Most of what a launch will ever be, it becomes in minutes. We went back through the launch record this week with one question: how much of a coin’s life actually happens inside its first hour. The answer is almost all of it.
The number that matters is 86. That is the percent of traced launches that make their all-time high, at least within the six hours we follow them, inside the first 60 minutes. More than two thirds of those peak inside five minutes.
Here is the full split, from 2,471 launches we can trace. These are launches the live board caught as they appeared on pump.fun, not a hand-picked list.
| When a launch makes its high | Share |
|---|---|
| Inside 5 minutes | 65% (1,611) |
| 5 to 15 minutes | 13% (319) |
| 15 to 60 minutes | 7% (185) |
| After the first hour | 14% (356) |
The late group is real but thin. One launch in seven peaks after the hour mark. That is the group that keeps a launch feed watchable. It is not the group that describes a typical launch.
The median launch tells the same story, flat. One minute in, it is worth $2,936. One hour in, $2,758. Six hours in, $2,754. The whole slide fits inside hour one, and the remaining five hours move the median by about four dollars. Per coin, the median value at hour one and at hour six is the same. Nothing happens, and then it keeps not happening.
Most of these launches never went anywhere to begin with. In the full traced set, 25% ever reached a $5,000 market cap, 15% reached $10,000, 5% reached $100,000, and 2% crossed $1,000,000. A launch standing still at $2,900 is not an early failure of a big coin. It is the median outcome.
What the hour mark actually shows
For the timing claims below we use the 1,770 traces that genuinely extend past the hour mark. Of those, 1,499 sat below $5,000 at the hour mark. Seven of them ever reached $10,000 afterward. Seven out of 1,499, half a percent. At this sample size, a coin that has shown nothing after an hour had almost no second act.
Quiet is only half of the first hour. The coins that ran rarely waited. In that same traced group, 334 ever touched $10,000. 96% of them touched it inside fifteen minutes of creation. And 37% were already back below $5,000 by the hour mark. The whole ride, up and down, fit inside hour one.
One more cut. A third of traced launches (578 of 1,765 with both readings) were worth more at the hour mark than at minute five. Two thirds were not. The hour is mostly downhill or flat, not a ramp.
There is a version of this that sounds tragic, spikes collapsing. The record does not show that. The 1,611 launches that peaked inside five minutes held a median value of 90% of peak at the hour mark, and 90% again at hour six. The peak was not a flash. It was the ceiling, and the coin sat just under it all day.
What that means if you read the board
A launch is not warming up during hour one. There is no settle-down phase after which a coin reveals itself. In this record the reveal came in minutes or it did not come, and by the hour mark most launches sat in one of two piles: already done, or never started.
That is worth knowing before you attach meaning to a coin being a few hours old. Age on its own is not a signal. If anything, the trace says an hour of patience mostly buys you the same coin at 90 cents of its peak dollar, or a coin that never left $2,900. The bigger caps that do climb run into a dead band between $400,000 and $1.4 million, a gap we measured separately. And all of these prices move on a bonding curve, which is the mechanism every launch rides before it graduates.
None of this tells you what any single launch will do next. It tells you what 2,471 of them did, and when.
- Watch launches appear in real time on the firememecoins.com board.
- Run any launch through the free pumpcheck for its holder counts and pool state.
- Background on the six-hour window is in our first launch-data post, and the stall band in where pump.fun coins stop.
Method
The sample is our launch corpus: 4,659 launches observed on pump.fun between late November 2025 and November 4, 2026. Of those, 2,471 have a price trace with four or more points inside the first six hours, and all figures about peaks and market cap thresholds use that set. Median market caps come from 2,400-plus readings at each mark, per minute one through minute 360. The claims about the hour mark use the 1,770 traces that extend past 55 minutes, and the $10,000 timing uses the 334 of those that ever touched it. Traces are sampled, not continuous: a coin can move and round-trip between two points without being seen. Six hours of follow-up means a coin that runs on day two is counted at its six-hour value, not its later peak. Percentages are rounded, so bucket shares can read as 100 without being exact.
