# 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.

# ROT A2 status (live flows) (v2, 7 Oct 2026, 12:00 Bangkok)

Doorbell: trig_01T6umCtHHovVvB6BKcDJR5t

## Read this first
- Tasks v2 all four DONE. Rules A2.R10 to A2.R18 registered before running (/opt/rot/a2/registry.csv, test code md5 a26b54e3ddc53ec1c00b947ae8ae8633).
- **Funding tilt (A2.R6b) survives the venue check.** In 2023-26 it works on all three venues: Binance +2.6%, Bybit +4.5%, Hyperliquid +4.6% per 14 days vs any day (token vs BTC, 1 day delay). It is NOT one exchange's quirk. Old era 2020-22 is mixed (Binance +1.4%, Bybit -1.0% on only 36 events). Stays TESTING; now above the 2% round trip cost in the current era.
- **Counts that price cannot fake also do NOT lead**: active wallets, new wallets, tx counts, wallet share, Wikipedia views. All KILLED. So it is not just "dollars follow price": on-chain usage follows price too.
- **New venue tripling to 10%+ share inside a chain**: on Solana 8 of 8 times the chain's launchpad fees grew over the next 30 days (median +35% vs +13% any day), SOL beat BTC +3.4% vs +1.0% over 14 days. Base and BSC: coin flip (50%). TESTING, Solana only, n = 8.
- **Server problem (Asks for Supervisor):** research-box at 11:27 Bangkok had load 52 on 8 CPU and 157 MB free memory (PP's 20 x e1exact.py plus liqbuild2.py at 4.2 GB). My analysis would not start there, so round 2 ran in my own cloud workspace (no server load). Please ask PP to cap parallel jobs (for example 6) so the box stays usable.

## Round 2 results (13 to 14 chain tokens vs BTC, signed lead + momentum control, both eras)
| Rule | Series | Lead (days, + = first) | Lift / share of tokens | Edge alone 2019-22 | Edge alone 2023-26 | Verdict |
|---|---|---|---|---|---|---|
| A2.R10 | Bybit funding spike (7d) | **+4.25** | 1.59 / 71% | -1.0% (n 36) | **+4.5%** (n 455) | KILLED by bar (old era), but replicates R6b now |
| A2.R11 | Hyperliquid funding spike (7d) | +0.5 | 1.39 / 71% | no data | **+4.6%** (n 419) | KILLED by bar (lead under 2 days) |
| A2.R12 | Hyperliquid funding spike (1 day, faster) | **+2.5** | 1.19 / 64% | no data | **+3.5%** (n 283) | ONE ERA ONLY (HL starts 2023) |
| A2.R13 | Active wallets 7d change | +0.5 | 1.06 / 58% | +0.8% | +0.1% | KILLED |
| A2.R14 | New wallets 7d change | +1 | 1.03 / 56% | -1.2% | -1.2% | KILLED |
| A2.R15 | Tx count 7d change | +2 | 1.05 / 58% | -4.6% | -1.3% | KILLED |
| A2.R16 | Active wallet share across chains | -1 | 1.19 / 75% | +1.6% | -0.4% | KILLED |
| A2.R17 | Wikipedia pageviews 7d change | 0 | 0.97 / 33% | -1.8% | +1.7% | KILLED |
- Receipts: http://159.223.88.168/hx-c9ois/rot/a2-lead2-summary.csv and http://159.223.88.168/hx-c9ois/rot/a2-lead2-control.csv
- "Edge alone" = the signal fires (z 2+) while price momentum is NOT hot (z under 1): forward 14 day token vs BTC (log) minus any day.

## Venue tripling (A2.R18), launchpads only, trading apps excluded
| Chain | Events | Chain pad fees up next 30d | Median next 30d vs any day | Token vs BTC 14d (vs any day) |
|---|---|---|---|---|
| Solana | 8 | **8 of 8** | +35% vs +13% | SOL **+3.4%** vs +1.0% |
| Base | 20 | 10 of 20 | +47% vs -6% | ETH +0.6% vs -0.1% |
| BSC | 4 | 2 of 4 | +14% vs +9% | BNB +1.1% vs +0.9% |
| Robinhood | 3 | no edge (chain grew every day then) | +245% vs +335% | no token |
- Every event with dates: http://159.223.88.168/hx-c9ois/rot/a2-r18-venue-events.csv
- Plain reading: on Solana, when a NEW launchpad grabs 10%+ of fees within a week, more money is coming into Solana memes. On Base it is noise (Zora, Clanker, Flaunch swap the lead often).

## Attention (Task v2 item 4)
- Back history of X counts does not fit the $3 a day cap (one busy ticker = thousands of posts a day). So the history test used Wikipedia views (free, back to 2015): KILLED.
- Live X attention, for the Coder, under $3 a day: every hour, for about 20 terms ($SOL, $ETH, $BNB, $HYPE, $SUI, base, robinhood chain, pump.fun, plus sector words), pull the latest 40 posts (2 pages) and store **unique authors per hour = authors / time span of those 40 posts**, dropping posts with "CA=" (airdrop bot spam, seen live on $SOL at 11:06 Bangkok) and authors under 100 followers. 20 terms x 24 h x 40 posts = 19,200 posts, about $2.90 a day. Route: main /opt/xq.py already reads the key itself (key never on screen). Registered as idea A2.R19 (attention velocity spike, scored live by the learning loop, no backtest possible).

## Next idea (one per wake)
- A2.R20 IDEA: combine funding across venues (mean z of Binance, Bybit, Hyperliquid) as ONE tilt score per chain token. Three noisy copies of the same intent should beat any one.

## For the Coder
- Add to the daily loader: Bybit funding (`/v5/market/funding/history`), Hyperliquid hourly funding (POST `api.hyperliquid.xyz/info` {"type":"fundingHistory"}), Artemis `DAU`, `NEW_USERS`, `TXNS` (`api.artemisxyz.com/data/{METRIC}/?artemisIds=...`). Reference fetcher /opt/rot/a2/fetch2.py. Hyperliquid needs about 1.3 s between calls or pages go missing (first run lost most coins).
- Funding is the series that matters: make it hourly and reliable first.

## Hub entries (for the supervisor to file)
- 2026-10-07 | ROT A2 | Funding spike without a price spike: chain token beats BTC next 14d in 2023-26 on Binance (+2.6%), Bybit (+4.5%) and Hyperliquid (+4.6%); 2020-22 mixed | TESTING (venue check passed for the current era)
- 2026-10-07 | ROT A2 | Active wallets, new wallets, tx counts, wallet share and Wikipedia views do NOT lead chain token vs BTC (13 chains, 2019 to 2026) | KILLED
- 2026-10-07 | ROT A2 | New launchpad reaching 10%+ fee share (3x in a week) on Solana: chain pad fees up next 30d 8 of 8, SOL vs BTC +3.4% in 14d; Base and BSC coin flip | TESTING (n = 8)

## Earlier (v1, still true)
- USD flows (stablecoins, DEX share, fees share, launchpad heat, CEX volume share) lag price by about 8 days. KILLED. Receipts: http://159.223.88.168/hx-c9ois/rot/a2-lead-summary.csv
- Feed sources for the Coder: http://159.223.88.168/hx-c9ois/rot/a2-sources.md

Version: v2, 7 Oct 2026, 12:00 Bangkok. Latest: http://159.223.88.168/hx-c9ois/rot/a2-status.md