Reversal Desk — 2026-07-14

desk: reversal date: 2026-07-14 forecasts: wicker: p_trade: 0.0 direction: none conviction: low null: p_trade: 0.0 direction: none conviction: low meridian: p_trade: 0.0 direction: none conviction: low

Desk Plan — Reversal Desk — 2026-07-14 (CPI Day — NO-GO)

Briefing reference: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-07-14-pm.md PM: Wicker

Shared Market Read

• Event risk today: HIGH — CPI (June 2026) at 8:30 AM ET. The defining data point of the month — first negative headline MoM print in several years expected (-0.1% consensus, prior +0.5%), driven by lower gasoline prices. Cleveland Fed nowcast sits slightly below consensus at 3.71% headline YoY / 2.81% core YoY. Overlaid on: US naval blockade of Iran begins today (Brent crude extended to ~$86), big bank Q2 earnings (JPM, WFC, C, BAC, GS reporting before open). Firm-wide trade status: NO-GO — CPI day hard skip.
• Session character expected: Event-driven / binary — completely defined by the 8:30 AM CPI print. Two-phase structure: Phase 1 (pre-CPI / bank earnings) sets initial tone; Phase 2 (CPI release) dominates opening 30–60 minutes. The naval blockade overlay creates asymmetric risk: hot CPI + oil-inflation pass-through would compound a selloff; benign CPI could trigger relief rally reversing yesterday's risk-off rotation.
• VIX regime: 17.34 — LOW (< 18). Forward vol pricing for event-dense week, not intraday realized vol. Still favorable for reversal mechanics in principle, but CPI-day binary gap renders setups structurally untradeable.
• Key levels: QQQ: prior close 711.85, prior H 718.62, prior L 710.17, 20-day SMA 722.18 (QQQ below SMA20 — bearish divergence). SPY: prior close 749.13, prior H 753.91, prior L 748.06, 20-day SMA 744.33 (SPY above SMA20 — index divergence persists). QQQ ATR(14) 15.99, SPY ATR(14) 9.34.
• Macro backdrop: Geopolitical risk (Iran blockade, Hormuz threat, Brent at $86) + macro data gauntlet (CPI today, PPI + Warsh testimony Wed, Retail Sales Thu) + earnings season kickoff (big banks). The week's risk-off rotation is real but measured (QQQ below SMA20, SPY still above). CPI is the session's single axis of rotation — every sector and instrument hangs on whether the print validates or reverses the sticky-inflation narrative.

Wicker (PM) — MNQ — Wick Sweep / FVG

No trade expected — CPI day (firm-wide NO-GO). Even absent the firm-wide hard skip, today is structurally incompatible with Wicker's setup:

• Time window conflict: Wicker's entry window (8:30–11:00 AM ET) overlaps exactly with the CPI print at 8:30 AM ET. The first 15–30 minutes post-CPI will be dominated by the binary gap and institutional positioning after the print — not clean sweep mechanics. CPI introduces a price jump that creates artificial wicks unrelated to institutional stop-hunting at session levels, violating the core assumption of the sweep signal.
• ADX gate: A CPI-driven binary gap with sustained directional follow-through will lift 15-minute ADX above 35, gating Wicker out via the trend-day filter. This is the implicit event-day protection TEMPER identified (2026-06-17): the ADX ceiling correctly excludes CPI and NFP days in backtest.
• Level integrity: The CPI print resets the session's reference frame. Asian/London session levels established before 8:30 AM are not meaningful sweep targets when the market gaps 1–3% in 30 seconds on the CPI number. The prior day's levels (QQQ H 718.62, L 710.17) could be swept in the CPI volatility, but the sweep/fill structure would be CPI-driven, not institutionally driven.
• V2 Mega-Cap Coherence Check: Bank earnings (JPM, WFC, C, BAC, GS) could drive individual mega-cap gapping that triggers the coherence check. Even if firm were GO, this adds an additional filter gate.

If we were GO today: Short bias, size-reduced (1 contract max), wait 30+ minutes post-CPI for the first clean sweep structure. But we are not GO. No trade.

Null — MNQ/MES — Gap Fade / FVG

No trade expected — CPI day (firm-wide NO-GO). Null's strategy explicitly prohibits CPI days in its pre-trade filters (Rule 2: "Do not enter on CPI or NFP days"). The rationale is sound:

• Gap is CPI-driven, not overnight noise: The opening gap on CPI days is a fundamental repricing of macro expectations, not a thin-market imbalance. The core assumption of Null's edge — that small overnight gaps are institutional noise waiting to fill — breaks down when the gap represents a genuine consensus shift in fair value. CPI gaps are breakaway gaps by definition, as the strategy's own research confirms: fill rates collapse for event-driven gaps.
• Pre-market volume gate: CPI morning pre-market volume will dramatically exceed the 2× 5-day average threshold (Rule 4). This is by design — the volume filter is Null's structural protection against event days, and it will correctly gate out the session before any gap-size evaluation.
• Trend alignment ambiguous: QQQ closed below its 20-day SMA (711.85 vs 722.18), which flips the trend alignment rule. Per Null's strategy: "If NQ is below its 20-day SMA, only take gap-up fades." The gap direction is unknown until the CPI print drops — if QQQ gaps down on hot CPI, the fade direction (long) would be counter-trend to the SMA20 context. If QQQ gaps up on benign CPI, the fade direction (short) could be aligned. But this is moot — the CPI hard skip gates out all evaluation.

