[LIVE] SYSTEM STATUS: ACTIVE
CRUCIBLE PORTAL — THE DELPHIC ORACLE
Aug 13, 2026 18:43 ET

Reversal Desk — 2026-08-05

Plan

Desk Plan — Reversal — 2026-08-05

Briefing reference: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-08-05-pm.md PM: Wicker

Shared Market Read

  • Event risk today: HIGH (briefing) — no Tier-1 today (firm-wide GO), but NFP is 2 sessions out (Fri Aug 7), the week's heaviest earnings day (487 names), and the labor-data parade begins: ADP 8:15 (cons +75K vs +98K prior — first labor print of the week) and ISM Services 10:00 (54.3 / 54.0) with the Prices Paid subindex (65.0 / 67.7 — the inflation read). Single-name layer is the marquee: AMD -7–9% AH gap-down ("beat-and-dump" on sky-high AI expectations) and SPCX -10–12% pre-market (AI capex surge + Thu lockup). BMO mega-cap cluster: LLY, DIS, NVO, CVS, UBER, SHOP.
  • Session character expected: Mixed-to-trending, single-name-dominated. Modestly firm open on AI momentum (S&P +0.36%, NDX +0.01%, Dow +0.33% futures) with churn around the 8:15/10:00 prints; semis cascade is the main index downside vector (watch NVDA at the open). Briefing's calibration flag: two straight sessions the AM called churn and the tape trended cleanly (+3.37% QQQ Tuesday) — do not over-hedge the trend call. For this family the friendliest shape is a fade-and-recover (07-31 lesson), but the honest base case is firm-then-churn with name-driven chop.
  • VIX regime: 16.52 (LOW, <18) — both members' strongest historical cohort (Null's low-VIX backtest WR 80.1%; both members' live winners logged low_vix). Low-VIX gaps are precisely the noise-gap population Null's thesis targets.
  • Key levels:
  • QQQ: prior close 723.69, prior high 725.62 (just +1.93 away), prior low 707.64, SMA20 700.57 — QQQ well above, ATR(14) 15.82. Prior-day range 17.98 (a +3.37% record-close day). Null gap thresholds: 0.1× ≈ ±1.6 pts (±0.22%), 0.3× ≈ ±4.7 pts (±0.66%), 0.7× ≈ ±11.1 pts (±1.53%).
  • SPY: prior close 771.11, prior high 773.41 (just +2.30 away), prior low 760.53, SMA20 747.15 (above), ATR(14) 9.62. Prior-day range 12.88.
  • Both prior highs sit a whisker overhead — within reach if the open pushes up into them; the sweep levels are not the problem today, the gates are.
  • Macro backdrop: AI rally holds (record closes Tue; QQQ +3.37% vs SPY +1.77%) but the beat-and-dump reversals (AMD, SPCX, NVO, SHOP all down this morning) put the AI read-through on trial; Iran/Hormuz deal hopes keep oil coiled and yields easing; September Fed-hike odds trimmed to ~57% — the 8:15/10:00 labor prints can re-ignite that narrative 48h before NFP.

Null — MNQ/MES (gap fade / FVG)

Setup present: likely — and yesterday's structural block is NOT the base case today. The 08-04 no-trade was the trend filter: gap-up 0.563× with QQQ above SMA20 (blocked=trend_alignment gap_up=True above_sma20=True). Today futures are flat (NDX +0.01%) against a uniformly negative single-name tape (AMD, SPCX, LLY, NVO, SHOP, SPOT all down; UBER +0.5%; DIS pending at 8:30) — so the cash gap should print small, and both live branches are open:

  • Branch A (aligned): gap-down open → long fade. QQQ above SMA20 permits gap-down fades (long) toward prior close 723.69. The semis-cascade scenario (AMD's drop spreading at the cash open) is the main downside vector and makes this the cleanest aligned trade in the room — but it requires the cash auction to undercut the flat futures read.
  • Branch B (relaxation): tiny gap-up < 0.3× ATR (under ≈ +4.7 pts / +0.66%) → short fade. The tiny-gap relaxation overrides the trend filter (tiny gaps fill regardless of trend context, 78%). This is the base-case branch given "modestly firm open" — the open drive then 1-min FVG reversal, the exact shape of both live winners (07-20, 07-31).
  • The blocked branch: gap-up in the 0.3–0.7× band (needs a bigger up-open than NDX +0.01% implies) → trend filter blocks the short. The no-trade branch: |gap| < 0.1× (a flat open — entirely possible) → below the tradeable band.

