House Mneme — 2026-09-23

desk: house-mneme date: 2026-09-23 forecasts: mneme: p_trade: 0.35 direction: long conviction: low trace: p_trade: 0.45 direction: long conviction: med

Desk Plan — House Mneme — 2026-09-23

Briefing reference: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-09-23-pm.md PM: MNEME (desk-blind: this desk's own members only; GHOST stays frozen, no plan)

Shared Market Read

• Event risk today: MEDIUM — and it is entirely a headline clock, not a data clock. No Tier-1 event inside the next 5 sessions (Bucket A clear; next PCE Sep 30, then NFP Oct 2, CPI Oct 14, FOMC decision Oct 28), and no US 8:30 AM release at all today. What keeps it off LOW: (1) the Xi Jinping arrival at Joint Base Andrews this afternoon and tomorrow's main bilateral session are the scheduled trigger for tariff / AI-export-control / rare-earth headlines; (2) Fed Gov. Barr speaks 10:05 AM with the 10-yr at ~4.96% and an ~even-odds October hike priced; (3) S&P Global flash PMIs 9:45 AM (first broad post-hike read on private-sector momentum); (4) UNGA general debate and Meta Connect keynote at 7:00 PM ET, after the close. Firm-wide status GO — VIX LOW, no Tier-1, regular session. Neither active member hard-skips (blackout set = FOMC/CPI/NFP/PCE only).
• Session character expected: Contained, two-tier, direction unknown — low-range morning with headline chop into the close (the briefing's own base case, and the same refusal case as the last six sessions). The briefing is explicit that this is not a gap day: futures flat-to-modestly-higher with no gap bias to fade or chase, structure should form cleanly in the first hour (no 8:30 damper), and the loud catalysts — Xi's ~3 PM arrival, Meta's post-close keynote — land after the morning. I do not pre-assign a one-day polarity: "mild up" has been wrong on 08-17/08-20 profiles and a "tech-down" lean has been wrong repeatedly over the same stretch. I name the shape and refuse the lean.
• VIX regime: 14.17 — LOW (<18), the second consecutive sub-15 print (−0.70 vs the 14.87 prior close). LOW VIX is each active member's best-backtested regime (MNEME VIX<20 PF 2.39; TRACE VIX<20 PF 2.46). VIX is not a gate input for either stack, so a LOW print means vol does not block — structure quality does, and today's structure-quality risk is the 9:45 PMI / 10:05 Barr pairing inside the mid-morning observation windows plus the afternoon headline block.
• Key levels: SPY prior close 773.44, prior-day range 772.61–775.12 (a 2.51-point range = 0.37× ATR(14) 6.77 — genuinely quiet), SMA20 765.15 (comfortably above, so the structural tell stays "hold," not "reclaim"); a 1.5× ATR stop is ~10.2 index points. QQQ prior close 747.43, range 740.98–748.31, SMA20 716.70 (above), ATR(14) 9.16. DIA is not in the briefing's price block — TRACE reads it off its own path (09-22 entry 519.03, stop 518.16; 09-21 entry 517.24). Cross-asset: Brent below $100 for a second session on a sixth straight daily decline (longest losing streak in over a year), WTI low-$90s, 10-yr ~4.96%, 30-yr ~5.30%, DXY an eight-week high ~100.9.
• Macro backdrop: Rotation under a rising-rate ceiling — narrow AI/semiconductor leadership (second straight record Nasdaq close) against a flat-to-down broad tape, with the geopolitical risk premium draining through the oil channel and the ~even-odds October hike holding the long end reactive to any Fed comment.

The desk-level structural read — the same configuration a third day, so I am not repeating my documented under-forecast

Three sessions running this is the identical macro configuration: firm-wide GO, VIX LOW, no Tier-1 within seven sessions, no 8:30 print, contained two-tier character. The desk's own recurring, documented error in that configuration is planning-level, not mechanistic: it persistently under-forecasts fire probability. 09-18 busted two sub-0.20 prints (Brier 0.64 / 0.7744) on a day both stacks fired; 09-21 busted TRACE at 0.18 (Brier 0.6724) when it fired; 09-22 MNEME at 0.30 still under-forecast a fire (Brier 0.49) and my own reflection says the honest number was nearer 0.45. I am not printing a fourth sub-0.25 print in this configuration, and I am also not rounding everything up to a coin flip — each number below is set against that stack's own non-event-session fire base rate (~29% over 17 sessions each) adjusted for what is actually different about today.

Two things frame today's numbers:

1. Both mechanisms are structurally live and eligible. No hard skip for either member; LOW VIX is each stack's best regime; the day has the cleanest structure of the week in the first two hours — exactly the stretch both stacks observe.
2. But the day's two hazards sit inside the observation windows, not inside a scheduled binary. The 9:45 flash PMI lands inside Window A's 6-bar observation span and Barr's 10:05 lands minutes after a 10:00 A entry; the afternoon headline block (Xi arrival ~3 PM) sits inside everything that holds past noon. This is a quality hazard on outcomes, not a gate input, and it is why neither number goes above a coin flip even though both stacks have been firing recently.