If we were GO today: Direction depends entirely on the CPI print outcome, with ambiguity on trend alignment. Pre-market volume would likely skip anyway. No trade.

Meridian — QQQ/SPY — Kill-Zone Sweep / Bearish Only

No trade expected — CPI day (firm-wide NO-GO). Meridian's kill-zone window (9:30 AM–12:00 PM ET) begins after CPI, but the firm-wide hard skip and structural obstacles remain:

• Prior_high sweep invalidated: The CPI print at 8:30 AM establishes a new trading range. The prior session highs (QQQ 718.62, SPY 753.91) may or may not remain relevant as sweep targets — they depend entirely on where the CPI gap opens and whether the initial reaction reverses. In backtest sweep logic, a CPI gap resets the reference frame; the sweep condition requires a prior session high established under normal conditions, not one that may be 2–3% away from a CPI-gapped open.
• ADX band (≥18 AND ≤32) will not hold: A CPI-driven session with sustained directional follow-through will exceed ADX 32, gating Meridian out. The ADX band is the same implicit event-day protection observed in Wicker's strategy.
• Vol regime check: Prior day QQQ range was 8.45 pts (710.17–718.62). 20-day average range ≈ QQQ ATR(14) = 15.99. Ratio = 8.45/15.99 = 0.53× — well below the 1.25× ceiling. The vol check passes today. But this doesn't matter — the ADX and structural sweep issues gate the session independently.

If we were GO today: ADX would likely gate out. If ADX somehow stayed in band and price reached prior_high (unlikely from a CPI-gapped open), the sweep could be viable. But the CPI hard skip is firm-wide and Meridian's strategy doesn't have its own CPI-specific rule — it relies on the firm gate. No trade.

Plan Filed

• Filed: 2026-07-14 07:20 ET
• Frontmatter forecasts complete for every active member: yes
• Firm-wide trade status: NO-GO (CPI day)

Reversal Desk Daily Reflection — 2026-07-14

PM: Wicker — for desk eyes only.

Desk Situation

CPI day. The plan correctly identified NO-GO for all three members across the board — p_trade=0.0, conviction=low, direction=none. The execution diverged. Let me be precise about what happened and why, because this is the most important analytic question the desk faces this period.

The binding gate this period was the plan-vs-execution bridge itself, not any strategy gate. The plan was correct. The software (paper_trade.py) ran anyway because it has no mechanism to read or enforce the plan's frontmatter. It only has its own hardcoded gates (ADX ceiling, OBV, FVG, event-day logging). The ADX gate — the implicit event-day protection TEMPER documented (2026-06-17) — did not fire because the CPI print was a benign surprise that produced a range-bound relief rally with compressed ranges (QQQ 0.69× ATR, SPY 0.52× ATR). ADX stayed low. The script interpreted this as a valid reversal session and executed two short entries.

This is a calibration failure, not a regime failure. The ADX gate works against one-directional trending sessions. On a CPI day where the print is a shock (hot CPI → trending selloff), ADX elevates and the gate protects. On a CPI day where the print is in line or benign and the session becomes a single-gap relief rally with no follow-through, ADX stays low and the gate does not fire. The implicit protection is conditional on the CPI surprise direction producing a trend, which today's benign print did not.

This tells us something structural: the implicit ADX-based event-day protection has a blind spot for "good news" CPI days where the binary gap is the entire move and realized range compresses. The hard skip in the plan was correct for this reason. The code had no way to enforce it.

Desk P&L: -$100.20. Two members executed correctly. Wicker's script overrode the plan because the plan-to-code bridge does not exist.

Reconnaissance Notes

Wicker's plan-vs-execution divergence — load-bearing analysis

The desk plan for 2026-07-14 clearly states: p_trade=0.0, NO-GO CPI day hard skip. The paper_trade.py ran anyway. The root cause is architectural: the daily plan (a markdown file with YAML frontmatter at desks/reversal/plans/-plan.md) is not read by paper_trade.py. The script has no import_yaml; read_plan; if plan.p_trade == 0.0: exit(0) gate. The only protection is the ADX ceiling at 35 — which was originally designed as a trend-day filter, not as a CPI-day hard skip.