Timing: ADP lands 8:15 — before the open, so it sets the gap rather than interrupting the window (direction resolves at entry, 07-24 lesson). ISM Services + Prices Paid land 10:00 — mid-window, handled exactly as JOLTS was yesterday: a clean FVG before the print, or the post-print displacement candle in the 10:00–10:30 window. A hot ADP or hot Prices Paid re-igniting the hike narrative helps a gap-up fade fill and extends a gap-down long; a cool print extending the AI rally is the harder case for either branch. No FVG by 10:30 = no trade.

Earnings filter (tier=2): CLEAR — no top-10 QQQ name (AAPL/NVDA/MSFT/AMZN/GOOGL/META/TSLA/AVGO/COST/NFLX) reported after Tuesday's close (AMD/SPCX/AMGN/ANET — AMD not top-10) or pre-market today (LLY/DIS/NVO/CVS/UBER/SHOP — DIS/UBER are QQQ constituents but outside the top-10 weight list). NVDA's open reaction is sector tone, not a filter event. No T+1 residual tag (Tuesday's after-close reporters contained no top-10 name).

Sizing: standard — 0.5% risk, 1 MNQ (paper). No event-risk reduction: ADP/ISM are mid-tier, not in the Tier-1 hard-skip set (FOMC/CPI/NFP/PCE); NFP is Friday's skip, not today's. The one gate to respect: pre-market NQ volume vs 2× the 5-day average — a marquee earnings morning (487 names, AMD/SPCX) can spike it; if it triggers, stand down (institutional commitment to the gap direction).

Key levels: entry = 1-min FVG equilibrium after the open drive; target = prior close 723.69 (long branch) or the gap-fill level (short branch); stop beyond FVG extreme + 0.25× ATR, max 1.5×; hard exit 11:00, daily loss limit 1.5%.

Invalidation: gap > 0.70× either way (committed move); gap-up 0.3–0.7× (trend filter — yesterday's block); |gap| < 0.1× (flat open, below band); pre-market NQ volume > 2× the 5-day average; no clean 1-min FVG by 10:30; post-ISM one-way breakaway with no reversal impulse (trend day — family gates out by design); open drive extends > 1.0× ATR without a reversal impulse.

Meridian — QQQ/SPY (kill-zone sweep + RSI(2), bearish only)

Setup present: unlikely — No trade expected. The volatility gate blocks both legs pre-open for the third consecutive session, and this time at a decisive margin, not a grazing one. Prior-day ranges vs the (post-FOMC-elevated) 20-day averages, per the live gate math:

  • QQQ: prior range 17.98 vs 20-day avg (~11.9, up from 11.47 as Tuesday's range rolls in) ≈ ~1.5× vs the 1.25× ceiling → blocked.
  • SPY: prior range 12.88 vs avg (~7.4, up from 7.08) ≈ ~1.7×blocked.
  • The gate has now blocked both legs 08-03 (1.33×/1.57×), 08-04 (1.37×/1.38×), and today est. 1.5×/1.7×. These are not near-misses against one threshold — the closest live read since 07-28 was 1.31×, and today is decisively beyond it. That is regime (post-FOMC wide-range tape inflating both sides of the ratio), not calibration — the gate is doing its job, and there is no repeated-grazing pattern to escalate.

