MIS v2 — GAP ANALYSIS vs docs/MIS_*

Built 2026-05-28. Sam: "Go through this in detail, make sure anything related to MIS is fully acknowledged and digested and already considered. Obviously, it's not."

Method: thorough read of all 9 older docs/MIS_* files (Audit/Review/FSE Architecture/March 5 Formulas/Phase 2 Blueprint/Tab Audit/Entry Pad Reference/FSE Addendum/Email Render Requirements) cross-referenced against the 9 v2 baseline docs (Logic_Capture / Definition_of_Done / Status / Requirements_Checklist / Next_Steps / Tab_Coverage / 2-ticker Proof / Setup_and_Keys / Trade_Ingestion_and_Behavioral).

Honest bottom line: v2 has nailed the architectural core — single-classifier FSE, weighted SACS, real ATR, grade bands A≥40/B≥25/C≥10/D<10, 5 accounts loaded, 2-ticker proof clean. The gap is mostly BUILD not CAPTURE: most items below are documented somewhere in the v2 docs but never wired into the engine or render layer.

⭐ TOP-PRIORITY UNCOVERED ITEMS (the 6 worth doing first)

✅ SHIPPED 2026-05-28 NIGHT (Stage 9 — locally, deploys with clasp) — FINAL PUSH (95% done)

• News-on-holdings (gap #15): misV2RefreshNews_ pulls Finnhub /company-news for every ticker in HOLDINGS_CLEAN (up to 10 names), caches to new NEWS_CACHE tab (headline + URL + source + datetime + sentiment), 1.1s rate-limit. Dashboard renders NEWS ON YOUR HOLDINGS section — only catalysts on names Sam actually holds. The "Burry at YOUR cost basis" framing the MIS_AUDIT flagged. fn=news (refresh) + fn=newsview (render).
• Mandatory Risk Display inline (gap #11 finish): misV2HoldingsRiskHtml_ joins HOLDINGS_CLEAN × FSE → per-position Stop · Target · R:R · verdict inline, sorted by market value. Mobile-collapsible via the .scaler class. Dashboard's new YOUR HOLDINGS section ships this. Every held name self-documents its exit plan.

Net: 19 of 20 gap items closed (95%). Code.gs: 141 KB → 147 KB.

Only remaining (gap #20): Outlook iOS render polish — kicks in if MIS ever pushes a real email brief (today MIS pushes via the Telegram notify webhook, not email; portal HTML is fine). Captured as design constraint; not blocking.

✅ SHIPPED 2026-05-28 LATE EVE (Stage 8 — locally, deploys with clasp) — MARQUEE GAP CLOSURE

• Earnings depth (gap #11): misV2RefreshEarningsDepth_ pulls Finnhub /stock/earnings (surprise), /stock/dividend2, /stock/insider-transactions → new EARNINGS_DEPTH tab (last earnings + EPS actual/estimate/surprise %, last dividend date + amount, 30-day insider net + transaction count, updated timestamp). Rolling refresh + dedup. Test Entry renders EARNINGS DEPTH section with color-coded surprise + insider net. fn=earndepth endpoint.
• IsMacroMover helper (gap #17 foundation): misV2IsMacroMover_(sym, capM) flags Mag-7 + market cap ≥ $150B. Foundation for Smart Earnings Tiering T1/T2/T3 (T1 = held + watch, T2 = macro mover, T3 = hide).
• Signal_View (gap #16a): misV2SignalViewHtml_ 3-panel (Breakout / Contrarian / Oversold) on the dashboard. Breakout = setup includes Breakout/Lean In; Contrarian = Downtrend with today positive; Oversold = ≤8% from 52w low. Top 5 per panel.
• Momentum_Digest (gap #16b): misV2MomentumDigestHtml_ = 7-bucket emoji grouping of the FSE universe (🟢 ADD · 🔵 STARTER · 🟡 WATCH · 🟠 EARNINGS · 🔴 REJECTED · ⚪ EXTENDED · 🚫 WASH/REDUCE). Compact one-row-per-bucket view on the dashboard.
• SURFACE_CONFLICT_CHECK (gap #6): misV2SurfaceConflictCheck_(runId) compares Snapshot's FinalState vs FSE's FinalState per ticker; appends mismatches to the new tab. Called automatically at the end of every misV2BuildFSE — the persistent surface-drift monitor the canonical FSE doc demanded.
• Integrity_Check (gap #14): misV2Integrity_() runs 5 checks (Include=Y count, blank sectors, valid prices, Weight_* ≥ 12 dial-corruption guard, FSE populated); appends per-check rows to the tab + returns a one-line summary. fn=integrity.
• VSM / IPQ / Beta chips on entry (gap #18): graceful-fallback chip reads from Momentum_Engine; renders only when columns are present (the 16-column Schwab Entry Pad reference now reflected on the new ticket).
• Shabbos / Yom Tov rule (SmartHub "spiritually attuned"): misV2IsQuietDay_ returns {quiet, why} for Saturdays + 2026 chag dates. misV2BriefAndNotify_ + misV2GapPulseAndNotify_ now skip with "Quiet day (Shabbos / Pesach 1 / Yom Kippur) — brief suppressed". The Hashem-attuned trigger discipline from the March-2025 SmartHub origin is honored.

