> cat pump-graduation.md
pump.fun graduations, by hand
Hypothesis
On pump.fun a memecoin graduates when it crosses a visible threshold, and the attention a graduation draws arrives after it. So, by hand: buy five minutes after a coin graduates, sell at sixty minutes, and let the late crowd pay.
Test
A replay of the rule on a public record of graduations in a fixed window: 4,158 complete round trips after dropping the rows the corpus itself flags as synthetic and the 94 with no liquidity reading, each scored net of fees and price impact, with a bootstrap interval on the mean. Then the same rule on the larger pools only, an exploratory cut, where costs take a smaller share of each trade.
Cause of death
Priced right. Both intervals sit entirely below zero. The larger pools lose less, and still lose. Neither cut comes near break even. The audit's check of this cause: consistent with its own record.
gateNO EDGE
causepriced right
evidenceresult table with interval
auditone of 53 ideas · plan frozen 2026-10-17 · luck-corrected
the number−53.5%, interval −55.8% to −50.9%+ checker −53.5%, match
checked by the audit: its checker compared this cell (run 3, 2026-10-25)
Lesson
The crowd got there first.
This site is not advice and gives no signals or trade ideas.
fig. 1 Mean net return per trade, buying five minutes after graduation and selling at sixty. Dots are the means, lines the 95% intervals, the grey line is break even. The larger-pool row is an exploratory cut.
Data: Slink Dev (slink21taken), PumpFun Launch Corpus, 2026, on Hugging Face, CC BY 4.0.
| Cut | Mean | Interval |
|---|---|---|
| All trades | −53.5% | −55.8% to −50.9% |
| Larger pools, after costs | −14.3% | −20.0% to −7.8% |