Every row below was committed to disk before the outcome existed — timestamped, unedited, misses included. This page regenerates straight from that ledger. Nothing here is chosen after the fact.
Each launch gets a notability rank at birth — how many separate wallets minted the same name in one burst. Rank it against what actually happened next and a clean inverse pattern appears: the louder the swarm, the less often the coin graduated off pump.fun’s bonding curve.
| Notability band | Swarms watched | Graduated | Rate |
|---|---|---|---|
| 0–20 | 37,640 | 3.15% | 1,185 |
| 20–35 | 28,978 | 1.49% | 431 |
| 35–55 | 10,436 | 1.43% | 149 |
| 55–70 | 2,558 | 0.98% | 25 |
| 70–100 | 40 | 0.00% | 0 |
Reading it plainly: a quiet swarm graduated roughly 3.1% of the time; the loudest ones under 1.0%. Attention is not adoption — across 79,652 resolved cases, hype anti-predicted success. That is the usable signal, and it is the opposite of what feed algorithms want you to believe.
A watcher records pump.fun name-swarms as they mint. At commit time the algorithm knows only: swarm size, wallet history, name triggers, and the linked account — never the outcome. Resolution is mechanical (bonding-curve completion read from chain state) about 2.0 hours later, median. Both steps are timestamped; nothing is edited afterward; misses stay public. Sample window: 43.6 days, 79,652 commits, 0 still resolving.
Not financial advice. Most tokens go nowhere regardless of any signal; assume you can lose everything you put in. Page auto-regenerates from the ledger — last build 2026-09-26 04:05 UTC.