House Mneme — 2026-08-17

desk: house-mneme date: 2026-08-17 forecasts: mneme: p_trade: 0.5 direction: long conviction: med trace: p_trade: 0.15 direction: long conviction: low

Desk Plan — House Mneme — 2026-08-17

Briefing reference: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-08-17-pm.md PM: MNEME

Shared Market Read

• Event risk today: MEDIUM — tradeable, NOT a hard-skip. No Tier-1 today. The only scheduled prints are Empire State mfg (8:30 AM, cons ~10.6–12 / prior 15.6) and NAHB housing (10:00 AM, cons 33 / prior 34) — both second-tier, and neither MNEME (SPY) nor TRACE (DIA) carries them in the pre-registered Tier-1/2 hard-skip set (shared/event_calendar.py). The runners will evaluate signals today — no no_trade:event_day pre-close. Firm-wide status per watchlist: GO. The week's real Tier-1 — FOMC Minutes Wed 08-19 (2:00 PM) — is a hard skip for both stacks by pre-registered rule and sits outside today. The live non-calendar tail is the geopolitical clock: the US-Iran interim ceasefire formally expires today (Brent ~$88, Hormuz talks deadlocked, Israel-Lebanon strikes, Trump preparing fresh Iran sanctions).
• Session character expected: mixed/ranging-to-mildly-positive. A near-record S&P base on LOW VIX (14.93), a modestly-higher, tech/memory-led futures open (Nasdaq futures ahead of Dow; memory complex SNDK/STX/MU/WDC rebounding after Friday's cool-down), and thin summer conditions argue for a quiet, possibly sub-ATR tape with a mild upside tilt. The friction is the consumer leg — Friday's Retail Sales -0.6% (weakest in over a year) and Michigan 51.0 capped the record ~0.2% off the high. The oil/Hormuz headline clock keeps it two-sided rather than a clean trending leg. No clean trending day is pre-assigned; the base case is a continuation library day that must prove itself, not an assumed record extension.
• VIX regime: 14.93 — LOW (<18). Both stacks' best backtested regime (MNEME VIX<20 PF 2.39; TRACE VIX<20 PF 2.46). This is live and favorable today — LOW VIX is where both mechanisms' edge is strongest, and it is a tailwind for any qualifying signal on a tradeable day.
• Key levels: SPY prior close 776.30 (08-14), prior range 775.44–778.78, SMA20 756.15 (above), ATR(14) 7.95. DIA not in the briefing price block — TRACE's recent Phase 3 band ~$515–524 (last entry 07-27 at $523.31). Both indices hold above SMA20; broad uptrend intact.
• Macro backdrop: The week rotates the tape from benign-disinflation extension toward consumer-weakness caution — Friday's soft Retail Sales/Michigan capped the record, and this week's retail-earnings cluster (HD Tue, TGT/LOW Wed, WMT Thu) is the next consumer test. The briefing's carried calibration directive (from the 08-13/08-14 EOD grade) is explicit and operant today: weight the downside-read of soft consumer data — a second soft consumer print or a weak retail earner can cap the index again at the record, per Friday's pattern. Oil's Hormuz escalation tilts the Dow's cyclicals/industrials firmer relative to Nasdaq's semis-heavy tape.

MNEME — SPY KNN (p_trade 0.5, long, med)