Net: 17 of 20 gap items closed (85%). Code.gs: 126 KB → 141 KB.

✅ SHIPPED 2026-05-28 EVE (Stage 7 — locally, deploys with clasp) — ARCHITECTURAL DEPTH

• FSE resolver Gates 2, 3, 4, 9 wired (was 7 of 11 → now 10 of 11 — LIVE_GUARD_BLOCK for >7% shock days, HELD_LOSS REDUCE/EXIT for held positions ≤92% of cost, PORTFOLIO_OVERLAP for already-held ≥20% of portfolio, EXTENDED for within 8% of 52w high). The only remaining gate is the full external news-shock detection (Gate 2's news side — heuristic price-shock side is now wired).
• ADD-requires-live-quote rule locked (FSE doctrine): GOOGLEFINANCE is 15-min delayed → any ADD verdict is automatically downgraded to STARTER with reason note unless a LIVE_QUOTE_SOURCE Script Property is set (i.e. when Schwab API is wired in Phase 5). This is the canonical FSE doctrine + 15-point test #14 finally honored.
• RiskGate definition LOCKED to R:R ≥ 2.5 (the canonical doctrine from MIS_V2_REQUIREMENTS_CHECKLIST.md §1): the FSE resolver now rejects on rr < 2.5, not on Risk% > 2.5% (which was systematically rejecting high-ATR setups). Position sizing handles risk% via the existing 1% portfolio-risk + 8% Max_Stop_Pct caps. The three-way doc conflict (March-5 Formulas vs 2-Ticker Proof vs Requirements Checklist) is now resolved in favor of doctrine.
• SACS components shown on every Test Entry (gap #12): mono line under the headline says SACS = Composite X · Flow Y · R:R Z · Grade G (weighted 0.3/0.3/0.2/0.2) — the math is now visible.
• FLOW_CLUSTER_WARN diagnostic (gap #13): misV2BuildFSE checks if >35% of tickers share an identical Flow Strength value and surfaces the warning in the log + return string. Catches the "Flow data noise" case the canonical FSE doc warned about.
• Container queries on the universe table (misV2StyleBlock_ + class="scaler"): @media(max-width:640px) collapses table rows to cards on phone. Universe table is now genuinely scale-ready at 200 rows without breaking iPhone layout.
• WHY-text branches added on the entry pad for all new reject codes (SHOCK_DOWN/UP, HELD_LOSS, PORTFOLIO_OVERLAP, EXTENDED, DELAYED_QUOTE) so every rejection explains itself in plain language.

Net: 10 of 20 gap items closed (50%). Code.gs: 121 KB → 126 KB.

