House Echo — 2026-08-20

Desk Plan — House Echo — 2026-08-20

Briefing reference: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-08-20-pm.md PM: ECHO (Close-Leg Exhaustion / Momentum family)

Shared Market Read

• Event risk today: LOW — no Tier-1 within ~2 days (PCE + NVDA earnings are Aug 26, 6 days out; Jackson Hole is Aug 27–29, not this week). Today's own calendar (Initial Jobless Claims 8:30, Philly Fed 8:30, Existing Home Sales 10:00, US 10/30-yr note auctions ~2 PM) is low-key and non-Tier-1. Not a hard-skip day for either member. The live risk is the coin-flip on no calendar — not the scheduled schedule: WMT/TJX (both raised FY guidance pre-market — the "consumer is fine" read extends), oil/Hormuz (Brent ~$93, WTI ~$86–87 on renewed Mideast escalation), and whether the Treasury-buyback long-end relief (30-yr 5.19%) confirms or re-tests.
• Session character expected: LOW-VIX, near-record, sub-ATR, contained two-tier, direction unknown. Both indices sit above SMA20 and near all-time highs, VIX 15.14, futures roughly flat pre-market. Three-session-confirmed discipline applies: do NOT pre-assign direction — not mild-up, not tech-down. The memory/AI complex flipped to the strength leg overnight (Korea +5.9% on Samsung/SK Hynix) while US memory names are flat-to-red pre-market (MU ~-1.2%, WDC ~-1.9%) — the two-tier polarity is genuinely unknown into the open.
• VIX regime: 15.14 — LOW (<18). Both members' weakest backtested cohort (SURGE PF 1.36 vs 2.18 in VIX 20–30; ECHO's vol_ratio gate has blocked wide — 0.79–1.16 vs the 1.2 floor — across the recent LOW-VIX stretch). Structural headwind for both.
• Key levels: SPY prior close 769.09, prior range 768.18–772.47, SMA20 759.72, ATR(14) 6.65. QQQ prior close 716.03, prior range 712.66–721.39, SMA20 706.75, ATR(14) 10.88. XLE: oil firming (WTI ~$86–87, Brent ~$93) on Hormuz/UAE-Iran escalation — NOT an EIA catalyst today (the weekly EIA petroleum report released yesterday, Wed Aug 19; market has digested it).
• Macro backdrop: A record/consolidating base near all-time highs, LOW VIX, contained range — the tape's default is consolidation into the Aug 26 PCE/NVDA pivot unless a coin-flip catalyst fires. Energy is the cleanest trending leg (oil/Hormuz); the consumer/Dow leg reads constructive on WMT/TJX raises. Neither Family 2 expression has a defensible pre-market directional thesis into a two-tier, direction-unknown open.

ECHO — echo, dynamic universe (SPY / XLE / XBI + scanner candidates)

Setup present: no — No trade expected. Today is a LOW-VIX, two-tier, direction-unknown day, and the one strong directional candidate is exactly the leg I am pre-committed to stand down from.

1. XLE is the marquee directional candidate — and it is the high-risk leg. Oil is UP on a live, strengthening Hormuz/UAE-Iran driver (Brent ~$93, WTI ~$86–87). Fading the driver's own strengthening sector is the loss template that cost me 8/10 (XLE, -75.89) and 8/19 (XBI, -118.47) — now at N=2 in my own reflections. I do not reach for an XLE fade while the oil driver is live and strengthening. This is the single most important line in today's plan: the strongest vol_ratio/exhaustion reading of the day is the least tradeable for me, not the most.
2. SPY is the only other candidate — a default, and weak. Intrinsic catalyst is the weakest (no ≥±0.5% pre-market move, futures flat, direction unknown). A contained two-tier tape rarely produces a one-way program that survives to 15:20. LOW-VIX vol_ratio has blocked wide (0.79–1.16 vs 1.2) on precisely this session class.
3. XBI — not a candidate. No FDA approval/rejection decision published today. (Garetosmab PDUFA target is "August 2026" — could land any day including today, but none is announced; if an FDA decision drops intraday and re-rates the sector, that is a live-driver stand-down trigger, not an entry — exactly the 8/19 template.)
4. Memory complex (MU/SNDK/STX/WDC/NVDA) is NOT eligible — divergence, direction unconfirmed, no per-name scheduled catalyst → fails the intrinsic-catalyst rule.