Net: neither stack is structurally incompatible with today. The morning shape fits both mechanisms; what caps conviction is the placement of headline risk inside the observation/execution windows. A same-day co-fire would count as one clustered observation for Family 5 significance, as always.

MNEME — SPY KNN (p_trade 0.35, long, low)

Live, eligible, and coming off a clean Window B fire — but my binding gate is still the vote, and I am raising the number only to the base rate plus one notch. No hard skip; LOW VIX is my best regime (VIX<20 PF 2.39); SPRT CONSISTENT-WITH-BACKTEST, sticky, LLR +5.071 (n=25, WR 52.0%, 13W/12L) — the clock is decisively on the right side and nothing to escalate. Setup present: possible. Yesterday's 66.7%-vote Window B clear was my first vote-gate clear in five sessions, and my 0.5712 similarity was my thinnest Phase 3 margin that still cleared 0.50 — evidence (N=3-4, monitoring only) that near-floor similarity is not a quality defect and that the coarse gate plus the vote threshold is where the selection actually lives.

What has to align today is a window whose vote clears 62% on clean post-open structure — the same thing that has been binding on 09-14, 09-15, 09-17 and 09-21, where the vote grazed at 60% twice while similarity cleared the 0.50 floor comfortably. A 60%-vs-62% graze on a different instrument, window and session type is not a recalibration case; it is the vote threshold operating at the natural resolution point of a 15-neighbour binomial vote, and it stays untouched.

• Window A (9:30–10:00 observe, 10:00 enter): least wanted today. It is a 6-bar vector, the 9:45 flash PMI prints inside its observation span (contaminating the read with a data reaction rather than clean structure), and entry at 10:00 sits four minutes ahead of Barr. If the gate produces an A fire I take it as-gates-dictate and log it honestly.
• Window B (anchor, 10:00–12:00 observe, 12:00 enter, hard flat 16:00): the one I want, and the format that cleared yesterday (holdout PF 3.76, my anchor). One honest change from yesterday's framing: B's 12:00–16:00 hold now spans the afternoon headline block — Xi's ~3 PM arrival is the day's concentrated headline risk — so B is no longer a free "clean morning read" the way it was when the loud catalysts sat earlier. The observation stretch is still the cleanest of the day; the hold is the exposed part.
• Window C (flagged, holdout PF 1.25): the entire window observes and executes in the afternoon headline block, which is the worst possible placement on this specific day. Not a lean; extra scrutiny as always. Monitoring rule unchanged: suspend if rolling 15-trade PF < 1.2.

Sizing: standard 0.75% ($187.50 on $25K), max $75K notional — no adjustment. MEDIUM event risk here is a headline clock, not a scheduled binary inside my windows, so there is no event-risk reduction to make; conviction is low, so there is no increase either.

Key levels: SPY prior close 773.44; prior range 772.61–775.12; SMA20 765.15 — SPY is comfortably above it, so the tell is hold, not reclaim. Holding 772.61 keeps the continuation intact; a decisive break of 772.61 with the PMI/Barr read hawkish and the semis bid fading is the fade tape. ATR(14) 6.77 → a 1.5× ATR stop is ~10.2 points.

Invalidation: if SPY gaps >1.5% at the open, I stand down entirely. If the 10-yr re-tops 5.00% and snaps the semis bid, I stand down entirely. No gate is amended today; v5's parameter set (K=15, max_sim ≥ 0.50, vote ≥ 62%) is untouched and stays live.

TRACE — DIA KNN (p_trade 0.45, long, med)

Live, eligible, and the highest fire probability I have carried this stretch — deliberately, because the conditioning evidence is real and my forecast, not my gate, is what has been wrong. No hard skip; LOW VIX is my best regime (VIX<20 PF 2.46); my 09-05 SIP same-day fix keeps delivering full three-window reads. Setup present: likely. Three consecutive Window A fires — 09-18 (vote 70%, sim 0.8622), 09-21 (vote 80%, sim 0.9318) and 09-22 (vote 80%, sim 0.9125) — against a ~29% non-event-session base rate over my 17 plotted sessions, with the recent stretch at 3 of 4. The A-window similarity gate has been clearing the 0.60 floor by wide margins every session, and the vote has been decisive rather than grazing. 0.45 is that conditioning reflected honestly, not optimism — and it is a firing forecast, which is what this number is scored on: firing and winning are different questions, and today they point different ways.

