The Pump.fun Death Rate: What 120K Launches Tell Us About Token Mortality

By Rekt Robert — 2026-07-04 — the-autopsy

It's hard to talk about pump.fun without sounding like you've spent too much time staring at charts, but the numbers from this past week are stark enough to be worth saying out loud: over half the tokens that launched simply didn't make it. They never graduated to Raydium. They just... expired.

We were watching the market across a 168-hour window in early July, tracking every new token that hit pump.fun's bonding curve. The scale is what gets you first. In seven days, we observed 121,768 launches. That's not a typo. That's nearly 17,000 tokens per day. That's one every five seconds, more or less.

And 61,668 of them died without graduating to Raydium.

That's a 50.6% mortality rate. Slightly better than a coin flip.

What "Dead" Means

We should be precise about this. A token dies on pump.fun when it fails to reach the bonding curve threshold required for migration to Raydium. The curve has a cap. Hit it, and you graduate. Don't hit it, and you're stuck in the pool forever, usually with a handful of holders watching their positions bleed toward zero.

Only 836 tokens actually made the jump to Raydium in that same window. Less than 0.7% of launches.

The rest—59,264 tokens—hung in a kind of limbo. They didn't die, technically. The contract still exists. The holders can still theoretically trade. But they're orphaned. No migration. No secondary market liquidity. No path forward. In practice, you could argue they're dead too, just slower about it.

The Mechanics of Failure

Most launches die almost immediately. You see the token post. Thirty minutes pass. Volume is anemic. The social signal never materializes. A few holders sell their initial allocation just to recover gas, and the price drifts down hard. By hour two, no one new is buying. By hour four, the original deployer has given up marketing it. By day three, the bonding curve is a ghost town.

This isn't tragedy. It's arithmetic. Pump.fun democratized token launching—anyone can do it for a few SOL. And they do. Constantly. Most of those launches are unserious, undercooked, or just noise. Some are rugpulls waiting to happen. A tiny fraction are actually worth someone's time and capital.

The 50% death rate isn't a market failure. It's what you'd expect when the barrier to entry is negligible.

What the Survivors Look Like

The 836 tokens that graduated to Raydium represent something different. Not all of them are winners—some graduate and still collapse. But they cleared a filter that 121,000 others didn't. They had enough social momentum, enough capital inflow, enough coordinated buying power to move the needle.

When we look at the ones that made it, certain patterns are obvious in retrospect. They usually had coherent narrative. They usually had some version of a community before launch. They usually landed with better timing—not necessarily better marketing, but better *timing*, when the market's appetite was higher.

Most importantly, the ones that survived to graduation had staying power in the first 30-60 minutes. That window is nearly everything. If a token can't build momentum fast in those first two hours, it almost never recovers.

The Honest Take

We're analysts, not traders, but anyone serious about pump.fun has to reckon with this baseline. Half the tokens you're looking at won't exist in any meaningful sense six months from now. Some won't exist next week. The signal-to-noise ratio is brutal.

That doesn't mean launches are a waste. It means you have to be ruthless about what you're actually looking for. Most people aren't. Most people just buy because the coin is new and cheap and maybe this one will 1000x. That's not a strategy. That's lottery tickets with slightly better odds than usual.

The tokens that graduate are the tokens that built something—even if it's just convincing enough narrative and community to move a bonding curve. That's not an accident. That's coordination. That's belief, translated into capital movement.

What We'd Do Differently

If we were actively trading this, we'd stop treating every launch like it has a shot. We'd start by assuming 50% death rate and ask which category each token falls into before it launches. We'd spend far less time on technical analysis of meaningless pump.fun charts and far more time on the first 60 minutes of community signal. We'd probably ignore 90% of what passes through pump.fun and focus on the 1-2% that showed real early traction.

We'd also accept that most of the wins, when they come, will feel boring in the moment. The token that graduates quietly, builds a decent community, and 3-5x over a month looks less exciting than the rugpull that 50x in three days. But it's the boring one you actually profit from.

The death rate isn't changing. It's baked into the model. The only question is whether you're honest about it.

Tags: pump.fun, rug, token launch