The median pump.fun launch does not crash in its first six hours. It does not run either. It sits there.
Our launch corpus holds 1,360 coins with a full six hour price trace. The median coin’s highest print across the whole trace sits 0.1% above the price we first caught it at. Not 10% above. Not double. One tenth of one percent above where it started, at its very best moment.
The numbers
A “first print” is the first price our own poller caught after launch. The trace after that is a sampled price path, print by print, out to six hours.
| Measure | Median across 1,360 traces |
|---|---|
| First print | $3,102 |
| Highest print of the six hours | $3,206 |
| Last print | $2,916 |
Read those as two medians, not as one coin’s story. Half the sample started below $3,102, and half of them never printed above $3,206 at any point in six hours. The middle of this market sits between those two numbers, and almost nothing in it moved.
The counts behind the medians:
- 819 of 1,360, or 60%, hit their high print inside the first five minutes.
- 76% never gained even 5% over their first print at any point in the six hours.
- 83% never gained 20%.
- 109 coins, 8%, doubled at some point.
- 22% ended at or below half their first print.
- 54% finished within 10% of where they started.
“Peaked inside five minutes” sounds dramatic until you ask what the peak was. Of the 819 coins that hit their high print in the first five minutes, 599 of them, 73%, got there with a peak no better than 0.1% above the first print. A coin can set its all time high in minute three and have no high to speak of.
One tenth of one percent was the best moment of the median coin. Six hours for that.
What it means
The middle of this market is not a tragedy. It is a shelf. Coins launch, trade a little, and stay put.
Only 22% of the sample ended at half or less of its first print. The blowup is a tail outcome, not the middle one. The median coin mostly just stopped being interesting.
Put the two tails side by side and the shape is clear. 109 coins in 1,360 doubled at some point in six hours. 298 ended at half or less of their first print. The middle of the distribution is fat: 54% of the sample finished within 10% of where it began. This is a shelf with two thin edges, not a crash next to a moonshot.
The runners are that small edge, and it lines up with what we measured from the other end. The ladder post counted that three in four launches never reach $5,000 of market cap at all. The first hour piece found most peaks land before the first hour is out. This cut adds the missing piece. For the median coin, the peak is barely a peak. It is the launch price with a rounding error on top.
So a coin that is hours old and has not moved is not a stalled setup. It is the median outcome, not the exception to one. Whatever it does next is a separate question, and this record does not answer it. It only says where the middle of the pile sat: nowhere much, for six hours.
The live board shows these coins as they arrive, with the book on each card. PumpCheck pulls the pool and the holder list for any single one of them. The book is who holds what, and the pool is the money actually sitting in the market. Both tools show what a coin is, not what it becomes. The flat middle above is why that is the honest thing to show: most of these books never move enough to be worth calling.
Method
Sample: 4,659 pump.fun launches observed by our own price poller since late October. Of those, 1,360 had at least four prints and a last print at least 5.5 hours after launch. Every number above is a median or a count of that 1,360, recomputed on the day of this post. A trace is sampled, not every trade, so a spike between prints can be missed. The strict six hour cut also favors coins that lived long enough to be caught repeatedly, which likely excludes some instant deaths and flatters the middle of these numbers. Counts describe this sample only. Nothing here predicts any coin’s next six hours.