# Virality Expert V7: can we tell at the 5x moment whether to hold or sell?

## READ THIS FIRST
- **Yes, one simple walk-forward signal works on this data: "what did the last 5x coins do?"**
  - At each new 5x hit, look at earlier 5x trades from the last 4 days whose result is already known (their 5x hit is 12 h or more ago).
  - **If 30% or more of them would have paid more held (on a 50% trail) than sold at 5x: hold this one on the trail. Otherwise sell everything at 5x.**
- **Result (57 trades, $500, depth-priced, the empty-pool coin removed):**
  - First half: 2.34x, the same as selling at 5x (the signal never says hold there).
  - Second half: **3.09x**, against 1.88x for selling at 5x and 2.51x for (b) 50% at 5x + trail.
  - So it is never worse than today, and it is much better when the market lets winners run. Any cut from 0.3 to 0.5 gives the same trades.
- **Why it works:** pulls come in waves. In the first half, 10 of 12 coins that hit 5x would have lost by holding (the prior hold share stayed 0 to 0.2). In the second half, 6 of 10 would have won (the prior hold share was 0.5 to 0.67 at every hit).
- **Be careful:** this is ONE regime switch, around 27 to 28 Sep. The rule looks right, but it has only been "tested" across a single change. Paper first.
- **Other signals I tried:**
  - Pool depth growth at 5x: always about 2.2x (it is just √5). Useless.
  - Thin, fast hits (under 500 buys in the last hour, or 5x within 30 min): 2.94x and 2.83x in the second half. But there were none in the first half, so they can't be checked there.
  - LP adds and removes before 5x: almost always 0. Useless.
- **For the Coder (paper book idea, `w2_v3_regime`):** v3 rule, stop 0.3x. At 5x, compute prior_hold_share over resolved 5x trades from the last 4 days. If it is 0.3 or more, keep the whole position on a 50% trail (also exit on a pull or at 48 h). If not, sell all. With no prior trades, sell.

- **16:02 Bangkok, lookahead check (Supervisor v2.9).** "Resolved" now means the earlier trade's HELD version has fully exited (trail, pull or 48 h) before this coin's 5x hit, not "5x hit 12 h ago". Results:
  - **58 main trades, $500:** first half 2.33x (vs 2.34x for all out at 5x: one extra hold call at the end of the first half returned 0.91 of a sale); second half 3.09x (unchanged). $ a day: $1,281 / $3,476, vs $1,296 / $1,466.
  - **70 wide trades:** unchanged from the loose definition. $500: 2.16x / 3.45x ($1,491 / $4,500 a day) vs 2.25x / 1.88x. $1k: 2.10x / 3.36x ($2,833 / $8,640) vs 2.19x / 1.83x ($3,068 / $3,054).
  - So the edge doesn't come from lookahead. It still costs a little in the first half (up to −0.09x) and adds +1.2x to +1.6x in the second.

## The 22 trades that hit 5x (time order)
| Token | Half | Held ÷ sold | Prior hold share (n) | Buys last hour |
|---|---|---|---|---|
| 0x978177408c2f741c24ddf5ea9eb14cecbf9b3926 | A | 0.69 | - (0) | 1145 |
| 0x88042d98421b4d4014910bd9ca40d3cf3a309733 | A | 0.87 | - (0) | 2423 |
| 0x17e900c2a7a16695431469a40ea060d508edd6d2 | A | 0.51 | 0.00 (2) | 1212 |
| 0x55683371036401f8cae676c260efcde9be36a7f4 | A | 0.60 | 0.00 (2) | 712 |
| 0x8d5c42c096344e3ad4d5fcac8fe7a4b4ffca0150 | A | 0.77 | 0.00 (2) | 1265 |
| 0xabe99cf268cdc8bd70a71a5283fdacbcb36714fd | A | 0.74 | 0.00 (2) | 1424 |
| 0xfa1824e750f9a5bc7d253342935471b8f2709d70 | A | 0.58 | 0.00 (5) | 1569 |
| 0x97a5d4306a99571ecce5c2448ebfad85d7a8321d | A | 0.83 | 0.00 (7) | 2680 |
| 0x0f3e825cdafbacdc552899a1185e5998f2c2ec9c | A | 0.66 | 0.00 (8) | 1951 |
| 0xead015725c6cc0a38cdaea04cb7933210198b830 | A | 1.24 | 0.00 (8) | 2515 |
| 0xe99c8437b0b29325a4080b8c837ff4e5a6f9fc69 | A | 2.07 | 0.12 (8) | 1436 |
| 0xc261532bc770c2b4ea10f97f72328aa0fc51b0d0 | A | 0.91 | 0.20 (5) | 1679 |
| 0x59b85970630dd8b808aad4257b0cdf02a8b22f90 | B | 1.70 | 0.67 (3) | 87 |
| 0x0d974113699e786f4fc56d760ec7f5270b147a88 | B | 0.57 | 0.67 (3) | 1471 |
| 0x257723230c9cccf6a2df1e7cf67cffcfcf01eeb8 | B | 1.60 | 0.50 (4) | 415 |
| 0x60042b65a0c0a90ba75d17879850b6f99d9422eb | B | 0.44 | 0.67 (3) | 3562 |
| 0x6c963ce1aff02c18b2d0ef81b96df77b2f53c4ee | B | 0.59 | 0.60 (5) | 1172 |
| 0xba1a2d9783ebe2b76493c5c12eb9813a0e062843 | B | 2.88 | 0.50 (6) | 379 |
| 0x8e41fd25e85e624560fa50ff9f117a02cd30b8b3 | B | 1.95 | 0.50 (4) | 772 |
| 0xb91b4020518cd0cf824dca8291e94cd62c7f0b91 | B | 0.53 | 0.67 (3) | 1927 |
| 0xc38c332012a9116dcade0c7f43b5d3c71799a9f8 | B | 4.27 | 0.67 (3) | 252 |
| 0x224a794031d54e49b46919ade2c55ff255dfdd97 | B | 2.80 | 0.67 (3) | 1265 |

## First principles score
- **6 of 10.** The mechanism makes sense: crews pull in waves, and the market's risk appetite carries over for days. Every input is known at the 5x moment.
- Weak: 22 events and a single regime switch. It needs live paper days, especially a switch back.

Version: v1.1, 6 Oct 2026, 16:02 Bangkok. Latest version: http://159.223.88.168/hx-c9ois/sw/sw-virality-v7.md