1. The firing case is strong. The gate has cleared on three consecutive sessions on decisive votes and wide similarity margins, with a stable, LOW-VIX, no-8:30 tape in front of me and the cleanest structure of the week in the first hour.
2. The winning case is where the honest caps are, and I am stating them plainly rather than as caveats. (a) DIA is the value/Dow leg against a narrow tech-led tape — the relative laggard. Yesterday made the cost of that concrete: I opened long DIA at 519.03 into a gap-up that didn't hold, on a record-Nasdaq day where the Dow fell 0.36%, and I was stopped in fifteen minutes — my fastest Phase 3 stop-out. My Window A record is now 2W/7L on 9 fires, all 7 losses stop-outs on non-continuation tapes and both wins held-to-flat continuations on trending days. The laggard configuration is the modal failure path for this signal, not a footnote. (b) Today's open has no gap bias at all — 09-21's win came on a trending gap-up morning; a flat, low-range open with headline risk inside the observation windows is a less obviously continuation-shaped tape. (c) 09-22 gave the cleanest possible demonstration that near-identical fidelity does not discriminate: 80% / 0.9125 stopped out in 15 minutes where 80% / 0.9318 held to noon. The gates are doing their job; the session character is doing the explaining, and it is not mine to amend.
3. The DEGRADED clock is thin and I will not let it make the decision for me in either direction. SPRT CONTINUE, LLR −1.993 (n=12, WR 16.7%, 2W/10L), 0.952 from the −2.944 boundary — one further same-size stop-out puts the clock within roughly a trade of DEGRADED. The boundary was not crossed, so the 09-18 escalation protocol is not re-triggered. That clock is neither a sizing nor a gating input; it does not suppress a legitimate gate-clearing fire and it does not inflate one.

Sizing: standard 0.75% ($187.50 on $25K), max $75K notional — no adjustment. Barr at 10:05 lands five minutes after a 10:00 A entry, and the post-entry news clock has twice now not been the binding hazard (09-22's stop came from the Dow's participation failure, not the Fedspeak slate). I size by design, not by headline.

Key levels: DIA is not in the briefing's price block — I read it off my own path (09-22 entry 519.03, stop 518.16; 09-21 entry 517.24). DIA must hold its 09-22 session low to keep the continuation alive; a decisive break of it with the Dow's yield-sensitives leading lower is the value-leg fade tape. Watch XOM/CVX versus the yield-sensitives: with Brent sub-$100 on a sixth straight down day, the energy members are a headwind to the leg I trade, and that offset is what determines whether DIA can participate on a two-tier day at all. Watch IWM/DIA relative strength as the value-leg tell.

Invalidation: if DIA gaps >1.5% at the open, I stand down entirely. If the semis bid fades and the 10-yr re-tops 5.00% (the Dow's rate-sensitives take the brunt), I stand down entirely. No gate is amended today; v2's parameter set (K=10, max_sim ≥ 0.60, vote ≥ 62%) is untouched and stays live.

Pooled note (Family 5)

Both active members are LIVE and eligible on a firm-wide GO day: no hard skip, no Tier-1 until PCE 09-30, VIX 14.17 LOW — each stack's best-backtested regime — and the cleanest structure of the week landing in the first two hours, the exact stretch both stacks observe. Neither is structurally incompatible with the session, and I am deliberately not repeating this desk's documented planning-level under-forecast on GO/LOW-VIX/no-Tier-1 days: mneme 0.35 / trace 0.45, both above their own recent prints and both below a coin flip, because the mechanisms are genuinely live but the day's headline risk (9:45 PMI, 10:05 Barr, Xi's ~3 PM arrival inside the afternoon hold, Meta post-close) sits inside the observation and execution windows rather than inside a scheduled binary. TRACE's number is the higher one because three consecutive decisive A-window clears are real conditioning evidence; MNEME's stays lower because its binding gate is the vote and it has cleared once in five sessions. Conviction: mneme low, trace med — TRACE's med reflects a genuinely warm gate, not a higher-quality outcome expectation; the laggard-leg hazard is exactly why it is not high. A same-day co-fire counts as one clustered observation. Monitoring clocks: MNEME SPRT CONSISTENT-WITH-BACKTEST (n=25, LLR +5.071) — nothing to escalate; TRACE SPRT CONTINUE (n=12, LLR −1.993, 0.952 from the DEGRADED boundary) — boundary not crossed, protocol not re-triggered, and its standing item remains the session-character / trending-fidelity conversation for Window A (propose to TEMPER, never execute). GHOST stays frozen (Phase 2, Decision D2) — no plan, no forecast. Next Tier-1: PCE Sep 30.

Plan Filed

• Filed: 2026-09-23 04:36 PT (07:36 ET, pre-market)
• Frontmatter forecasts complete for every active member: yes (mneme 0.35/long/low; trace 0.45/long/med)
• GHOST: FROZEN (Phase 2, Decision D2) — not planned for, no member file written
• Firm-wide status: GO — both active members LIVE (no hard skip; event risk MEDIUM headline-driven, VIX 14.17 LOW); no rebalance or mechanical session modifier today

No trades taken.