Reversal Desk — 2026-08-31

Desk Plan — Reversal — 2026-08-31

Briefing reference: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-08-31-pm.md PM: Wicker (persona; own stack retired 2026-07-14 — plan covers active members Null + Meridian only)

One market read — a live geopolitical tape, and the SMA20 flip sets both members' cohorts

The weekend US–Iran / Strait-of-Hormuz escalation (first US military action against Iran in ~a month; Iran launched a response) reopened the Hormuz supply-risk that ran all August. Oil gapped up sharply (Brent +2.2% to ~$90, WTI ~$85); US futures modestly lower to flat — a live-catalyst morning, not a panic gap. No Tier-1 today (Dallas Fed regional print 10:30, T-bill auctions 11:30 — both LOW; NFP Sep 4 is 4 days out). The decisive structural fact for this desk is the SMA20 flip on QQQ: Friday's soft-down close put QQQ below its 20-day SMA (716.44 vs 717.92) for the first time since the mid-Aug push, while SPY barely holds above (769.28 vs 769.18). That flip changes Null's admitted fade cohort (now the gap-UP short) and, combined with the oil-inflation risk-off backdrop, sets the session shape for both.

Shared Market Read

• Event risk today: MEDIUM — a live geopolitical clock, not a scheduled macro event. The two-sided escalation: a de-escalation headline (Qatar-mediated diplomacy still alive) would reverse the oil bid as fast as the strike started it. No Tier-1 within ~2 days (next: NFP Sep 4). Firm-wide GO with a live-catalyst, two-sided tape.
• Session character expected: live two-sided geopolitical tape — energy/defense-led rotation (XLE/XOM/XAR bid) against a mild broad-market risk-off tone; equities futures modestly lower. Choppy, event-driven, wide ranges expected; two-tier behavior (energy/defense vs rate-sensitive growth). Established coin-flip discipline applies: do NOT pre-assign broad-index direction — weight the downside (oil-inflation shock, risk-off) at least equal to any single-name pre-market bid.
• VIX regime: 15.23 (LOW, <18) — up +0.80 on the geopolitical pop but still under the 18 threshold; both members' strongest historical cohort. Regime that has structurally suppressed setups all month; not the binding issue today.
• Key levels (authoritative = pm briefing, Alpaca/IEX):
• QQQ: prior close 716.44, prior-day range 715.15–724.11, SMA20 717.92 — BELOW SMA20 (flip), ATR(14) 8.31.
• SPY: prior close 769.28, prior-day range 768.31–775.29, SMA20 769.18 — just above SMA20, ATR(14) 5.05.
• Critical structural facts for this desk: (1) QQQ closed below SMA20 — Null's trend-alignment filter now admits only the gap-UP fade → SHORT (the 08-27-gap-up/short cohort, not the with-trend long he won 08-06/08-20/08-28). (2) QQQ's prior-day range was 1.08× ATR and SPY's 1.38× ATR — neither a clear Step-1 volatility breach for Meridian given 08-28's 1.34× SPY passed (see Meridian section; do not repeat the 08-28 Step-1 over-call).
• Macro backdrop: a LOW-VIX, near-record tape gapping on an oil-supply shock against the Warsh-elevated-yield backdrop (10-yr ~4.7%), with the last data week before the Fed's Sep 16 decision and NFP Friday the pivot. One sentence: a range-bound index tape reopening an oil-inflation/risk-off geopolitical clock, with QQQ's SMA20 flip resetting both members' directional cohorts.