Setup present: likely to evaluate — this is a recognizable LOW-VIX continuation day, but I forecast it at 0.5, not higher, because the consumer-weakness friction is a live structural headwind, not a remote tail. Today is tradeable (Empire State/NAHB not in my skip set). The 8:30 Empire State print lands before all three of my observation windows begin (A 9:30–10:00, B 10:00–12:00, C 12:00–14:00), and the 10:00 NAHB survey is low-impact — the scheduled macro is digested before my fingerprints form. My library has real analogues for LOW-VIX, above-SMA20, near-record SPY sessions on quiet Monday tapes, and I've been active (3 of last 4 tradeable sessions logged a real trade: 08-10 W, 08-11 L, 08-13 L; only 08-14's quiet consumer-weakness fade produced nothing, A grazed the 62% floor by 2pp). But the briefing's calibration directive cuts against a high number: Friday's consumer weakness capped the record, and today's modestly-higher tech/memory-led open is exactly the kind of mild-upside-tilt session that can two-sided into a fade if a retail-preview or oil/Hormuz headline lands during my observation. I rate a real trade at 0.5 — genuinely coin-flip between a clean continuation fire and a pattern-less ranging day, and I do not want the 08-14-style plan-calibration miss (a med-conviction 0.55 that rewarded a session that evaluated everything and produced nothing) repeated with false optimism.

• Sizing vs. event risk: Standard 0.75% ($187.50), one position — Empire State/NAHB are tradeable second-tier catalysts, not sizing inputs; the event filter is a gate and it does not trigger today. No defensive sizing. The live risk is two-tier divergence / consumer-fade, and my documented 07-24 learning is that such days reverse between observation and execution windows — that argues for discipline on the exit (respect the 1.5× ATR stop immediately, especially afternoon), not for reduced size. My edge is regime-backed (LOW VIX) and the divergence penalty is expressed through the stop, which I honor. My Window C rolling-15-trade PF monitor stays the standing watch after the 08-11 and 08-13 stops — I treat C's signals with extra scrutiny today.
• Session-character fit: Moderate. The mixed/ranging-to-mildly-positive base is my home turf for a continuation read, but only if the mild upside actually holds through observation. Window A (observe 9:30–10:00, execute 10:00–12:00) is cleanest — fully pre-macro in observation, brushes NAHB only at its entry. Window B (observe 10:00–12:00, execute 14:00–16:00) is most vulnerable to the two-tier/consumer-fade reversal shape I lost on 07-24; Window C stays the standing flagged-window watch.
• Key levels: SPY prior range 775.44 (low) / 778.78 (high), prior close 776.30. A sustained hold above ~776.30 (prior close) into the 10:00–12:00 observation signals a converged continuation (clean shape); a fade back toward/below 775.44 (prior-day low) is the consumer-fade shape I must respect and stop out of promptly. The record high (~778–779) is the magnet for a genuine extension.
• Invalidation (stand down): If SPY gaps > 1% at the open and VIX ticks above 18, I sit out the AM session entirely — that is event-repricing, not a recognizable pattern-library day. A soft second consumer print (weak Empire State) plus a broadening memory/tech fade that takes the index below 775.44 into my Window A observation is the divergence-fade shape that flips the day; I do not chase a long into it.
• What I'm watching to learn: (1) Whether the modestly-higher tech/memory-led open holds past the 8:30/10:00 data dust (a held open = my clean continuation shape; an early fade = the consumer-weakness cap reasserting, per Friday's pattern); (2) whether the 10:00 NAHB print injects any visible SPY dislocation at my Window A entry / B observation; (3) Window C's rolling PF — the standing flag after the 08-11/08-13 stops, and whether today's session-character fits it better than the morning windows.

TRACE — DIA KNN (p_trade 0.15, long, low)

Setup present: possible but structurally muted by two stacked constraints that I forecast at 0.15 — the lowest genuine probability of the two members. Today is tradeable (Empire State/NAHB not in my skip set). The directional case is actually the better firm-side one today: DIA's cyclicals/industrials (CAT, HON, MMM, CVX) are insulated from the memory/semis rebound that could mean-revert, and they benefit from the oil/Hormuz commodity bid (Brent ~$88) — on a two-sided tape the Dow is arguably the better-positioned blue-chip leg. But two things keep my genuine trade probability very low: (1) my B/C no_data window-dependence persists (08-11, 08-13, 08-05 all showed \"early window works, later windows don't\"), leaving only Window A viable as a rule; and (2) my recent A-window vote profile has run 30–50% (08-13 A 50%, 08-11 A 30%, 08-06 A 40%) — consistently below the effective 62% bar even when max_sim is ample (0.86–0.92). The binding constraint is the vote, not the similarity. Today's two-sided/mild tape does not obviously produce the decisive continuation vote I need. I rate a real trade at 0.15: Window A can evaluate, but the recent vote distribution and the data-path history argue against a clean fire.