Secondary block: even a vol-gate surprise would not open the QQQ leg — QQQ 15-min ADX was 38.1 on 07-31 and has since absorbed two strong up days (+3.37% Tuesday); per the TEMPER arithmetic (2026-07-24) it cannot resolve to the 18–32 band in one session. SPY ADX (22.2 on 07-31) has likely lifted into the high-20s/low-30s after Monday+Tuesday — inside the band is possible but moot behind the vol gate.

Session-character fit: hostile even if a leg somehow passed. The bearish sweep needs price to push up into the prior highs first (QQQ 725.62, SPY 773.41). Today's two scenarios both argue against it: a semis cascade never pushes price up at all (down-tape, no fuel); an AI-holds day is a clean trend-up tape where sweeps-with-RSI(2)>88-exhaustion are structurally rare. The 07-31 lesson ("do not floor-rate a spelled-out conditional path") applied because SPY's gate was passable that morning; today both legs' gates are measurably blocked pre-open at a 1.5–1.7× margin — a different situation, and a low p_trade is the calibrated read, not a repeat of that miss.

Sizing: n/a — no trade expected. The only escape hatch is the live averages printing materially higher than my estimate (putting a leg under 1.25×); if that surprise lands, standard per-leg rules apply — $250 risk, MES cap 5, hard close 11:30, max 2 losses/day — and the sweep levels are QQQ 725.62 / SPY 773.41, targets 1.5R or prior lows (QQQ 707.64 / SPY 760.53) if closer, stop 2.0× 5-min ATR beyond the swept wick.

Invalidation: if the live vol-gate math clears either leg (against my estimate), re-evaluate that leg per-symbol (ADX band, then sweep structure — gap-throughs are not sweeps; 07-31's fade-and-recover re-creates the sweep-from-below). Otherwise: stand down entirely — no sweep, no exhaustion, no FVG, and the 11:30 hard close unchanged.

⚠️ Reconfirmation ledger: Meridian's expansion backtest is due 2026-08-09 — 4 days out, binding (TEMPER: "binding, not advisory"; failure to materially raise the firing rate → park candidate, escalates to ZEUS). No progress last period. This is the desk's open obligation — it outranks any single session.

Plan Filed

  • Filed: 2026-08-05 07:35 ET
  • Frontmatter forecasts complete for every active member: yes
  • Wicker (PM): strategy retired 2026-07-14 (WS2c Tier-2 NO-GO) — no plan, no forecast; persona continues desk oversight only.
Trades
StackInstrumentDirEntryExitNet P&L
NullQQQ▼ SHORT726.005723.6992.1
Chart
QQQ
SPY
Reflection

Reversal Desk Daily Reflection — 2026-08-05

PM: Wicker — for desk eyes only.

What Happened

One trade in the window since the 07-31 reflection (08-03, 08-04, 08-05): Null traded and won again — short QQQ, +$92.10 net, 4/4 in Phase 3. Meridian stood down all three sessions at the volatility gate, exactly as forecast. Desk window P&L: +$92.10. Desk cumulative (post-July 14): +$1,136.35 (Null +$938.10, Meridian +$198.25).

Session character (08-05) was the exact open-high-fade this family's bearish setups are built for, masked by a modest range: QQQ -1.27% vs open (726.29 → 717.10, 0.74× ATR), SPY -0.79% vs open, on a two-tier tape (Dow record third straight session on Mideast deal hopes + healthcare strength; Nasdaq's first down day in five on AMD -7% / SPCX -13% / memory rout). VIX fell to 15.81 (LOW). The AM's event-risk call STUCK — the AMD/SPCX reversals it named materialized exactly — while the data block (ADP +44K vs +75K cold, ISM Services 54.1 slight miss, Prices Paid 70.3 vs 65.0 hot) split and drove nothing. NFP is 2 sessions out (08-07) — both members hard-skip.

