Briefing reference: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-08-28-pm.md PM: Wicker (persona; own stack retired 2026-07-14 — plan covers active members Null + Meridian only)
Yesterday (08-27) was a corporate-repricing day: Null deterministic NO-GO (NVDA earnings filter), Meridian degraded-by-gap. Today the roles flip. The single live risk is Fed Chair Warsh's Jackson Hole keynote at 10:00 AM ET (a speech, not a decision → MEDIUM per calibration), the AI leg that led Thursday (+9% NVDA) is giving back pre-market, and futures are flat-to-modestly-down with a Dow-up/Nasdaq-down rotation toward value-defensiveness. That landscape:
One read, two diverging calls. Do not blur them.
Setup present: possible-to-likely, contingent on a gap-DOWN open — the with-trend LONG cohort. Null's window is open: no Tier-1 macro today (Warsh is a speech, not Tier-1), and none of yesterday's after-close reporters (MRVL, ADSK, WDAY, AFRM, ULTA, ESTC, RBRK, GAP) is a QQQ top-10 name, so the mega-cap constituent earnings hard filter does not fire. QQQ is above SMA20 (721.20 vs 716.49), so the trend-alignment rule admits only the gap-down fade → LONG (the cohort that won live 08-06 and 08-20; the 08-27 gap-up was correctly skipped as counter-trend). Pre-market Nasdaq futures dipping / AI give-back makes a small gap-down the lean — but IEX pre-market is thin with no reliable gap print yet, and per the TEMPER 07-24 lesson gap direction resolves at the cash open, not pre-market, so this is a live-trigger day, not a pre-committed one.
The honest gate stack for today: gap must land in-band (0.1–0.7× ATR ≈ 0.8–5.6 pts on QQQ), gap must be down (up → counter-trend short, skip), pre-market volume ≤ 2×, then a clean bearish-reversal 1-min FVG off the open drive within 9:30–10:30. The one new consideration is the 10:00 Warsh event sitting inside Null's 9:30–11:00 window: the reversion typically fills in the first ~30 min (before the speech), and that is the strong case; a setup that only appears close to 10:00 carries speech-shake risk. Sizing: standard (risk 0.5%, one position) — no event-risk reduction warranted on a speech, but favor the entry that fills before 10:00 and treat the keynote as a hard reason not to hold a weak open position into it; the 11:00 time exit is unchanged. Key levels: entry off the reversal FVG; primary target the gap fill at prior close 721.20 (prior low 714.59 the stretch); stop beyond the FVG. Invalidation: gap-up at the open (skip), gap > 0.7× ATR (breakaway, no trade), pre-market volume > 2×, no qualifying FVG by 10:30, or the 10:00 speech producing a breakaway-type move that kills the fill thesis. p_trade 0.40, direction long, conviction med.
Context/calibration: SPRT CONTINUE (n=9, W9/L0, 100% WR, LLR +1.907); 9 of the first 30 on the 2026-WR watch (≤60% triggers review — currently 100%). Last live trade 08-25 was the with-trend gap-up short, ninth straight win (+$246.80). The 08-26 PCE and 08-27 trend-filter days were correctly flat. Today the with-trend long is the live cohort.
No trade expected — and the likely binding gate is Meridian's own Step-1 volatility check, before any sweep logic runs. Yesterday (08-27) was an above-ATR session for QQQ (1.26×) and SPY (1.34×), the direct input to the "prior-day range > 1.25× 20-day average → no trades" gate — the same pattern that vol-blocked QQQ/SPY on 08-03/04/05 after a wide prior day. This is a read pending live gate values (I don't hold the exact 20-day average-range series to confirm the 1.25× breach), but it is the expected blocker for both index legs.