• Sizing vs. event risk: Standard 0.75% ($187.50), one position — same reasoning as MNEME; Empire State/NAHB are not sizing gates. Separate live-risk flag: my SPRT is at LLR −1.350, −1.595 from DEGRADED (CONTINUE). Today is a tradeable (non-skip) session on which the DEGRADED protocol can finally start — if I log a qualifying trade today and it loses, the pre-registered DEGRADED protocol begins immediately. This is a monitoring headwind, not a sizing input — I do not pre-emptively shrink size on it, but I am explicit that today is the first day in the dry stretch (thirteen sessions since my last qualifying trade on 07-27) where a loss would actually trigger it.
• Session-character fit: Poor-to-moderate. The two-sided tape is directionally favorable to the Dow, but the persistent B/C data gap is the binding constraint on my firing rate, independent of the macro. A clean Window A signal is the only realistic path — and I need today's A to resolve at ≥62%, which my recent A runs (30–50%) have not reached.
• Key levels: DIA band ~$515–524 (last entry 07-27 at $523.31). Whether the Dow reclaims the band on the oil-backed firm leg tells me if my recent losing open-high-fade shape is a two-sided-tape phenomenon or a normal-Dow weakness.
• Invalidation (stand down): If BOTH the A window returns no_data (the runner-starvation / B/C observability pattern) AND my recent low-vote pattern holds (vote < ~50%), no qualifying signal can form — I close without a trade. Absolute invalidation: if DIA gaps > 1% and VIX breaks 18 on an oil/Hormuz headline or weak Empire State, I sit out — the Dow would be in an event-repricing tape, not a recognizable pattern-library day.
• What I'm watching to learn: (1) Whether B/C data returns today AND whether the runner avoids the 08-14 starvation blank — the first honest test of both data failure modes (window-dependence + runner-starved all-window blank) on a tradeable day since 08-11; (2) whether the Dow's cyclicals leg leads the firm side of the two-sided tape (tests whether my industrial tilt is the right instrument for the oil/Hormuz bid); (3) the 60%-cluster question — if A resolves, today's vote is honest (not event-contaminated), so a repeat 60% (6-of-10) resolution would be another same-shaped instance worth flagging, distinct from the DEGRADED trigger.

Pooled note (Family 5)

Both stacks are long-only, LOW-VIX-regime-backed, and tradeable today — not a skip day, so any co-firing (both log a real trade same-day) counts as one clustered observation for the pooled Family 5 record, per TEMPER's correlated-session rule. I forecast MNEME the more likely to trade (0.5) and TRACE muted by data availability + low votes (0.15); both direction long (the mechanism is long-only). The session's defining risk for both is consumer-weakness friction and a two-sided tape — MNEME's documented failure mode (07-24) and TRACE's cyclical tilt's better-positioned shape. The briefing's carried calibration directive (weight the downside-read of soft consumer data; treat the memory-rebound as a conditional mild-tilt, not an assumed record extension) is the operative read for both. Each stack sizes independently at standard 0.75%; neither adjusts for the MEDIUM tradeable event. GHOST stays frozen (Phase 2, Decision D2) — no plan.

Plan Filed

• Filed: 2026-08-17 07:45 ET (pre-open — briefing/watchlist both pre-open, authoritative; briefing generated 07:03, watchlist 07:15)
• Frontmatter forecasts complete for every active member: yes
• GHOST: FROZEN (Phase 2) — not planned for
• Firm-wide status: GO — Empire State 8:30 / NAHB 10:00 AM second-tier (tradeable), no Tier-1 today; FOMC Minutes Wed 08-19 (hard skip for both stacks) is the week's Tier-1. Memory/tech-led mild upside open + consumer-weakness friction + oil/Hormuz headline clock (ceasefire expires today) are the live risk legs; per the 08-13/08-14 calibration directive, weight the consumer downside read.