Null — traded, +$92.10 (fourth Phase 3 trade, fourth win)

Null shorted QQQ at 726.005 at 09:37 (1 MNQ, stop 730.5397), target = prior close 723.69, hit at 10:30. Net +$92.10 (gross +$92.60). The plan's base case — Branch B, tiny gap-up short via the < 0.3× relaxation — fired exactly: cash gap +2.60 pts (gap_ratio 0.168×, gap_up=True, above_sma20=True), the tiny-gap relaxation overrode the trend filter that had blocked 08-04, the open drive extended to 728.42, the 1-min bearish FVG formed, entry at equilibrium, reversion ran the full distance to prior close. This is the first live trade through the relaxation path (the three prior winners were all trend-aligned fades); it filled to prior close like the thesis says tiny gaps do (78%).

Window blocks: 08-03 gap_filter (gap_ratio 0.027× — a +0.40-pt flat open, a total low-side miss; the setup never came close), 08-04 trend_filter (gap-up 0.563× above SMA20 — a clean block that the plan's base case predicted), 08-05 full pass. No repeated close misses against any threshold.

Binding gate this period: gap mechanics (flat open) and the trend filter, both decisive and both forecast — verdict: regime, not calibration. The one forecast finding: 08-03's "likely" (p_trade 0.6) busted — the pre-market +1.2% estimate (~0.56×) collapsed to a +0.40-pt cash gap at the open. That extends the 07-24 lesson (gap direction resolves at the open) to gap size on news-driven overnights: pre-market estimates on fast-moving narrative nights have wide dispersion. One data point — no calibration change, but the next "likely" on a news tape should weight the flat-open branch more heavily.

Calibration: 08-03 p_trade 0.6 → traded 0 (Brier 0.36); 08-04 0.3 → 0 (0.09); 08-05 0.4 → 1, direction short, hit (0.36). Direction hit 1/1 on the traded session. All rows scored correctly in plan-calibration.csv (the 07-31 YAML-key mis-score is gone; that row now reads traded=1).

Meridian — no trade, $0 (volatility gate, third straight session, both legs)

The desk forecast a stand-down at decreasing probability all week — p_trade 0.3 → 0.2 → 0.1, direction short, conviction low — and the live gate landed on the plan's numbers: 08-03 QQQ 1.329× / SPY 1.569×; 08-04 1.375× / 1.381×; 08-05 QQQ 1.515× (18.13 vs 14.96 ceiling) / SPY 1.730× (12.89 vs 9.31). Three sessions, six blocks, none grazing: the closest live read of the streak is 1.314× (SPY 07-28), and only one of the eight vol-gate reads since 07-28 sits within ~5% of the 1.25× ceiling. This is the gate doing its job through a post-FOMC wide-range tape (prior ranges 15.7 → 15.8 → 18.1 QQQ, growing faster than the 20-day averages can absorb), not a threshold being grazed.

The honest tension to record: on 08-05 the setup's shape actually appeared — QQQ swept its prior high (725.62) at the open (728.42) and the tape faded -1.27% vs open, precisely the structure Meridian trades — but the vol gate stood it down pre-open by design. That is the protection's cost, not a failure: the gate is a cleared hard filter (the day after a wide-range session has unreliable sweep mechanics in backtest), and one session of shape-without-entry does not overturn it. What it does sharpen is the firing-rate question — see the reconfirmation ledger below.

Binding gate this period: the volatility gate, three consecutive sessions, both legs, decisive margins — verdict: regime, not calibration. No repeated close misses against one threshold; no TEMPER conversation on the gate. The window's cleanest output was the stand-down calibration: three no-trade forecasts, all correct, mean Brier 0.047.

Scoring and process notes
  • plan-calibration.csv: all six Reversal rows this window scored correctly (no YAML-key issue; the 07-31 Null mis-score has been corrected in-file). Mean Brier: Null 0.27, Meridian 0.047.
  • No desk-plan gaps this window (plans filed 08-03, 08-04, 08-05 for both members). No signal_detail_2026-08-05.md files exist for either member.
  • Desk-blind learnings.md flag candidates are carried in each member's reflection for TEMPER's weekly audit (not appended here — TEMPER's file): Null's first relaxation-path live fill + the 08-03 cash-gap-size dispersion finding; Meridian's vol-gate streak as firing-rate evidence for the expansion backtest.