Even if the vol gate passes, the session character denies the setup. Meridian needs price to drive up into the prior high and reject with RSI(2) ≥ 88 on a trend-character tape inside the ADX 18–32 band. On a pausing two-tier tape drifting lower ahead of the 10:00 speech, an upside drive-to-sweep is structurally improbable — the 06-18 lesson ("reachable in principle ≠ likely to be reached") applies verbatim: a down-drifting/pre-event-pin tape is the wrong structural backdrop for an upside sweep. Hawkish Warsh = a downside move without a prior upside sweep; a no-guidance reiteration = muted relief. Either path fails to produce the one-sided upside over-extension the setup requires, and a bearish-only mechanism shorting into a coin-flip speech is exactly the wrong risk on a pre-assigned-direction-free day. IWM (the thin n=15 expansion leg, honest live anchor ~56%, ADX in-band 20.3 yesterday) is the only theoretically-live leg, but on a value-led pause a small-cap upside sweep is unlikely and its live expectation is thin. Sizing: none expected; if every gate somehow cleared on a leg before 10:00 (vol passes, ADX in-band, upside sweep + RSI(2) ≥ 88), standard $250/leg but flag event proximity — do not loosen the sweep/RSI/ADX/vol stack to manufacture frequency (standing rule). Key levels: prior highs QQQ 721.32 / SPY 772.32 / IWM ~299.95 (per open). Invalidation: a correctly-filtered no-trade is the base case; stand down entirely on the pre-event pin. p_trade 0.10, direction short, conviction low.
Context/calibration: SPRT CONTINUE (n=2, W1/L1, LLR −0.175); spy_leg n=1 WR 100%; iwm_leg n=0. The 08-25 QQQ 5-sweep/no-RSI-exhaustion (max 46.5 vs 88) was regime, not calibration; the exhaustion confirm does real filtering — that watch continues. Today the pre-event pin is the structural block.
| Stack | Instrument | Dir | Entry | Exit | Net P&L |
|---|---|---|---|---|---|
| Null | QQQ | ▲ LONG | 719.155 | 721.2 | 81.3 |
| Meridian | QQQ | ▼ SHORT | 720.94 | 718.5904 | 186.97 |
PM: Wicker (persona; own stack retired 2026-07-14 — desk covers active members Null + Meridian only) Briefing reference: dadbrain/Analysis/briefings/2026-08-28-eod.md Period covered: 2026-08-26 (prior desk reflection 2026-08-25) through 2026-08-28.
A three-session period built around the week's binary cluster, resolving in a clean, both-members-win Friday. Net desk paper P&L over the period +$268.27 — all of it on 08-28, with two hard-skip no-trade days ($0.00) on 08-26/08-27.
gap_ratio=0.135 (in-band 0.1–0.7×), gap_up=False, above_sma20=True, VIX 14.4 (LOW). Tenth straight win. SPRT now n=10 W10/L0 LLR +2.119 CONTINUE (+0.825 to CONSISTENT-WITH-BACKTEST). Cumulative +$1,885.70, WR 100% (10/10) — 10 of the first 30 on the 2026-WR watch (≤60% triggers review; currently 100%).spy_leg n=1 WR 100%, iwm_leg n=0. Cumulative +$385.22, WR 66.7%.Net desk paper P&L for the period: +$268.27 (both wins on 08-28; $0.00 on the two hard-skip days).
A binary cluster that resolved in two legs, then a clean fadeable Friday. 08-26 was the PCE Tier-1 decision day + NVDA after close — a firm-wide NO-GO where neither mechanism has a live window (the exact event-day-protection case each gate was cleared to handle). 08-27 was the NVDA repricing morning: a deterministic NO-GO for Null (mega-cap earnings filter) and a structurally-degraded open for Meridian (the +1% gap consumed the prior-high sweep trigger). Friday (08-28) was the play: a choppy/ranging two-tier tape on the hawkish-Warsh Jackson Hole keynote (10:00 ET) — QQQ -0.66% vs SPY -0.25%, Dow ~flat, VIX 14.43 (LOW). SPY range printed 1.38× ATR (a genuinely wide two-way band around the speech) but the close was small with a flat last hour — churn, not trend. That tape handed each member exactly one clean fade:
Both members captured a different side of the same two-sided keynote churn. Desk plan calibration was strong on the trade day: Null forecast 0.40 long (traded, direction hit, Brier 0.36); Meridian forecast 0.10 short (traded, direction hit, Brier 0.81 — underforecast, see below).
| Member | Period trades (P&L) | WR (cumulative) | Net P&L | SPRT LLR | SPRT status |
|---|---|---|---|---|---|
| Null | 1 (08-28 +$81.30) | 100% (10/10) | +$1,885.70 | +2.119 | CONTINUE |
| Meridian | 1 (08-28 +$186.97) | 66.7% (2/3) | +$385.22 | +0.222 | CONTINUE |
Desk net P&L for the period: +$268.27 (both wins on 08-28).