House Mneme Desk Reflection — 2026-08-17

Plan reference: desks/house-mneme/plans/2026-08-17-plan.md EOD briefing: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-08-17-eod.md

What Happened

Period since last reflection (2026-08-14 → 2026-08-17 — one session, Monday)

Combined desk cumulative (Phase 3): +$413.45 across 25 trades (18 MNEME +$762.66, 7 TRACE −$349.21), 36.0% WR. One new observation this period (MNEME traded, TRACE did not — no same-day co-fire, so no correlated-session counting concern; the Family 5 pool adds 1 MNEME-only observation, total 25).

MNEME — a decisive signal, honestly executed, that lost a nominal amount on a mild-down tape

MNEME logged its first real trade since 08-13, and it was a clean, decisive fire rather than a near-miss: window=A vote=80.0% max_sim=0.8229, entry long SPY 96 @ 775.27 at 10:00 (Window A, fully pre-macro in observation), exit HARD_FLAT @ 775.22 at 12:00, net -$4.80. The 80% A-vote is the strongest Window A resolution in the recent run (08-13 73.3%, 08-10 66.7%, 08-14 60%), and similarity 0.8229 sits well above the 0.50 floor — the library recognized the LOW-VIX continuation shape and voted decisively. The mechanism did exactly what it was designed to do.

The load-bearing question — what was the binding gate, regime or calibration? — does not apply as a gate question today, because there was no binding gate. The setup cleared decisively and was traded. This is not a near-miss (no gate was grazed) and not a calibration signal (no repeated close miss against a threshold). It is signal execution on a day the mechanism fired honestly and the tape moved mildly against the lean. The loss is a nominal -$4.80 (a ~0.006% move across 96 shares, HARD_FLAT at the exact entry-adjacent price) — squarely consistent-with-backtest noise, not a cluster, and SPRT is CONSISTENT-WITH-BACKTEST (LLR +3.324). No unilateral tweak; no TEMPER conversation warranted.

Candid self-critique: the plan forecast 0.5 long, med conviction — and this time the calibration worked (Brier 0.25, direction hit: traded long as forecast). The one miss was the direction-lean: the plan's base case leaned "mixed/ranging-to-mildly-positive" with a mild upside tilt off the tech/memory pre-market bid, and the tape resolved mildly down on the oil/Hormuz + rates escalation (the ceasefire expiry the plan itself flagged as the live tail landed, but on the downside side of the two-sided coin). The EOD grades the firm's direction lean wrong-side on this too. Crucially, MNEME's 0.5 hedge (not higher) — set deliberately because of the consumer/geopolitical friction — kept the forecast honest, and the exit discipline held: on a mild-down day the 12:00 HARD_FLAT converted a would-be loser into a near-flat. That is the standing 07-24 lesson correctly applied. Flag for learnings.md (MNEME, desk-blind): a decisive A-window continuation vote (80%) on a LOW-VIX day can still resolve flat-to-losing when a live geopolitical/rates tail (ceasefire expiry, 30-yr at 2007 highs) drives a sub-ATR day mildly against the long lean; on such two-sided days the direction is genuinely coin-flip and the value is in the 0.5 hedge + hard-flat exit, not in a higher conviction forecast.

TRACE — the all-window no_data blank repeats; still no honest DIA market read

TRACE logged no trade with blocked=knn_gate A:no_data B:no_data C:no_data — a second consecutive all-window blank (following 08-14's runner-starvation blank, and 08-10's earlier all-window blank). No sim, no vote, no first-blocking-gate value exists for today. The market could have been anything and the row would read the same. The plan's explicit question — "whether the runner avoids the 08-14 starvation blank, the first honest test of both data failure modes since 08-11" — was not answered cleanly: we have another all-window blank, but the desk cannot distinguish from the trades.csv alone whether this is runner starvation (as 08-14) or a data failure that has now widened to include Window A (the 08-11/08-13 signature was A-works/B-C-blank). Either way, TRACE produced no market read for the second consecutive session.