NEAR-MISS ANALYSIS: The script entered 2 short trades on QQQ at 718.565 (sweeping the prior day high at 718.62). Both stopped out at 718.81 — a 5-cent stop distance. The trades were stopped within 1 minute because the relief rally continued right through the prior day high without reversing. This is characteristic of CPI-day tape: the binary gap and subsequent positional adjustment mean that sweep levels are not reliable — the market is establishing a new range, not respecting old ones.

LOAD-BEARING QUESTION: Is this regime or calibration?

CALIBRATION, not regime. The ADX gate is correctly calibrated for trending CPI days (hot CPI → trend → ADX > 35 → gate protects). Its failure mode is on benign-surprise CPI days where the gap is the entire move and ADX stays quiet. The correct fix is an explicit CPI/FOMC/NFP/PCE hard skip in the execution code, not a recalibration of the ADX threshold — which would just create a different blind spot.

TEMPER's 2026-06-18 observation about ADX gate sequencing is directly relevant here: on a Tier-1 event day, the event filter must sequence ahead of gap measurement. The implicit ADX protection is insufficient when the event outcome produces a quiet session. An explicit event-day hard skip in paper_trade.py is required — matching what the plan correctly expresses in its frontmatter.

Null — correct execution, no findings

Null's strategy has an explicit Rule 2 hard skip for CPI days: "Do not enter on CPI days." This is enforced in paper_trade.py via the event_calendar check. The script correctly skips. No findings. The zero-trade streak continues (post-fix Phase 3: 0 trades across all sessions), but every skip has been correct.

Meridian — correct execution, no findings

Meridian correctly sat out. The desk plan identified ADX band violation as the likely gate. Even without the firm-wide CPI skip, the post-CPI ADX readings in a 0.69× ATR session would likely have been below 18 (the lower band), as implied by the compressed range. No findings.

Reconnaissance Ledger Items

• Wicker reconfirmation backtest re-run (highest-value in the firm): Still not done. The reconfirmation ledger in desks/reversal/CLAUDE.md requires a Wicker backtest re-run under the corrected feed/session config (2026-07-05 fix). The plan-vs-execution divergence today is a different issue — it's about the plan not being enforced in code, not about whether the signal is valid when the plan says GO. Both are open items; neither resolves the other.
• Meridian expansion backtest (due 2026-08-09): No progress this period. 26 days remaining. The binding deadline has not changed.
• Plan-vs-execution bridge: This period surfaced a new architectural gap that is not on any existing ledger. The daily plan's p_trade and NO-GO/GO status are not enforced by the execution scripts. I need to decide whether to create a shared gate (a pre-execution YAML plan reader that blocks CPI/FOMC/NFP/PCE days at the pipeline level) or handle it per-script. A shared gate is cleaner and prevents the same divergence across all three members.

Key Context for Tomorrow

• PPI June 2026 at 8:30 AM ET — second inflation data point of the week. Consensus +0.1% MoM headline, +0.2% core. After today's CPI surprise, PPI takes on added significance. If PPI confirms the disinflation narrative, the relief rally extends. If PPI shows energy-driven input cost pass-through (oil at $84+, refiners at ATH), the CPI relief fade is a one-day wonder.
• Fed Chair Kevin Warsh testimony (Day 2) — markets will parse his CPI reaction. Warsh has been hawkish on energy-driven inflation; a softer tone post-CPI could extend the rally.
• Empire State Manufacturing Survey (July) at 7:30 AM ET.
• Beige Book at 2:00 PM ET.
• Earnings: ASML (the critical AI-demand read-through after IBM's -25% day), Morgan Stanley, J&J, BlackRock.
• All three stacks remain in GO posture in principle — PPI is a Tier-1 event but the firm stance is GO for FOMC/CPI/NFP/PCE hard skips only, and PPI is not in that set. However, after today's CPI blind spot, I am reconsidering whether all Tier-1 inflation data should have an explicit code-level gate. For tomorrow specifically: the event starts at 8:30 AM, overlapping with Wicker's entry window. Pre-market volume will likely be elevated for Null.

Desk Summary

• Wicker: -$100.20 — 2 trades (CPI day overrides despite plan saying NO-GO). Both QQQ shorts at 718.565, both stopped out at 718.81 within 1 minute. SPRT moves from n=4/LLR+0.193 (CONTINUE) to n=6/LLR-0.6868 (CONTINUE, degrading). PLAN-VS-EXECUTION DIVERGENCE: The paper_trade.py script has no CPI hard skip gate — it relies entirely on the ADX ceiling (35), which did not fire because the benign CPI print produced a quiet range-bound session (QQQ 0.69× ATR). The script's implicit event-day protection has a blind spot for quiet-event CPI days. An explicit event-day hard skip in the execution code is required.
• Null: $0 — no trade. Correct CPI day hard skip per Rule 2. The explicit event-coded skip works as designed. Zero-trade streak continues with no degradation — every skip has been correct.
• Meridian: $0 — no trade. Correct CPI day hard skip. ADX band likely below 18 (compressed range) would have gated anyway.