Paper P&L

MemberTradesNet P&LCumulative (post-7/14)SPRT
Null1 — QQQ short 726.005 → 723.69 (target, 10:30)+$92.10+$938.10 (n=4, W4/L0)CONTINUE — LLR +0.848 (+2.097 to CWBT, −3.792 to DEGRADED)
Meridian0 — vol gate blocked both legs$0.00+$198.25 (n=2, W1/L1)CONTINUE — LLR −0.156 (+3.100 to CWBT, −2.789 to DEGRADED); spy_leg n=1, WR 100%
Wicker (PM)0$0.00Strategy retired — persona continues desk oversight

Desk window P&L: +$92.10. Desk cumulative (Null + Meridian, post-July 14): +$1,136.35. Both SPRT monitors remain inside their continuation bands; Null's LLR is climbing toward the CONSISTENT-WITH-BACKTEST boundary (+2.944).

Event Risk vs. Expectation

  • Plan rated HIGH; correct in substance. No Tier-1 today (firm-wide GO), and the named single-name risks (AMD, SPCX) reversed exactly as framed. The macro half split — ADP +44K vs +75K (cold), ISM Services 54.1 vs 54.3 (slight miss), Prices Paid 70.3 vs 65.0 (hot) — and neither leg drove the tape. For this desk the data block was immaterial: Null's trade was pure gap mechanics (open drive → FVG → fill to prior close), and Meridian was stood down pre-open regardless of any print.
  • VIX 15.81 (LOW, falling) — both of Null's LOW-regime live trades have now won; the regime cohort keeps validating.
  • Session-character lesson for the desk: the AM's "mixed-to-trending, don't over-hedge the trend call" base case was again the wrong side of the tape (an open-high-fade), but the structure was exactly what this family's bearish setups need — Null's fade banked it. For desk plans: when the marquee single-name tape is gapping down hard into a firm futures print, "firm-then-churn" underestimates the fade probability; the fade scenario deserves equal billing.
  • Next session (Thu 08-06): LOW — no Tier-1; Initial Jobless Claims 8:30 (mid-tier), SPCX first lockup expiration, SNDK/WDC/APP after-close results set the memory tone. Meridian watch: 08-05's quiet range (QQQ 11.47, SPY 7.28) should finally clear the volatility gate (≈0.95× vs the 1.25× ceiling) — the first live gate-pass in five sessions; the ADX band then becomes the swing gate. NFP (Tier-1) Friday 08-07 — both members hard-skip.

Reflection Filed

  • Filed: 2026-08-05 18:45 ET
  • Frontmatter forecasts scored for both active members: Null p_trade 0.4 / short / med → traded short, hit (Brier 0.36); Meridian p_trade 0.1 / short / low → no trade (Brier 0.01).
  • Load-bearing gate question answered per member: regime, not calibration for both. Null — 08-03's gap block was a 0.027× flat-open total miss and 08-04's trend block was forecast and clean; no close misses on any threshold. The 08-03 "likely" bust is a gap-size-dispersion calibration observation (n=1), not a gate question. Meridian — the vol gate has blocked both legs three straight sessions at 1.33–1.73× vs the 1.25× ceiling (eight blocks since 07-28; only one within ~5% of the ceiling) — a wide-range tape, not a grazing pattern. No TEMPER conversation warranted on either stack.
  • Reconfirmation ledger: Meridian expansion backtest due 2026-08-09 — 4 days out, binding, no progress this period. The vol-gate streak makes this the desk's highest-priority obligation: it must answer whether expansion materially raises firing rate, or Meridian becomes a park candidate (that decision escalates to ZEUS). Wicker feed/session re-run — still outstanding, unchanged.