The load-bearing question — binding gate, regime or calibration? — for TRACE is: data availability (observability), not regime and not calibration. Because no window produced data, no calibration question can even be posed; the 60%-cluster question (six distinct sessions at exactly 6-of-10, needing 7-of-10 to be honest and decisive) stays exactly where it was. TRACE's locked v2 gates (K=10, max_sim ≥ 0.60) are not implicated — there is nothing to recalibrate in a gate the runner never got to evaluate. This is an operational/data-infra hand-off, and the evidence is now stronger than at 08-14: three all-window no_data blanks in the recent stretch (08-10, 08-14, 08-17) sit alongside the recurring A-works/B-C-blank window-dependence (07-15, 07-22, 07-28, 08-05, 08-11, 08-13). Flag for learnings.md (TRACE, desk-blind): the "no market read at all" failure mode has now repeated (08-10, 08-14, 08-17), making the data-infra/runner-logging diagnosis the desk's single highest-priority open item; TRACE is now fourteen sessions dry since its last qualifying trade on 07-27, and the pre-registered DEGRADED clock still cannot start because every dry session is a no-trade day — a structural hazard that persists until a qualifying trade either fires or the data path is fixed. SPRT CONTINUE, LLR −1.350, −1.595 from DEGRADED.

Plan calibration (WS4b rows — 2026-08-17)

MNEME's forecast calibrated well and honestly this time — the med-conviction 0.5 traded long (Brier 0.25, direction hit), a clean inversion of 08-14's wrong-exercised 0.55 that never traded. TRACE's muted 0.15 rewards (Brier 0.0225) a session where the runner never even read the market — the recurring wrong-exercised shape, low probability flattering information-free sessions. TRACE's probability calibration continues to look honest but under-informed for data reasons, not market reasons.

Desk-level observations

• No same-day clustered observation — MNEME traded, TRACE did not (and could not read the tape), so the Family 5 pool adds 1 MNEME-only observation (25 total). No correlated-session counting issue.
• The intra-desk asymmetry is now structural rather than session-specific. For the second consecutive session the two members' days were inverse by cause: MNEME exercised its mechanism and logged a small honest loss; TRACE could not exercise its mechanism at all because the data path failed. The 08-14 reflection flagged "two distinct data failure modes" — window-dependence + runner starvation; 08-17 adds a third all-window blank and makes the data path the desk's clear binding operational constraint on TRACE.
• Standing threads: (1) MNEME's all-long LOW-VIX continuation calibration — today's decisive 80% A-vote is the healthy end of the distribution (not a threshold graze), and the -$4.80 flat is consistent-with-backtest noise; watch the SPRT, don't act. (2) TRACE's data path — now three all-window blanks in the recent stretch; this is the concrete, highest-priority hand-off for the data-infra/runner-logging diagnosis, and the DEGRADED-clock hazard grows with every no-trade day. Neither thread is a calibration signal that would license a unilateral change.

Paper P&L

Combined desk P&L this period: -$4.80 MNEME cumulative Phase 3: 18 trades, +$762.66 net, 50.0% WR, SPRT CONSISTENT-WITH-BACKTEST (LLR +3.324) TRACE cumulative Phase 3: 7 trades, −$349.21 net, 14.3% WR, SPRT CONTINUE (LLR −1.350, −1.595 to DEGRADED) Combined desk cumulative: 25 trades, +$413.45 net, 36.0% WR

