# Tuesday, August 4, 2026

## TODAY — Top 3

1. **Wire the AMSB indicator into `indicator_alerts` with `alert()` JSON**
   The CRWD trade had to be captured by hand overnight. The 16 `alertcondition()` lines can only send a fixed sentence — no price, no stop, no target, no bar time — so no AMSB firing can ever reach the database. One `alert()` block turns every firing into a row. That's exactly the filtered sample the +0.48R number is missing.

2. **Decide: demote `revbreak_long` out of the validated set**
   Still blocked on you, 7 days now. n=259, 18% win, −0.46R avg, lower bound −0.54. The most-firing play is the biggest bleeder, and it keeps firing while the decision sits.

3. **Guard the SIP composite gap calc — SNAP printed `gap:-100.0%`**
   A −100% gap is a missing prior close, not a stock. It shipped into today's new `/sip/csv/sip_2026-08-04.csv` archive, so a bad number is now permanent in a file traders read.

---

## What shipped

- **Adaptive Mean + σ Bands v5.2 tool playbook** — [playbooks/adaptive-mean-sigma-bands.html](/playbooks/adaptive-mean-sigma-bands.html). Eight slides: one engine / three zooms (daily · 30-min · 2-min), 9 tracked trade types, and the CRWD case — 8/4 close 211.22, +6.63R off 191.21, projected 209.15 tagged and **paid**, now sitting at +1.93σ. Tool source published alongside at `playbooks/_tools/adaptive_mean_swing_v6.pine`; index got a tool card.
- **Stat-honesty correction inside that playbook** — the deck's "VWAP L: +.48 ExpR over 168 signals, 26% win" is wrong twice: 26% is not the win rate, and those 168 signals are not this exact setup. Paste-ready slide fixes are in the page, and the whole thing is banner-labeled **HYPOTHESIS**, not a validated dollar edge.
- **Stockbee 20% breadth scanner pair** — [system/stockbee20/](/system/stockbee20/): `stockbee20_up.pine` + `stockbee20_down.pine` + a 5-minute `HOW_TO.md`. Two TV alerts post to the existing SMB webhook, land in `alerts_tradingview.indicator_alerts`, get counted and split by tape type, and surface on the multi-tf-stack cheat card, Cross-Signal Tensions, and all three MGP editions.
- **Daily SIP as CSV** — `/sip/sip_today.csv` (always latest) plus `/sip/csv/sip_YYYY-MM-DD.csv` (permanent archive), written on every build: 7:00, 8:45, and the 15-minute watchdog. Columns carry rank, score, reason, which scanners found it, earnings tier/badge, and catalyst news.

## What moved forward

- **Every pipeline ran clean.** All three Domino briefs emailed (MGP 05:41, mid-day 11:30, EOD recap 16:30) and all three MGP v2 editions (06:48, 11:38, 16:40). No silent regression tonight.
- Scheduled surfaces all refreshed on time: morning + EOD briefings, multi-tf-stack (+ preview verdict), swing radar 15:40, setup league 16:30, scorecard + you-vs-you 16:48, Morning Command 16:50, Atlas flow status 20:40.
- **HUD work happened in the dev tree** — `api/trader.php`, `hud/js/render.js` and `hud/index.html` were all edited between 18:17 and 18:26 ET. Recorded from file timestamps only; the nightly run does not open cockpit code, so what changed isn't established here.
- The AMSB alert-plumbing spec is fully written — the exact `alert()` payload shape, the four required receiver fields, and the `trade_id` pairing rule are in the playbook. Tomorrow's task is typing, not designing.

## What's blocked

- **`revbreak_long` demotion** — waiting on you since 7/28. It is the only P0 blocker on the board.
- **BOAT v4.6** — commit to the cap=1 selectivity build and attack costs, or shelve it. Open since 8/2.
- **scientific_layer migrations 001–009** — still needs a psql tunnel run; unlocks the daily-summary view, sizing-calculator DB wiring, and the rolling-stats / setup-performance views.
- **AMSB hold rules** — the tool exits at its own target, so it has nothing to say about "I should have held for 6R." Three candidate hold rules are written up and untested. Don't put the tool's expectancy next to a hold claim until they run in the a2 bench.
- **Duplicate MGP sends, low grade** — every edition emailed twice today (06:48/08:35, 11:38/11:43, 16:40/16:44). Same pattern on 8/3. Idempotency isn't suppressing the second fire. Noise, not damage.
- **HUD `api/error_log` unreadable** — the dev-tree error log was being written at 23:08 ET and would not open over WebDAV during this run. Unread, so unknown. Worth a look if the HUD misbehaves tomorrow.

## Files touched

- `playbooks/adaptive-mean-sigma-bands.html` — new 8-slide tool playbook, CRWD case, stat-honesty fixes
- `playbooks/_tools/adaptive_mean_swing_v6.pine` — v6 indicator source published
- `playbooks/index.html` — tool card added (backup `.bak-pre-tool-card-2026-08-04_2243`)
- `system/stockbee20/stockbee20_up.pine`, `stockbee20_down.pine`, `HOW_TO.md` — new breadth scanner pair + setup doc
- `sip_page/j3p_sip_today_page.py` — daily SIP CSV writer (latest + dated archive), py_compile clean on a2 3.9
- `trading/dev/a1p_journal/accountability-v2/` — `api/trader.php`, `hud/js/render.js`, `hud/index.html` edited 18:17–18:26 ET (dev tree; content not inspected)
- `AGENT_SYNC.md` — earnings agent logged the CSV ship and flagged the SNAP gap bug
- `system/J3P_SWING_RADAR.html`, `J3P_SETUP_LEAGUE.html`, `J3P_LIVE.html`, `command.html`, `multi-tf-stack/`, `briefings/`, `mgp/` — scheduled refreshes, all on time