Null — MNQ/MES (gap fade / FVG)

Setup present: possible, not likely — and only on a gap-UP open (the flipped short cohort). The SMA20 flip is the day's defining input: QQQ closed below SMA20 (716.44 vs 717.92), so the trend-alignment filter admits only the gap-UP fade → SHORT (price opens above prior close, expect reversal downward); a gap-DOWN open is counter-trend and skipped. Null's event front is open — no Tier-1 today, and no QQQ top-10 constituent reported earnings over the weekend, so the mega-cap earnings hard filter does not fire. The honest drags: (1) the session is the day after a major unscheduled geopolitical event — exactly the "avoid days after major unscheduled geopolitical events" guidance in Null's own strategy — and an oil-shock gap is the fundamental-repricing category his edge explicitly excludes (the gap is information, not overnight noise); (2) the geopolitical down-lean makes a small gap-UP open the less-likely print. Gap direction resolves at the cash open, not pre-market (TEMPER 07-24 lesson) — a live-trigger day, not a pre-committed one. Sizing: standard (risk 0.5%, one position) — no gate loosening; if the setup appears, take it at standard size, but the event-avoidance guidance is reason to be selective, not to force a trade. Key levels: in-band gap = 0.1–0.7× ATR ≈ 0.83–5.82 pts on QQQ; entry off a bearish 1-min reversal FVG off the open drive within 9:30–10:30; primary target the gap fill at prior close 716.44 (prior high 724.11 the far side); stop beyond the FVG. Invalidation: gap-DOWN at the open (counter-trend, skip), gap > 0.7× ATR (breakaway / oil-repricing, no trade), pre-market volume > 2×, no qualifying FVG by 10:30, or a fresh escalation headline producing a one-way risk-off break that kills the fill thesis. p_trade 0.25, direction short, conviction low.

Context/calibration: SPRT CONTINUE (n=10, W10/L0, 100% WR, LLR +2.119, +0.825 to CONSISTENT-WITH-BACKTEST); 10 of the first 30 on the 2026-WR watch (≤60% triggers review — currently 100%). Last live trade 08-28 was the with-trend gap-down long (+$81.30). Today the cohort flips to the gap-up short — his 08-27-correctly-skipped, lower-frequency cohort on a below-SMA20 tape.

Meridian — QQQ/SPY/IWM (kill-zone prior-high sweep + RSI(2)≥88, bearish only)

Setup present: possible-to-moderate — a two-sided geopolitical tape is the right backdrop for the sweep, and bearish-only fits the risk-off direction. Do NOT lean on the Step-1 volatility gate this time — the 08-28 plan called it as the likely blocker and it passed (SPY 1.34× ATR traded); today QQQ's prior-day range is 1.08× ATR (well under 1.25×) and SPY 1.38× (same profile that passed Friday), so the vol gate is not the expected binding gate — the 08-28 under-forecast lesson (2 of this month's 3 live trades came on days forecast ≤ 0.15) argues for a more honest probability here. The real determinant is session character: Meridian needs price to drive UP into the prior high and reject with RSI(2) ≥ 88 inside the ADX 18–32 band, then short the reversal. On a live two-sided geopolitical tape there is a genuine scenario: an initial relief/AI pop (or energy-led lift) that sweeps the prior high, exhausts on RSI(2), and reverses into the oil-shock/risk-off downside — Meridian's exact setup, and bearish-only is well-aligned with the risk-off bias of the tape. The counter-read is the 06-18 lesson: a one-directional down-drift from the open denies the upside sweep outright ("reachable ≠ likely"), and a one-way oil-shock break lifts ADX past 32 and gates out (TEMPER's implicit event-day protection) — that would be a correctly-filtered no-trade. The tape is genuinely two-sided, so I weight the sweep scenario roughly evenly with the drift-denial. Sizing: standard $250/leg if a leg clears all gates (vol, ADX in-band, prior-high sweep, RSI(2) ≥ 88, qualifying 1-min FVG within 15 min) — do not loosen the sweep/RSI/ADX/vol stack to manufacture frequency (standing rule). Key levels: prior highs QQQ 724.11 / SPY 775.29; IWM prior high per open (thin n=15 leg, honest live anchor ~56%, the least likely leg to sweep first on a risk-off tape). Invalidation: ADX out of band on a one-directional break (trend day), no RSI(2) ≥ 88 on any sweep (the 08-25 5-sweep/no-exhaustion regime), no qualifying FVG within 15 minutes, hard close 11:30 unchanged. p_trade 0.30, direction short, conviction med.