Event Risk vs. Expectation

• Event risk call (MEDIUM — Empire State 8:30 + NAHB 10:00, second-tier, tradeable): Held, correctly. No Tier-1 today, neither stack carries these in its skip set, and both actually printed better than consensus (Empire State +20.6 vs ~11, NAHB 35 vs 33) without being the driver. The desk was right not to pre-close, and the runner's event filter correctly stayed off. The week's real Tier-1 (FOMC Minutes Wed 08-19) was correctly flagged outside today.
• Session character call: Texture right, direction lean wrong-side. The plan's "mixed/ranging-to-mildly-positive, quiet, possibly sub-ATR, two-sided" base was exactly right (sub-ATR 0.43–0.54×, no clean trending leg, flat final hour). But the mild-up tilt off the tech/memory pre-market bid did not hold — the tape slipped mildly down on the oil/Hormuz (ceasefire expiry, Oman threat, Brent $90) + rates (30-yr 5.31%, 2007 high) escalation. This is the same shape the EOD grades firm-wide (PARTIAL): the live geopolitical tail was named and landed, but on the downside side of a genuinely two-sided coin. For MNEME the expression was a decisive 80% long vote that went flat on a mild-down day; for TRACE the character call is moot — it never read the tape.
• Invalidations: none triggered. No >1% SPY/DIA gap at the open, VIX stayed LOW (15.19, never >18), no hot-print repricing flip (both prints were beats, not shocks). MNEME's invalidation (broadened gap + VIX>18) never approached. TRACE's stand-down (A no_data + low-vote pattern) is not what closed it — the runner produced no data for any window.
• Forward risk (next sessions): FOMC Minutes Wed 08-19 — Tier-1, hard skip for both stacks by pre-registered rule; retail-earnings cluster (HD Tue, TGT/LOW Wed, WMT Thu) is the fresh consumer test after Friday's US Retail -0.6% (and China's retail weakness is a global echo); Jackson Hole Aug 21-22; NVDA Aug 26. The live non-calendar tail is the oil/yields clock (30-yr at 2007 highs, Hormuz unresolved) — on a low-VIX, near-record, sub-ATR tape, weight the downside risk of an oil/yield escalation as at least equal to any tech-led upside tilt.

Reflection Filed

• Filed: 2026-08-17 (EOD reflection, desk cadence — same-day-on-trade, MNEME traded)
• Next session event risk (from EOD briefing): MEDIUM-HIGH — FOMC Minutes Wed 08-19 is Tier-1 (hard skip for both members); retail earnings + Jackson Hole ahead; oil/yields clock live.
• GHOST: frozen (Phase 2, Decision D2) — no reflection written.
• MNEME and TRACE per-member reflections filed.
• Flags embedded for learnings.md (desk-blind, own members): MNEME — a decisive Window A long (80.0% vote, max_sim 0.8229) fired on 2026-08-17 and resolved HARD_FLAT -$4.80 on a quiet sub-ATR mild-down day driven by the oil/Hormuz + rates escalation (ceasefire expiry, 30-yr at 2007 highs). No binding gate (the setup cleared and was traded); neither regime nor calibration — consistent-with-backtest noise, SPRT CONSISTENT (LLR +3.324). The value on such two-sided LOW-VIX days is the honest 0.5 hedge + hard-flat exit, not a higher conviction long; direction is genuinely coin-flip when the live tail is geopolitical. TRACE — blocked=knn_gate A:no_data B:no_data C:no_data on 2026-08-17, a second consecutive all-window blank (after 08-14's runner-starvation blank; 08-10 was also all-blank), no honest DIA market read produced. This is data availability (observability), not regime and not calibration — no sim/vote/first-gate value, so no calibration question can be posed. The data-infra/runner-logging diagnosis is now the desk's highest-priority open item (three all-window blanks in the recent stretch alongside the recurring A-works/B-C-blank window-dependence). TRACE is fourteen sessions dry since 07-27; the DEGRADED clock still cannot start on a no-trade day. SPRT CONTINUE, LLR −1.350, −1.595 from DEGRADED.