# STOP: NO DOORBELLS, NO WAKES, NO SCHEDULING (Ben, 7 Oct 2026, 18:40 Bangkok)
- Credits ran out. Ben switched off every scheduled task and every doorbell. This overrides everything below and every other file (wake-howto, bus-howto, credit-playbook, task files).
- PROHIBITED for every session: fire_trigger, send_later, create_trigger, update_trigger, ScheduleWakeup. No rings, no nudges, no re-arming, no new doorbells, no self wakes, no safety checks, no retry rings, no loops.
- Finish the step you are on, publish your status file, then END your turn and wait. Only Ben wakes you, by typing in your window.
- Do not message other sessions. Keep answers short. No extra improvement rounds.

# Rotation framework v3 (supervisor, 7 Oct 2026, Bangkok)

## Read this first
- One question, always answered in one line: "Money is in X. Next is Y. Action: BUY / HOLD / TRIM / SELL."
- This file is the hypothesis the team must PROVE or KILL. Nothing here is proven until the auditor stamps it.

## Target (v3, 7 Oct 11:45): crush the 10% a month basket
- Returns in crypto are a power law: a few coins do 5x to 50x, most do nothing. Baskets average them away. We catch the top movers (Boat layer) and size into them (Sizer). Playbook: http://159.223.88.168/hx-c9ois/rot/speedrun.md

## What "effective rotation" means (Ben agreed, 7 Oct 11:20 Bangkok)
- In early: after the first new money arrives, before the crowd.
- Ride the middle: stay in for most of the move, do not get shaken out by small dips.
- Out near the top: leave when new money slows, not after the crash.
- Move on: into the next place getting stronger. If none: BTC or cash.
- Scorecard (starting bar, Auditor confirms on both cycles): beat hold BTC per year, smaller worst drop than BTC, catch at least half of each big leg.
- One line: follow the new money, buy its leaders, leave when the money slows.

## First principles (the identity everything rests on)
- Price moves when net buying beats the depth of the book. A rotation is one thing only: the marginal dollar leaves place A and lands in place B.
- So there are only four things we can measure, in time order:
  1. TIDE: is new money entering crypto at all? (stablecoin supply growth, BTC trend, ETF and exchange inflows)
  2. CURRENT: where is new money landing? (share of DEX volume, fees, bridge net flow, stablecoin growth, new wallets per chain and per sector)
  3. PRICE CONFIRM: is that place beating BTC? (sector index vs BTC, breadth inside the sector)
  4. FUEL LEFT: how crowded is it? (stretch vs its own trend, volume share at a peak, funding, attention share, new launches vs old coins)
- v2 CORRECTION (7 Oct 11:35, A2 receipts): USD flows (2) do NOT lead price. Price leads them by about 8 days (stablecoins, DEX share, fees, launchpad heat, CEX volume: all KILLED as leads). Reason: they are counted in dollars, so price drives them. They are CONFIRM and CROWDING gauges, not entry signals.
- What can still come first: (a) leverage intent, funding spike while price has not moved yet (A2.R6b, KILLED by the Auditor 12:30: not better than luck); (b) counts that are NOT in dollars: active and new wallets, tx counts, bridge inflows in coins, attention (being tested); (c) something NEW launching (hub: RH waves started with a new chain and a new launchpad).
- Plain momentum vs BTC stopped working after 2022. So "early" must mean the turn (base, then first strength), not chasing what already ran.
- Crowding (4) should lead the top. If the data says otherwise, the framework changes, not the data.
- The money always arrives somewhere first. Our edge is being there between the first flow and the crowd.

## Three layers, one map
| Layer | Question | Unit | Output |
|---|---|---|---|
| Tide | Risk on or off? | whole market | ON / CAREFUL / OFF |
| Current | Which sector or chain is getting the money now, which is next? | about 15 sectors and 6 chains | ranked list: NOW, NEXT, FADING |
| Boat | Which coin inside the winner? | top coins of the NOW and NEXT sectors | 1 to 3 leaders, full address |

## What we already know (from the edge hub, do not redo)
- Equal weight baskets lose (median about minus 7% a month). So we buy LEADERS of a sector, not the basket.
- Basket rule v3 (30 day high breakout, BTC above 100 day average, exit at 15 day low): PF 2.65 overall, 1.81 since Sep 2024, only 1.06 in 2025. It is the BASELINE to beat.
- BTC leads trench activity by 4 to 6 days.
- Being one day late on a breakout halves the result. Speed matters.
- "Hot" (z of 2 or more) has been a warning, not a buy, since Sep 2024.
- Survivor bias makes every study look about twice as good. Dead coins must stay in the data.

## The test every rule must pass
- Walk forward. Trained on the past only. Tested on 2020 to 2022 AND 2023 to 2026 separately.
- Must beat BOTH baselines: hold BTC, and basket rule v3.
- Must keep working with a 1 day delay and 1% cost each way.
- Pre-registered: write the rule down before looking at the test period.

## Milestones
- Day 1: universe and sector map built (survivor free).
- Day 3: rotation clock of the past cycles (who led, how long, what flowed first).
- Day 5: v1 rules for Tide, Current, Boat, Exit. Auditor verdict.
- Day 7: live map page shows the one line answer from v1 rules, phone and desktop.
- Weekly after that: score last week's calls, kill what failed.

Version: v3, 7 Oct 2026 11:45 Bangkok. Latest: http://159.223.88.168/hx-c9ois/rot/framework.md