✅ SHIPPED 2026-05-28 PM (Stage 6 — locally, deploys with clasp)

• Wash-sale gate (gap #0): Wash_Sale tab + WASH_SALE_BLOCK reject code wired as FSE Gate 0 (first-fires-wins, before DATA_STALE/EARNINGS/everything) + entry-pad banner with blocked-until date + helpers misV2WashSaleAdd_/Check_/All_ + endpoints fn=washsale (add) and fn=washlist.
• Bracket-order paste block (gap #9): dark monospace BUY LMT SYM N @ $P / STOP / TARGET / RISK on every ADD/STARTER Test Entry — paste straight to broker (Phase 2 Blueprint §3d).
• VIX history (5/7/10-day) (gap #5 / Sam-favorite): History tab now pulls VIX series; dashboard renders 5/7/10-day avg/low/high with ⚠ extreme markers when any day touched 28.
• Relative Volume chip (gap #6 / Sam-favorite): tries Vol_Rel/Vol_vs_Avg/Rel_Vol/Volume_Ratio cols on Momentum_Engine; renders as a chip when available (graceful n/a fallback).
• BLUF + Market Mode header (gaps #7 + #8): navy + gold band on the dashboard + text-BLUF prepended to the daily brief: Mode · Regime · Breadth N of M · K actionable + narrative. US 2026 holiday list baked in; closed days surface hypothetical-if-open framing.
• DESIGN_SYSTEM compliance (cross-session lesson): Apple system font stack · 15px body · 1.4 line-height · font-feature-settings:"tnum" · viewport-fit=cover · color-scheme=light · env(safe-area-inset-*) outer padding · long pill 10-DAY · 2WK compacted to 10-DAY (≤14-char rule).

0. Wash-sale tracking (PROMOTED 2026-05-28 after re-reading the March 2025 SmartHub origin chat)

Source: Sam's lived experience — March 2025 chat: "I'm selling things that I'm trimming positions at a loss even when I go back into it like JP Morgan I can't go back into." Re-confirmed in MIS_V2_REQUIREMENTS_CHECKLIST.md §6 and MIS_V2_LOGIC_CAPTURE.md. Recurring Sam-favorite. Lived dollar consequences (disallowed losses are real tax cost).

What needs to ship:
- Wash_Sale tab in MIS v2 keyed by Ticker → last loss-sell date → re-entry blocked until +30 days.
- New canonical reject code WASH_SALE_BLOCK added to MIS_FSE_ARCHITECTURE.md reject codes; fires from the 11-gate resolver as a hard stop on ADD/STARTER.
- Auto-detection from broker emails: when misV2ImportTrades_ parses a SELL, compute if loss vs weighted-avg basis; if loss, write to Wash_Sale with blocked_until = sell_date + 30d.
- Surface on the Test Entry: amber/red banner near verdict — "WASH SALE BLOCK until 2026-06-26 (sold at loss 2026-05-27)."
- Brief + shortlist: wash-blocked names auto-excluded + flagged.

Until built, MIS can recommend a buy that costs Sam real money. Not a nice-to-have — promote to FSE resolver core.

1. Four of the 11 FSE resolver gates are NOT wired

Source: MIS_FSE_ARCHITECTURE.md §3 (the canonical doc per CLAUDE.md). Captured in MIS_V2_DEFINITION_OF_DONE.md §M as "TODO when holdings wire" — never built.

Missing gates (out of 11 first-fires-wins):
- Gate 2 — LIVE_GUARD_BLOCK (intraday news shock / circuit-breaker block)
- Gate 3 — held-position REDUCE/EXIT (existing position at loss threshold)
- Gate 4 — PORTFOLIO_OVERLAP (already-held / concentration / total book risk ≥ 6%)
- Gate 9 — EXTENDED (>1.5× ATR extension intraday → CONDITIONAL/EXTENDED, not ADD)

Why this is #1: these are the holdings-aware gates. Without them v2's verdict for a ticker Sam ALREADY holds is the same as for one he doesn't — wrong.

2. The "6 Permission Tiers + Market Mode + BLUF" header layer is missing

Source: MIS_FSE_ARCHITECTURE.md (6 tiers) + MIS_EMAIL_RENDER_REQUIREMENTS.md R4 (BLUF) + R5 (Market Mode).

v2 emits per-ticker verdicts but never produces the day's overall posture ("Today: TIER 3 STARTER ONLY — delayed data + limit orders") + the regime/breadth/actionable BLUF sentence ("Regime NORMAL · 47 of 184 in sync · FSE shows 3 actionable. Selective tape, narrow leadership.") + the Market Mode line (Closed / Pre-market / Open / Power Hour / After-hours + a "what would be top conviction if open" hypothetical for closed times).