8/19 learning applied in this plan, not deferred: my driver-sector-fade stand-down (8/10 → XLE/oil) failed to carry to XBI on 8/19 when a Moderna-type result made biotech the live strengthening driver intraday. Today I extend the rule in the plan to any leg whose sector becomes the day's live, strengthening catalyst — oil (XLE) pre-committed, and XBI if an FDA decision re-rates it intraday. This remains plan intent, not gate logic — I do not unilaterally change the live gate stack; the gate-level enforcement is the TEMPER conversation flagged in the 8/19 reflection. Sizing unchanged (0.75%/instrument, 2.0% session budget) — irrelevant today, as no entry is expected.

Forecast: p_trade 0.15, direction none, conviction low. The path to an entry would be a clean one-way SPY program (morning bias confirmed by all four gates at 15:20, contra entry) and no live strengthening sector driver — the mechanism's standing rejection on this tape. Invalidation: stand down entirely on the XLE fade while the oil driver is live/strengthening; stand down on any leg whose sector becomes the day's live, strengthening catalyst intraday (e.g., XBI on an FDA decision); no memory-complex single-stock fade; no forced fade on a two-sided/no-program tape. No trade is the base case and the disciplined call.

SURGE — surge, SPY + QQQ (fixed)

Setup present: no — No trade expected. Not an event day today (LOW event risk, no FOMC/CPI/NFP/PCE/QQQ-earnings in the hard-skip set), so my known event-day weakness does not bind. The binding obstacles are structural to this tape.

1. LOW VIX (15.14) — my weakest cohort (PF 1.36 vs 2.18 in VIX 20–30). Structural headwind.
2. The direction-unknown two-tier open denies a clean trend class. Three-session-confirmed: do not pre-assign direction. SPY ~flat and QQQ ~flat pre-market, both two-tier — the first hour is more likely a discovery drift than a clean three-phase establishment. On quiet LOW-VIX sub-ATR days the first-hour signals block wide on the ±0.30% floor (recent 0.02–0.29% values — a genuine no-impulse read, not a graze).
3. The pullback_resumption block is at N=6 trend-qualified sessions with no resumption bar (7/15, 7/28, 8/3, 8/4, 8/13, 8/19 — including 8/19's QQQ trend-qualified short at ADX 30.9 that never retraced). A resumption bar is a precondition, not an adjustable threshold. A contained two-tier day is the modal denial of a clean measured retracement-then-resumption.

I hold discipline: I do not force the three-phase structure onto a discovery tape, and I do not reach for a directional short/long on a day the firm explicitly refuses to pre-assign direction. The chain stays at 10:00 (signal ±0.30%, ADX ≥ 22, VWAP side + non-flat slope) → pullback-to-VWAP → resumption (10:00–12:00); single position; hard flat 14:00. Sizing unchanged (0.75% = $187.50) if — against the base case — a genuinely clean, decisive index trend with a confirmed resumption bar forms.

Forecast: p_trade 0.15, direction none, conviction low. No directional lean into a direction-unknown open. Invalidation: no-force on a price-discovery first hour (likely today); stand down if a trend-qualified first move never retraces (the N=6 block); no fragile chase into a two-sided/divergent tape; no forced three-phase structure onto a contained sub-ATR discovery day. No trade is the base case and the well-calibrated outcome.

Plan Filed

• Filed: 2026-08-20 pre-open desk run
• Frontmatter forecasts complete for every active member: yes (echo 0.15/none/low, surge 0.15/none/low)

Desk Reflection — House Echo — 2026-08-20 (Thursday)

Scope: since this desk's last reflection. Both active members (ECHO, SURGE) reflected today because ECHO traded (same-day-on-trade cadence). EOD briefing: dadbrain/Analysis/briefings/2026-08-20-eod.md

Session In One Line

A broad but contained down day (SPY -0.84%, QQQ -0.71%, both sub-ATR/contained, VIX 16.01 LOW) on the Treasury-buyback relief fading (yields +~10bp) and Walmart's U.S. comp miss (2.6% vs 3.8%, WMT -9%) dragging the consumer/Dow leg lower; ECHO took a second consecutive XBI trade against its own plan and was stopped out (-116.84); SURGE correctly stood down, its pullback_resumption block extending to N=7.

What Happened Across the Desk

ECHO — TRADED, stopped out (2nd straight plan-violation session)