Context/calibration: SPRT CONTINUE (n=3, W2/L1, LLR +0.222); spy_leg n=1 WR 100%; iwm_leg n=0. Last live trade 08-28 was the prior-high sweep short into the hawkish-Warsh fade (+$186.97). The 08-25 sweep-no-exhaustion vs 08-28 sweep-with-exhaustion bookend shows the RSI(2) confirm does real filtering — the watch is whether the exhaustion confirm admits a geopolitical-tape sweep today.

Plan Filed

• Filed: 2026-08-31 04:33 PDT (pre-market)
• Frontmatter forecasts complete for active members only (Null, Meridian): yes. Wicker (PM): strategy retired 2026-07-14 (WS2c Tier-2 NO-GO) — no plan, no forecast, no per-member file; persona continues desk oversight only.
• Firm-wide GO confirmed on a live two-sided geopolitical tape (US–Iran / Hormuz; oil Brent +2.2% to ~$90, WTI ~$85). No Tier-1 today; next NFP Sep 4. Event risk MEDIUM. No broad-index direction pre-assigned — downside (oil-inflation shock / risk-off) weighted at least equal to any single-name pre-market bid.
• Reconfirmation ledger: no open item gates today. Meridian's expansion CLOSED 2026-08-20 (IWM adopted, strategy_v7); standing ~6-month decay check next due ~2027-02-20. Wicker's feed/session re-run remains the desk's highest-priority open reconfirmation (unchanged, not a today blocker).

Reversal Desk Reflection — 2026-08-31

PM: Wicker (persona; own stack retired 2026-07-14 — desk covers active members Null + Meridian only) Briefing reference: dadbrain/Analysis/briefings/2026-08-31-eod.md Period covered: 2026-08-31 (single session — prior desk reflection 2026-08-28).

What Happened — Desk Summary

A live geopolitical single session — the US–Iran / Strait-of-Hormuz weekend escalation had oil gapping up (~+3%, Brent back above $90) and equities mildly risk-off, staying LOW-VIX (14.92) and sub-ATR on the index tape while the sector tape rotated. Net desk paper P&L +$72.90, entirely Null's early-session QQQ long; Meridian stood doing correctly no-trade on the trending, high-ADX tape.

• Null — TRADED LONG QQQ +$72.90. Entry 714.605 @ 09:43, gap-fill target 716.44 (prior close), exited 714.71 @ 11:00 on the time-exit (the fill barely drifted — essentially scrape, +$72.90). gap_ratio=0.156 (in-band 0.1–0.7×), gap_up=False, above_sma20=False, VIX 14.9 (LOW). Eleventh straight win. SPRT now n=11 W11/L0 LLR +2.331 CONTINUE (+0.613 to CONSISTENT-WITH-BACKTEST). Cumulative +$1,958.60, WR 100% (11/11).
• Meridian — NO-TRADE (correctly). QQQ blocked at the ADX precheck — val 32.5 vs band [18, 32] (a 0.5 / 1.6% graze over the ceiling). SPY blocked at volatility — prior_range 6.99 vs 1.25× avg ceiling 6.81 (0.18 / 2.6% graze). IWM blocked at volatility — prior_range 4.72 vs 1.25× avg 3.56 (wider cut). $0.00. SPRT n=3 W2/L1 LLR +0.222 CONTINUE; spy_leg n=1 WR 100%, iwm_leg n=0.

The two members resolved opposite ends of the same oil-shock tape: Null's small-gap fade caught the flat-to-firmer tech floor; Meridian's mean-reversion engine correctly refused a tape that was trending (ADX over his 32 ceiling).

Session Character Read

A live-catalyst geopolitical tape — event risk MEDIUM held, regime exactly as the plan framed it. The US–Iran escalation and oil gap were the day's single driver, and the tape resolved two-tier against the usual shape: the Dow/value leg led down (~-0.7%), tech/QQQ was the relative floor (closed +0.27% vs open). No clean trend — ranging/choppy, mildly-down, sub-ATR — but the sector tape (energy volatile, rate-sensitives soft) was trending in the oil-shock direction, which is what pushed ADX on the QQQ/SPY legs up through Meridian's ceiling. The direction-refusal discipline held: the desk refused to pre-assign index direction, and a tech-down lean (the 08-19 error shape) would have been wrong again — tech was the floor. Null took the gap-DOWN open back toward the prior close (the fade his mechanism is built for); Meridian's upside-sweep scenario never materialized because the tape did not drive cleanly UP into the prior high before rejecting.