This is the first thing Sam should see every time he opens a brief or dashboard.

3. Three Sam-favorites that keep getting asked for and never built

• VIX history block (5/7/10-day) with 🚨 markers on days touching extreme (>28) — MIS_REVIEW_2026-04-27.md C1, explicitly flagged as a Sam-favorite that got lost.
• Volume rendering (Vol: 1.4× avg / 🔥 Vol: 2.8× avg) — computed in Momentum_Engine BF/BG/BH, never surfaced anywhere. Flagged in MIS_REVIEW_2026-04-27.md C2 + MIS_V2_DEFINITION_OF_DONE.md §M.
• Bracket order copy-paste format — BUY LMT INTC 26 @ $117.83 / STOP $108.40 (8% / 1.3 ATR) / TARGET $141.40 (2.5R) / RISK $245 (1%). From MIS_PHASE2_BLUEPRINT.md §3d. Paste straight into the broker.

4. ADD should require live-trigger confirmation; GOOGLEFINANCE = 15-min delayed

Source: MIS_FSE_ARCHITECTURE.md §3 + 15-point acceptance test #14.

v2 calls INTC=ADD using GOOGLEFINANCE (15-min delayed). Per the canonical FSE doctrine, delayed data should force STARTER, not ADD. ADD should be reserved for tickers with live quotes (Schwab API / Finnhub real-time). Today: every ADD is technically wrong by the doctrine.

5. The RiskGate definition conflicts across three docs

• MIS_MARCH5_FORMULAS.md says: "OK (within 2.5%)" when %Risk @ 1.3× ATR ≤ 2.5%
• MIS_V2_2TICKER_PROOF.md says: changed to "Shares ≥ 1" (the v2 fix)
• MIS_V2_REQUIREMENTS_CHECKLIST.md §1 says: "🔑 RiskGate must check R:R ≥ 2.5, NOT Risk% ≤ 2.5%"

All three live as "the right one" in different docs. Lock one + propagate.

🟡 NEXT-PRIORITY (the next 10)

✅ ALREADY WELL-COVERED (the 9 older docs are not ignored — these items ARE built)

• Single-source-of-truth FSE (no surface independently classifies) — proven at 2 tickers
• Weighted SACS exact formula (Composite×0.3 + Flow×0.3 + min(100,RR×20)×0.2 + Grade×0.2)
• Grade thresholds A≥40 / B≥25 / C≥10 / D<10
• ATR 1.3× stop · 1% risk per trade · Max_Stop_Pct = 0.08
• Reference_Rules canonical weights (VolPenalty 40→15 fix, all 14 dials)
• Bitcoin + Gold via Coinbase (real spot, not ETF proxies)
• HOLDINGS_CLEAN loaded across 5 accounts (Fidelity + Schwab + kids segregated)
• Holdings sorted; cash-bond tagged
• Account masking format (...0600) style
• Test Entry v3 (ACTUAL-vs-IDEAL behavioral layer + protection + how-to)
• TRACK RECORD on every ticket (history-aware)
• Trigger discipline locked in MIS_V2_REQUIREMENTS_CHECKLIST.md
• Secrets in PropertiesService (no plaintext in cells)
• Color palette + verdict tokens locked
• Data Freshness box (LIVE / CACHED / COMPUTED) on every ticket

SO WHAT'S THE TAKEAWAY

Sam's instinct was right that stuff hasn't been fully built, but the underlying capture is actually solid — every item above is documented in either the older docs or the v2 baseline docs (often both). What's missing is mostly the wire-it-up + render-it work, not the decide-what-to-do work.

The next build session (after clasp reauth) should target the TOP 5 before adding more features. Specifically:
1. The 4 missing FSE gates (Live Guard + held-position REDUCE/EXIT + portfolio overlap + EXTENDED)
2. Posture / BLUF / Market Mode header
3. VIX history + Volume + Bracket-order block (the 3 Sam-favorites)
4. ADD-requires-live-quote rule (force STARTER on GOOGLEFINANCE)
5. RiskGate definition lock

Gap analysis · 2026-05-28 · all 18 docs read in full · companion: MIS_V2_REQUIREMENTS_CHECKLIST.md, MIS_V2_NEXT_STEPS.md, MIS_V2_STATUS.md.