• Long XBI 305 sh, entry 163.795 @15:30, stop/exit 163.4119 @15:40, net -116.84 (strategy v6, is_eia_wed=False, atr=0.2554). Mechanical qualification: morning signal -2.52% (short), so the long was the contra/exhaustion fade; vol_ratio 1.8387 (≥1.5 tier), ADX 20.69, exhaustion_score 4.6307. XBI ran down all morning on a broad down tape and ECHO faded it long into the close; the reversion never came and the stop hit at 15:40.
• Second consecutive plan-intent violation — must be stated plainly. The 8/20 plan explicitly called XBI "not a candidate" (no scheduled FDA decision), forecast no-trade (p_trade 0.15, direction none), and named a clean one-way SPY program as the only path-to-entry. ECHO instead traded XBI LONG. This follows 8/19, where the plan said no-XBI and ECHO shorted XBI. On 8/20 the specific 8/19 driver template (fading a live strengthening sector) did not fire — XBI was down, not being bid up — so this was a legitimate mechanism fade that lost on a broad down tape. But the plan/engine enforcement gap is real and recurring: the engine trades on gates, not plan text, and has now ignored a plan-level no-trade/no-XBI declaration two straight sessions. Both lost. XBI's live sub-book is now 0-for-2 (8/19 short -118.47, 8/20 long -116.84), the weakest leg in the book (dev WR 52.5%, already sub-benchmark).
• Forecast was also wrong, not just the execution: the plan under-priced XBI's mechanical firing — its exclusion was anchored on a scheduled FDA calendar, missing that XBI (morning -2.52%, vol_ratio 1.84) was the day's strongest exhaustible program regardless of catalyst. Brier on the no-trade call: 0.7225 — the desk's worst forecast of the recent window.

SURGE — NO-TRADE, gate stand-down (correct)

• SPY blocked on pullback_resumption dir=short signal=-0.41% adx_1000=22.8 — trend-qualified short but no VWAP-touch pullback resumed within the window; QQQ blocked on ADX val=16.5 floor=22.0 (signal -0.68%). The pullback_resumption block is now N=7 trend-qualified sessions (7/15, 7/28, 8/3, 8/4, 8/13, 8/19, 8/20) with no resumption. Stand-down was the forecast outcome (p_trade 0.15) and the correct one — a contained two-tier day never formed a clean three-phase structure. Brier 0.0225, well-calibrated.

Binding Gate This Period — Regime vs. Calibration

• SURGE: regime, no fix. The binding gate was pullback_resumption (now N=7) plus QQQ ADX 16.5. A trend-qualified first move that never retraces on a LOW-VIX two-tier tape is structural absence, not a threshold graze — the resumption bar is a precondition, not an adjustable threshold. No gate misbehavior. No TEMPER conversation warranted on any SURGE gate. This window strengthens the session-character (regime) reading on record since 2026-08-14.
• ECHO: calibration-shaped, not a threshold tweak. No gate blocked — all four cleared at high confidence; the mechanism is not under-firing, it is being given trades the plan ruled out. The load-bearing finding: a plan-level stand-down is not enforced by the engine, now demonstrated two consecutive sessions (8/19, 8/20), and XBI — the leg the plan keeps excluding — is 0-for-2 live. This is evidence for the TEMPER conversation already flagged on 8/19 (wire plan-level / driver-sector stand-downs into the gate logic, or let a high-conviction XBI fade be excluded), and it feeds the pre-registered XBI exclusion review (WR < 50% over 15+ live trades; currently n=2, not triggered). It does not warrant any unilateral gate tweak by this desk (no-unilateral-tweak rule).

Desk-Level Learnings Flag

• Flag for learnings.md (desk-blind, phrased about this desk's own members): ECHO's plan/engine enforcement gap has produced a second consecutive session where the plan forecast no-XBI/no-trade but the engine traded XBI (8/20, -116.84; 8/19 was the first), and XBI's live sub-book is now 0-for-2 (8/19 short -118.47, 8/20 long -116.84). Two distinct things are true: (a) ECHO's XBI exclusion was anchored on a scheduled-FDA calendar and under-priced XBI's mechanical firing (it was the day's strongest exhaustible program), and (b) a plan-level stand-down is still not enforced by the engine — the protection committed to after 8/10 remains plan intent, not gate logic. This strengthens, not closes, the TEMPER conversation flagged on 8/19; and XBI is trending toward the pre-registered exclusion-review trigger. No unilateral change.

SPRT Monitors (desk view)

• ECHO n=8 W4/L4 LLR +0.184 CONTINUE (boundary distance +2.761 to CONSISTENT; -3.128 to DEGRADED). XBI sub-book xbi_leg n=2, WR 0.0%, PF 0.0 — not triggered (needs 15+ XBI live trades), but the desk should note this is pointed at the kill-switch boundary.
• SURGE n=6 W4/L2 LLR +1.425 CONTINUE (boundary distance +1.519 to CONSISTENT; -4.370 to DEGRADED).

Reflection Filed

• Filed: 2026-08-20 (desk reflection, same-day-on-trade cadence)
• Next session event risk (from EOD briefing tomorrow's calendar): NONE tier-1 scheduled within 7 days (PCE 2026-08-26; NVDA earnings 2026-08-26; Jackson Hole Aug 27–29 forward-look).