Null's binding gates this period were two deterministic hard-skips, not gates being grazed:
is_event_day(tier_filter=2) closed the session before any entry logic ran. No gap was even measured. Correctly forecast p_trade 0.0, Brier 0.0.gap_ratio=0.135, small but not below the 0.1 floor) and the gap-fill target (prior close 721.2) filled in 4 minutes. The plan's 0.40 long was well-calibrated (Brier 0.36).The load-bearing answer: no binding-gate problem exists for Null this period. The two skip days were the hard-skip machinery working exactly as cleared (event days, post-earnings single-name repricing), not a threshold being grazed; the one tradeable day produced a clean with-trend win. No calibration action, no unilateral tweak. The 11:00 time-exit discipline wasn't even needed today (target hit at 09:40). The bookended SMA20 over-forecast rule flagged on 08-21/08-24 did not recur — Friday's long forecast was honest and correct.
This is the one genuine calibration note of the period, and it is not a gate-calibration issue:
The load-bearing answer: this is a plan-forecast under-calibration, not a regime block and not a gate-threshold graze. No gate misbehaved — the vol/ADX/sweep/RSI stack admitted a good trade (this is the mechanism working). The miss is the desk's p_trade being systematically too low on Meridian: this is now the second live Meridian trade this month (07-31 p_trade 0.15, 08-28 p_trade 0.10) taken on a day forecast ≤ 0.15, both winners, both poor Brier (0.7225, 0.81). At N=2 that is a monitoring note, not a threshold being grazed and not yet a TEMPER conversation — but the desk's Meridian forecasts are carrying less information than they should. Flag for learnings.md (desk-blind): Meridian has traded on 2 of the 18 scored sessions this month, and on both live-trade days the desk under-forecast p_trade ≤ 0.15; the Step-1 volatility concern has now been over-called (08-28 it did not bind). Monitor the pattern; do not loosen any gate. No unilateral tweak.
| Dimension | Null | Meridian |
|---|---|---|
| Traded? | Yes (08-28, +$81.30) | Yes (08-28, +$186.97) |
| Binding gate | None (hard-skips 08-26/27 worked; clean with-trend long 08-28) | None gate-wise — the expected Step-1 vol block didn't fire; plan under-forecast (0.10) |
| Near-miss reading | regime/no-action (no graze) | not regime, not gate-calibration — plan-forecast under-calibration (N=2) |
| Structural vs. transient | Structural — hard-skip machinery + with-trend cohort | Transient plan miss — vol gate passed; mechanism fired a good sweep |
| Action needed? | No | No gate change; monitor the ≤0.15 under-forecast on live-trade days |
Narrative finding: The desk's two members again told the same two-sided-keynote story from opposite sides — Null faded the gap-DOWN back to the prior close, Meridian swept the prior high and shorted the rejection. Both were with-regime reversal fades on a churning keynote tape, both filled their targets inside the morning window (09:40 / 10:06), neither held into event risk (Null's 11:00 time-exit; Meridian's target hit ~6 min after the speech). The period's single calibration blemish is Meridian's under-forecast — the plan predicted a Step-1 volatility block that never materialized, and the mechanism delivered its cleanest live sweep-with-exhaustion of the month. Null's calibration across the full period was excellent (0.0 / 0.0 / 0.36 Brier).
iwm_leg sub-book monitoring (carried). n=0; IWM ADX 14.4 off-band today (no surface). Watch for a coherent small-cap level on a non-event session.agents/null/reflections/daily_2026-08-28.md, agents/meridian/reflections/daily_2026-08-28.md (wicker retired — no per-member file)