# Virality Expert: the domino map, round 2 (V1 done, V2 first pass)

## READ THIS FIRST (for Ben)
- **Your theory holds, with one twist:** the biggest coins are slow burns, not explosions.
  - In their first hour after $500k they pull in FEWER new buyers than mid coins (22 an hour vs 112 to 151).
  - But they keep pulling them in: on day 2 they still get 26 new buyers an hour, while coins that stall are down to 2 or 3.
- **The dominoes, in order, for $20M+ coins** (median hours after the first $500k hour):
  1. quiet before: no posts with the contract address in the 24 h before;
  2. the first 10k+ account posts right around $500k;
  3. **snowball point at about 4.75 h**: new buyers keep growing for 90 min in a row;
  4. 2,000 holders by about 6 h;
  5. bigger buys arrive in waves: the average buy grows from $133 to $285, and $5k+ buys start after hour 4;
  6. 5,000 holders by about 9 h;
  7. $20M at about 22 h.
- **Where the others stop:** right after the first big X posts, before 1,000 to 2,000 holders. The posts come, the new buyers don't.
- **Follow-the-wallet is real but small:**
  - 67,585 repeated "B buys right after A" pairs, against 135 if the order were random.
  - 95% of them are bots copying inside 60 s.
  - Humans following a wallet (1 to 60 min later): 3,595 pairs. They explain less than 1% of a coin's buyers.
  - So most of the spread runs through channels wallets don't show: X, Telegram, DMs, boards.
- **Trade signal that survived both halves:** at hour 4 after $500k, buy if the coin was quiet before AND new buyers in hours 2 to 4 are still at least half of hour 1.
  - Average +84% per trade, middle trade +25% (half sold at 2x, rest trailed 50%, stop at -50%, 4% costs).
  - First half +93% (80 trades), second half +65% (37 trades). Paper only.
- **A fancy model did worse than these simple rules.** It lost money on unseen days.

## The numbers (569 Robinhood Chain coins that held $500k for a full hour, July to 4 Oct; 45 went to $20M+, 90 to $5 to $20M, 267 to $1 to $5M, 167 stalled under $1M)

| Medians | $20M+ | $5 to 20M | $1 to 5M | under $1M |
|---|---|---|---|---|
| New buyers per hour, hour 1 | 22 | 112 | 151 | 78 |
| New buyers per hour, day 2 | 26 | 24 | 11 | 2.6 |
| Crowd kept on day 2 vs hour 1 | 34% | 15% | 5% | 3% |
| Average buy, day 2 | $285 | $179 | $107 | $71 |
| $5k+ buys on day 2 | 31 | 5.5 | 0 | 0 |
| New buyers who were early in earlier winners (hours 1 to 4) | 19 to 23% | 34 to 36% | 35% | 33% |
| Early holders still holding 24 h later | 74% | 64% | 65% | 75% |
| Top 10 wallets' share at 24 h | 12% | 18% | 21% | 23% |
| Coins with a snowball point | 76% | 63% | 44% | 29% |

- **New people, not the usual crowd:** after hour 1, only about 1 in 5 of a $20M+ coin's new buyers were early in earlier winners. Other tiers: 1 in 3. The big ones reach new groups.
- **Snowball spells:** 2+ spells of growing buyers in 96 h: $20M+ 27% (first half) and 9% (second half), against 5% and 2% with no snowball.
- **Human follow waves:** a strong human-follow wave in the first 4 h: $5M+ 44% vs 31% (first half), 20% vs 11% (second half).
- **Separate follow groups after $500k:** $20M+ coins 10, stalled coins 3.
- **Bots:** in stalled coins the copy bots come BEFORE $500k (median 17 bot-follow buys before, 3 after). In winners they come after (2 before, 15 after). Bots leading is a bad sign.
- **X:**
  - Quiet before still helps: $5M+ 28% vs 18%.
  - Loud before (5+ posts) is worst in the second half: 4% reach $5M.
  - A 100k+ account posting before $500k is not bad: 33% of $20M+ coins had one, vs 14% of stalled coins.
- **Ben's sub-theories:**
  - Someone early tells friends: TRUE but small (human-follow links).
  - Big ones stay quiet, then pop: TRUE for "quiet" and "slow burn"; but they don't pop fast, they keep climbing.
  - Same-size scheduled buys: no signal found (bots copy, they don't buy in equal sizes here).
  - Influencers at the end: partly. The first 10k+ post comes at about $500k for every tier; what differs is whether new buyers follow it.
  - fomo trending, Telegram: still not observable on Robinhood Chain.

## Caveats
- The 569 coins come mostly from the runner tracker, which keeps coins alive in October. Stalled coins are under-counted, so all win rates are too high. The ORDER of steps and the gaps between tiers are what count.
- Hourly prices. Exits ignore liquidity pulls (most of these coins are in pools that were not pulled within 7 days, but that is not checked coin by coin).
- "Trench wallet" history exists only from 13 Sep, so the fresh-crowd step is tested on 129 coins only (too few $20M+ to judge).

## Needs
- From Price Predict:
  - the fair crossing list on all venues (every coin passing $500k, not only tracker survivors);
  - the signed buy tape after 30 min (who signed each buy), to rebuild the follow graph without bots;
  - trending board snapshots (PP7) from now on.
- From Ben (optional): Telegram call groups you know or trust, to test whether they come before or after the snowball point.

## Next (V3 to V5)
- **V3:** group the coins into spread patterns: led by quiet fresh crowd, by bot copy, by an early big X account, by crew/wash.
- **V4:** among coins that reached the snowball point, what separates $5M from $20M (new groups reached, buy-size waves, chain mood).
- **V5:** hand the hour-4 rule (quiet + crowd keeps coming) and a hold rule (second snowball spell) to the Coder as paper books.

## First principles score
- **Score: 7 of 10.** Mechanism tested: price rises only while new money keeps arriving.
- Measured directly: new buyers, buy sizes, holders and who follows whom. Shuffle test on follow links; both halves checked.
- Missing: fair coin list, signed tapes, the board and Telegram channels.

Version: v1, 6 Oct 2026, 13:10 Bangkok. Latest version: http://159.223.88.168/hx-c9ois/sw/sw-virality-v1.md