Does the clock matter?
Every pump.fun graduation with a standard seed (≥67 WSOL) from our archive, banded by the UTC hour it graduated. “Doubled” means the token peaked at least 2× its graduation price inside our tracking window.
Measured 05 Sep 2026 · 26,549 graduations over 95 days · rebuilt daily| Graduated | n | Doubled | Ended positive | Median peak | Early / late half |
|---|
| 00–04 UTC | 4,620 | 57.8% | 46.1% | +177% | 54.4% / 60.9% |
| 04–08 UTC | 3,703 | 60.1% | 49.6% | +211% | 50.4% / 69.4% |
| 08–12 UTC | 3,131 | 57.7% | 52.3% | +188% | 49.7% / 65.5% |
| 12–16 UTC | 4,707 | 54.0% | 44.7% | +136% | 46.6% / 61.9% |
| 16–20 UTC | 5,552 | 45.1% | 37.6% | +76% | 38.2% / 52.5% |
| 20–24 UTC | 4,836 | 48.1% | 37.7% | +90% | 41.6% / 54.5% |
The last column splits the archive into its first and second halves: the ordering holds in both, which is what separates a pattern from a fluke. The mechanism is plausible but unproven — launches during US afternoon hours compete with the day’s heaviest flow for the same capital.
What this is not. Better odds of a peak are not odds of profit — the median graduation still ends far below its peak, and most end negative at every hour. This shifts a base rate; it does not make a bad trade good. Costs do not care what time it is: fixed costs still take
~0.0085 SOL a round trip.