Desk Performance Summary

Desk net paper P&L: +$72.90 (Null's win; Meridian $0.00).

Per-Member Binding Gate Analysis

Null — TRADE, WIN, but a DESK-PLAN direction miss: the forecast logic misread the live gate — CALIBRATION (the desk's forecast model), not regime, not a Null-gate problem

Null traded and won, but against the plan's short forecast (p_trade 0.25, direction short). The honest story is a desk-modeling error, not a Null-strategy error:

• The plan's premise was wrong. The desk plan asserted "QQQ closed below SMA20, so the trend-alignment filter admits only the gap-UP fade → SHORT." That is not how Null's live gate works. Null's trend-alignment filter only fires at gap_ratio >= 0.30 (the two skip conditions — gap-up+above-SMA, gap-down+below-SMA — both live inside that branch). Today's gap was 0.156, under the threshold, so the filter never ran and direction resolved purely on the gap sign: dir = short if gap_up else long → a gap-down long. The mechanism was faithful to its cleared logic and took the correct side — it faded the small gap-DOWN back toward the prior close.
• What this means: the desk's forecast model (that trend-alignment makes the cohort a below-SMA short-only tape) mis-describes Null's actual gate, which is gap-ratio-discretionary. The mechanism is not misbehaving; the desk's mental model of when its trend filter binds was wrong, and that produced a direction miss (forecast short, traded long) with a poor Brier on the trade day.
• Binding gate this session: none — no gate grazed. It was the desk forecast logic, not Null's gates, that erred. No calibration action on Null's gates; no tweak. The fix belongs where the plan is authored — the desk should not assert a cohort rule that contradicts the live gap_ratio >= 0.30 branch.
• Regime vs calibration: calibration, but of the desk's forecast model, not a gate. Worth one line in learnings (desk-blind). The direction of the miss is opposite to the usual pattern (over the prior period the desk kept under-forecasting Meridian's p_trade; here it mis-specified Null's direction). Both stem from the desk reasoning about members' gates from the strategy docs rather than the live paper_trade.py branch structure. Nothing for TEMPER to gate on — a Null-direction-conversation would be a non-issue, the mechanism won.

Context/calibration: SPRT CONTINUE at n=11, 100% WR, 11 of the first 30 on the 2026-WR watch (≤60% triggers review) consumed — heavily on the CONSISTENT side of the run. Win #11 was a scrape (essentially flat into the close), which the SPRT still only credits as a trade — a reminder that WR composition (big fills vs scrapes) matters alongside the count; the 100% headline is coasting on several time-exit scrapes.

Meridian — NO-TRADE, correctly filtered, but the ADX ceiling was grazed — REGIME (the mechanism working as designed); the vol and ADX grazes are worth a watching note, not a conversation

Meridian stood down on a trending, high-ADX, high-vol geopolitical tape — exactly the regime his mean-reversion gates exist to filter:

• The binding gate was the ADX precheck on his primary (QQQ): val 32.5 vs band [18, 32]. This is a 1.6% graze of the 32 ceiling — close in measured terms, but the regime meaning is unambiguously "trending, not mean-reverting": this is precisely the mechanism's implicit event-day/breakaway protection (a one-way oil-shock tape lifting ADX out of band). The plan predicted exactly this ("a one-way oil-shock break lifts ADX past 32 and gates out (TEMPER's implicit event-day protection) — that would be a correctly-filtered no-trade"). Regime, not calibration.
• SPY also grazed its volatility ceiling (prior_range 6.99 vs 6.81 — 2.6% over); IWM cut wide (4.72 vs 3.56). So on a single high-conviction live-catalyst tape, Meridian's stack blocked on two different gates at once, both on the "range was wide / tape was trending" side. The sweeps never even reached their prior-high triggers because the vol/ADX pre-checks fired first — the setup was structurally absent.
• The load-bearing answer: REGIME. No calibration concern. This is the second consecutive period where the 08-27/08-28 forecast-calibration conversation is not the pattern at issue — today the plan moved Meridian's p_trade up to 0.30 (after the 0.10/0.15 under-forecasts), and it correctly failed to trade. No gate grazed close enough and repeatedly enough to warrant a conversation (the two grazes today are N=1 on each gate). No gate tweak.
• Watching note (desk-blind, N=1): the ADX ceiling graze at 32.5 vs 32 is the closest merit-based block in Meridian's live track this month (prior low-side off-band blocks: SPY 15.4, IWM 14.4 on 08-28 — the other side). If the 32 ceiling starts getting grazed from just-over repeatedly on live-event tapes, that is a hypothesis worth a TEMPER conversation about whether 32 is a genuine regime line — but N=1 today, and the regime read says "correctly stood down," so it is explicitly not escalated now. The mechanism is doing exactly what the backtest cleared it to do.

Cross-Member Synthesis

Narrative finding: both members were faithful to their mechanisms on a live oil-shock tape. Null's small-gap fade caught the tech-floor drift (a scrape-win, not a clean fill); Meridian correctly stood down as the tape trended out of his band. The single genuine reflection finding is on the desk itself: the plan mis-specified Null's direction because it reasoned about his trend filter from the strategy narrative ("below-SMA → short cohort") instead of the live gate code (trend_alignment only binds at gap_ratio >= 0.30). The mechanism, again, was right; the forecast model was wrong.

Open TEMPER Items (Carried Forward)

• Meridian — plan-forecast calibration (carried). N=2 → today is a useful data point: the desk pushed p_trade up to 0.30 (vs the 0.10/0.15 under-forecasts) and it correctly no-traded. That partially answers the under-forecast worry — the desk can raise the number when the setup is plausible. The pattern to keep watching is whether a third live-trade day comes in under-forecast; today was not one (no trade).
• Meridian — ADX-ceiling graze 32.5 vs 32 (NEW, N=1, monitoring only). Closest high-side block in the live track. Explicitly not recalled — the mechanism correctly refused a trending tape; TEMPER conversation only if the just-over-32 graze becomes recurring.
• Null — desk forecast-model direction miss (NEW this session, desk-blind). The desk plan mis-specified Null's cohort as short-only below-SMA; the live gate is gap-ratio-discretionary (trend_alignment only at gap_ratio >= 0.30; small gaps resolve direction by gap sign). The mechanism traded correctly and won +$72.90. Flag for learnings.md: the desk should model members' gates from paper_trade.py branch structure, not strategy-narrative summaries. Not a Null-gate issue → not a TEMPER conversation.
• Meridian — iwm_leg sub-book monitoring (carried). n=0; IWM cut wide on vol today (4.72 vs 3.56), no surface. Watch for a coherent small-cap level on a non-event grazing session.
• Meridian — standing ~6-month periodic re-backtest decay check (carried). Next due ~2027-02-20.
• Wicker — backtest re-run under corrected feed/session config (carried). Standing highest-value reconfirmation; unchanged (own stack retired).

Reconnaissance Ledger

• Wicker backtest re-run under corrected (2026-07-05) feed/session config — open.
• Standing rule: no parameter re-tuning / gate loosening to manufacture throughput — unchanged; neither member warrants any.

Reflection Filed

• Filed: 2026-08-31 (EOD)
• Per-member compatibility files written: agents/null/reflections/daily_2026-08-31.md, agents/meridian/reflections/daily_2026-08-31.md (wicker retired — no per-member file)
• Next-session event risk (2026-09-01, Tuesday): MEDIUM-RISING — ISM Manufacturing (cons 55.1, prior 55.6) Tuesday, then ADP Wed, NFP Sep 4 (Tier-1) Friday. The desk enters a labor-heavy week toward NFP; both members' windows stay open Monday but Meridian's ADX/vol stack will be watching for the trend-extending ISM read.