# 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.

# A1.22 Is the edge fading? (Analyst 1, 6 Oct 2026, 20:40 Bangkok)

## Read this first
- **No. The edge is flat, not falling. No trend in $ a day passes a basic noise test, over the 16 trading days or over the last 7.**
- **The one real trend is good news: new buyers per trade are rising** (+106 a day, p = 0.02 over 16 days).
- **One thing to watch:** $1M gates per day in the last 7 days slope down (−1.0 a day, p = 0.11). Today (6 Oct) is a part day with 2 gates so far.

| Trend line (per day) | 21 Sep to 6 Oct (16 days) | Last 7 days (30 Sep to 6 Oct) |
|---|---|---|
| $ a day, all out at 5x ($1k, A2.21 exact) | −$58 ± $244, p 0.81 = **flat** | +$696 ± $840, p 0.45 = flat |
| $ a day, kit hold (my estimate, see Limits) | +$566 ± $420, p 0.20 = flat to up | +$2,413 ± $1,584, p 0.19 = flat to up |
| New buyers per trade, 2 h after the gate (median) | **+106 ± 42, p 0.02 = rising** | +357 ± 191, p 0.12 = rising, not certain |
| $1M gates per day | 0.0 ± 0.2 = flat | **−1.0 ± 0.5, p 0.11 = watch** |
| Kit A, Kit B and 3,587-family launches per day | +1.8 = rising | +2.0 = rising |

- **Days until $ a day crosses $2,000 going down:** none on the fitted line, because it is not falling.
  - The most pessimistic version is all-out $ a day from 21 Sep to 5 Oct, leaving out today: −$160 a day, from a fitted $3,056. That would reach $2,000 in about 7 days.
  - But that slope is smaller than its own error (±$271, p 0.57), so it is noise, not a trend.
- **Halves for comparison (all out, $1k):** the first 8 trading days average +$4,532 a day, the last 8 +$4,227. Kit hold estimate: +$6,022, then +$9,630.
- **Kit launches did not fall.** Kit A stopped on 1 Oct, but Kit B took over: 9 to 50 launches a day since 1 Oct, 14 so far today.
- **First principles:** our profit is the next buyer paying more. The supply of that buyer is new wallets arriving after the gate, and that supply is growing (A1.18 showed most of them are fresh disperse-funded wallets). The flow of coins is flat. The share reaching 5x went from 43% (first 35 trades) to 37% (last 35), which is inside the noise (about ±12 points). The risk is not decay, it is noise: one day swings from −$3,161 to +$12,069, so any 7-day line has an error bigger than $2,000 a day.

## Per day (Bangkok), live rule: wide gate, buy at gate + 30 s, 0.3x stop, sell at end of next block, $1k real depth (A2.21)

| Day | $1M gates | Trades | Mean x (all out) | Hit 5x | $ all out | $ kit hold (estimate) | New buyers per trade, 2 h (median) | Kit A/B/3,587 launches |
|---|---|---|---|---|---|---|---|---|
| 09-18 | 1 | 0 |  |  | +0 | +0 |  | 0 |
| 09-19 | 0 | 0 |  |  | +0 | +0 |  | 0 |
| 09-20 | 0 | 0 |  |  | +0 | +0 |  | 0 |
| 09-21 | 4 | 5 | 1.09 | 20% | +446 | +446 | 825 | 0 |
| 09-22 | 6 | 7 | 2.72 | 57% | +12,069 | +12,069 | 981 | 0 |
| 09-23 | 7 | 7 | 2.16 | 43% | +8,150 | +8,150 | 850 | 0 |
| 09-24 | 6 | 3 | 1.75 | 33% | +2,260 | +2,260 | 628 | 1 |
| 09-25 | 15 | 6 | 2.44 | 50% | +8,636 | +18,149 | 757 | 37 |
| 09-26 | 5 | 5 | 1.08 | 20% | +407 | +2,928 | 828 | 17 |
| 09-27 | 5 | 3 | 3.10 | 67% | +6,289 | +6,174 | 1,224 | 10 |
| 09-28 | 7 | 2 | 0.00 | 0% | −2,000 | −2,000 | 346 | 20 |
| 09-29 | 9 | 6 | 1.63 | 33% | +3,752 | +1,517 | 617 | 15 |
| 09-30 | 8 | 4 | 1.23 | 25% | +932 | +3,293 | 854 | 11 |
| 10-01 | 10 | 4 | 3.49 | 75% | +9,975 | +18,600 | 624 | 9 |
| 10-02 | 10 | 4 | 0.21 | 0% | −3,161 | −3,161 | 1,074 | 31 |
| 10-03 | 4 | 2 | 3.11 | 50% | +4,214 | +4,214 | 3,024 | 26 |
| 10-04 | 10 | 5 | 1.92 | 40% | +4,623 | +15,092 | 1,037 | 50 |
| 10-05 | 5 | 5 | 2.20 | 40% | +5,996 | +16,062 | 1,344 | 23 |
| 10-06 (part day) | 2 | 2 | 4.74 | 100% | +7,483 | +21,423 | 3,716 | 14 |

- 14 to 17 Sep had no gates and no trades. Lines are fitted from 21 Sep, the first trading day, so the empty days do not fake an upward slope.
- CSV: http://159.223.88.168/hx-c9ois/sw/a1-22-per-day.csv

## Method
- **Trades and all-out $:** Analyst 2's A2.21 per-trade x at $1k (all frictions), grouped by Bangkok day of the trade.
- **Kit hold estimate:** for the 15 kit trades that hit 5x (Kit A 9, Kit B 6), hold the whole position from the 5x moment on a 50% trail from the peak, using each pool's own swaps up to 12 h after launch (A1.15 data). The ratio exit price ÷ price at the 5x moment is applied to A2.21's 5x x.
- **New buyers per trade:** wallets that signed a buy in the 2 h after the gate and had not bought that coin before the gate (A1.18 data, all buys of the 70 coins).
- **Gates:** the 114-coin sellable $1M gate set, by gate time. **Kit launches:** pools whose launch `to` code hash is Kit A, Kit B or the 3,587-byte family.
- **Trend:** ordinary least-squares line on daily values, with standard error and p-value. Under p 0.05 counts as real.
- I also checked PP's pp3-series (526 pools). Its new-buyer counts after the $1M crossing cover every coin that crossed (v4 and dust coins included), and they jump around too much day to day (median 1 to 1,305) to read a trend. I used my own counts for our 70 traded coins instead.

## Limits
- **The kit hold estimate runs high:** +$3,584 / +$8,338 a day vs A2.24's exact +$3,317 / +$6,683. It ignores exit depth and pulls, and values holds still open at 12 h at their last price. Use it for the direction only; A2.24 has the exact level.
- 16 trading days is short. The p-values say how weak any slope is.

## Asks for Supervisor
- No size cut on decay grounds. The edge is flat and the buyer supply is growing.
- Watch line for the Coder: if $1M gates per day stay at 4 or fewer for 3 days in a row, ring A1 to re-check the coin flow (Kit B launches are still rising, so a gate drop would mean fewer launches turning into runners, not fewer launches).
- I can re-run this table daily, alongside the paper book check.

First principles score: 85%. The money is A2.21's exact per-trade x and the buyers come from every buy transaction, with trends tested against their own noise. Missing: exact kit-hold x per trade (A2.24 per-trade file), and more days.

Scripts: /root/a1/a122/kh.py, day.py (review container). Latest: http://159.223.88.168/hx-c9ois/sw/a1-22-edge-decay.md