 (LOW) is MNEME's weakest regime (PF 2.39) — but MNEME still won today. TRACE's relationship with LOW VIX is not yet disaggregated (TRACE's holdout lacks a VIX regime breakdown by instrument), but today's loss in LOW VIX is directionally consistent with a KNN mechanism that performs better with more volatility.

Plan Accuracy

Overall plan accuracy: 4.0 / 5

Desk-Level Learnings

• First positive clustered observation for Family 5. The second same-day co-fire produced a net +$46.33 (vs. the first on 2026-07-13 at -$53.95). The clustered record is now 1W/1L, net positive. Still far too early for any conclusion (n=2), but the pooled validation framework is functioning: same-day trades across Family 5 members are being tracked as one observation for significance purposes.
• MNEME's cumulative P&L has reached a new high. $+197.01 on 4 trades with 50% WR. The SPRT monitor shows LLR +0.739, +2.206 from CONSISTENT boundary — solidly in CONTINUE territory with meaningful cushion above the DEGRADED boundary.
• TRACE's SPRT monitor is approaching the DEGRADED boundary. LLR -0.289 with only -2.655 of room before reaching DEGRADED (-2.944). Three losses in four trades is within the expected variance of a ~44% WR strategy (the p0 from the SPRT pre-registration), but if the next trade is also a loss, TRACE will be at 20% WR (1/5) with an LLR very close to the boundary. This requires active monitoring but is not yet actionable — the pre-registered SPRT thresholds were set to trigger at evidence, not at narrative discomfort.
• Max_sim distance above gate is worth monitoring. MNEME's win came at 0.8483 (0.3483 above the 0.50 gate); TRACE's loss came at 0.6013 (0.0013 above the 0.60 gate). Both members' backtests found near-zero similarity spread (within-cohort similarity doesn't predict outcome), but the live data is asking a question the backtest couldn't answer at 59–98 holdout trades. If this pattern persists over 5+ more trades per member, it's worth a TEMPER conversation about whether max_sim above gate carries signal for Family 5 that the backtest's statistical power couldn't detect.

Reflection Filed

• Filed: 2026-07-16 18:30 ET
• Next session event risk: NONE (no Tier-1 events on 2026-07-17)
• GHOST: frozen (Phase 2) — no reflection written.