We follow a sample of real Solana launches, record how each one fared, and file it under the UTC hour and weekday it went live. Every slot shows how many coins it's based on, so a lucky hour can't pass itself off as a pattern.
Dimmed bars rest on fewer than 20 sampled coins. Hours are UTC, all weekdays combined.
Every 20 seconds we read Jupiter's newest-launches feed. All launches are counted for traffic; a fixed hash-based sample is followed: 1 in 40 pump.fun coins, 1 in 5 from the other launchpads.
Share of sampled coins that, at the 1-hour check, trade above their first observed market cap and saw a trade in the past hour.
Share that left the bonding curve (Jupiter's graduatedAt) within 24 hours, among coins at least 24h old or already graduated.
Median of the highest market cap seen at our check-ins up to 24h, divided by the first market cap. Check-ins happen at 15m, 1h, 3h, 6h, 12h and 24h, so short spikes in between are missed and true peaks are understated.
Average launches per hour seen in that slot, a rough measure of how crowded it is.
"Every launchpad" weights each coin by the inverse of its sampling rate. Slots with fewer than 8 coins are striped and kept off the departures lists; hourly bars need 20.