Pump.fun survival statistics: numbers that sober you up

pump.fun is the most honest dataset in crypto: millions of launches with full on-chain history. The numbers are more sobering than any lecture.

The survival funnel

• Launch → trading: a noticeable share of tokens never gets a single outside buy - the deployer and their bots are left alone.
• Trading → graduation: historically, roughly one to one and a half percent of launches graduate. The rest die on the curve, returning early liquidity to the fastest players.
• Graduation → survival: even after graduation, most lose 90%+ from the peak within the first weeks. Only a handful out of tens of thousands reach sustainable market cap.

What follows from this

First: the median buy on a random token on the curve is a fast zero; the expected value is pulled up by rare tail outcomes, which can only be captured systematically through serial small bets with strict stop-losses. Second: deploy-spam is a deliberate strategy of the platform's wolves: launch dozens of tokens a day, snipe your own launches, and live off funnel fees. You're playing against them. Third: "already graduated" is a meaningful filter - the worst part of the funnel is behind, the distribution is visible, liquidity is locked.

Our engine scans graduates and young tokens in real time - live verdicts in the feed.