A launch that reaches $25,000 does not need a crowd. We split 293 of them from our launch corpus by how many unique wallets sat on the trade tape when we first read it. The lower half, fewer than 25 wallets, is a handful of accounts trading the same bag over and over. It spikes fast. By the last point in its trace the market cap is back near single digit thousands.
The upper half is a different animal. Twenty-five wallets or more, repeat volume near half, and a far better chance of still being worth something hours later. Same board, same candles, two different machines underneath.
Definitions, so the numbers mean something. A runner is a launch whose traced market cap touched $25,000 at any point in our record. The tape is the trade log we snapshot in a coin’s first minutes, a median of 31 minutes after creation for this set. Unique buyers are the distinct wallets on that tape. Repeat share is the percent of trades from wallets that bought three or more times.
Here is the narrow half. 158 runners. Median repeat share 95%. Median 16.5 trades per buyer, against 172 trades in total. A few accounts simply trade the bag repeatedly until the print looks big. The median peak was $95,999, and 76 of 158, or 48%, touched $100,000. So the print is real. The demand behind it is not broad.
It gets sharper at the extreme. 33 of the 158 had five or fewer unique wallets. Twelve of those 33 still touched $100,000. Five accounts can print a six figure market cap for a few minutes. That is what a thin tape can do on its own.
The broad half, 135 runners, looks like a market instead of a loop. Median repeat share 50.5%. Median 2.4 trades per buyer, median 499 trades in total. Median peak $181,728, and 78 of 135, or 58%, touched $100,000.
| Under 25 buyers | 25 buyers or more | |
|---|---|---|
| Runners | 158 | 135 |
| Median repeat share | 95.0% | 50.5% |
| Median trades per buyer | 16.5 | 2.4 |
| Median peak market cap | $95,999 | $181,728 |
| Touched $100,000 | 48% | 58% |
| Peaked inside 5 minutes | 92% | 55% |
The timing split is stark. 145 of 158 narrow runners, 92%, peaked within five minutes of creation. For the broad half it was 74 of 135, or 55%. The narrow spike is mostly a wick at the open. That fits the wider picture we measured in the first hour problem, where most launches never build on their first candle. It also fits the launch ladder, since three in four never reach $5,000 at all.
Then the round trip. Staying power can only be measured on runners whose trace ran long enough, past 5.5 hours. That is 204 of the 293: 82 narrow, 122 broad.
Of the 82 narrow, the median market cap at the last recorded point was $2,507. Eleven, or 13%, were still above $10,000. The median close sat at 1.6% of peak.
Of the 122 broad, the median at the last point was $11,381. Sixty-six, or 54%, were still above $10,000, and 38 held above $25,000. The median close was 8.1% of peak.
Neither half keeps the money. Even the broad runner gives back more than 90% of its peak at the median. The difference is that the broad one often leaves a market cap with some depth in it. The narrow one usually leaves a few thousand dollars and the same five wallets.
What this means when you read the board. A tall green candle on a fresh launch measures churn, not demand. The same print can come from five patient wallets or from five hundred, and the two shapes end very differently. The buyer count behind the tape is the number that separates them, and it sits on the card for every launch we score. The checker prints buyers and repeat share live for any mint, so the tape is checkable before anyone has to trust the candle.
None of this says broad means buy. A close at 8.1% of peak is still a close at 8.1% of peak. It says the tallest number on the screen, the candle height, is the least informative one. We looked at winners and losers through this same lens in an earlier buyers study, and the corpus keeps agreeing.
Method. The sample is our launch corpus, 4,659 pump.fun launches observed from creation. A runner is a launch whose traced market cap touched $25,000. 328 did, and 293 of those had a readable trade tape at snapshot time. The tape snapshot is taken in a coin’s first minutes and covers the most recent 499 trades: unique wallets, repeat share, trades per buyer. We split at 25 unique buyers, near the median of 22 for this set. Staying power was measured only on the 204 runners with a trace past 5.5 hours, 82 narrow and 122 broad, using the last recorded point of a sampled price path rather than tick data, so a later recovery can be missed. The snapshot lands after the peak for most narrow runners, so this is a description of two shapes, not a lead lag test. For scale, 239 of the 2,471 launches with a full six hour trace, about 10%, ever touched $25,000. Every number here describes that small and unusual slice.
