Of the 7 million tokens launched on pump.fun between January 2024 and March 2025, only approximately 97,000 maintained more than $1,000 in liquidity after launch. That is a 1.4% survival rate. The other 98.6% — over 6.9 million tokens — showed pump-and-dump characteristics: an initial surge followed by rapid collapse, usually within the first hour.
This is not a bug in pump.fun's design. It is the predictable output of a system where token creation costs nothing and most launches are engineered exits rather than genuine projects. The more useful question is not "why do most tokens fail?" but "what do the 1.4% that survive look like on-chain before they move?"
The data shows clear patterns. The failures and the survivors differ across four measurable dimensions that are visible before you buy: deployer history, buyer quality, holder concentration, and early volume composition.
The Four Failure Modes
Understanding why tokens die in the first hour starts with the mechanism of death. There are four common patterns:
The coordinated dump: The deployer, early buyers, and sometimes KOLs are coordinated. The deployer launches with a bundled buy to build initial momentum, socially amplifies the token, and sells into the retail buyers who enter after the call. Price spikes 400–800% in minutes, then collapses back below launch price within the hour. The bonding curve shows a clean pump-and-dump shape: rapid rise, sharp cliff.
The bot-dominated launch: Volume looks organic in aggregate but the early buyers are almost entirely bots — sniper bots that enter in the first block and MEV bots that extract value from every subsequent trade. There are no real holders. When the bots rotate out (usually within minutes), there is no demand to sustain the price. Token dies quietly without a dramatic dump. A related trick is outright manufactured volume — our explainer on how Solana volume bots fake activity and how to spot it covers how to tell wash volume from the real thing.
The abandoned launch: Developer deployed the token, made no effort to build community or liquidity, and simply waited to see if organic momentum appeared. It did not. Token sat at sub-$5K market cap with 20 holders for 48 hours and quietly expired. Occasionally one of these gets a second life when traders organize a community takeover and score the CTO before it pumps — a rare exception to the first-hour death rule.
The honeypot or freeze: Token contract has active freeze authority or a hidden mechanism that prevents selling. Early buyers discover they cannot exit. Token shows price appreciation on charts but is economically worthless for anyone who bought.
Each of these failure modes leaves a different on-chain fingerprint. None of them are invisible before you buy — they are readable at launch if you know what to check. Several of these patterns — especially the honeypot and freeze variants — overlap with outright scams, which we cover in our guide on how to spot and avoid rug pulls on Solana.
What Survivors Look Like at Launch
Analysis of the tokens that maintain liquidity and graduate to PumpSwap shows consistent patterns in the first 30 minutes:
Diverse early buyer base: Surviving tokens have their first 20–50 buyers distributed across wallets with different funding sources, ages, and trading histories. Failed tokens have their first 10–20 buyers concentrated among wallets created on the same day or funded from the same source — the hallmark of a bundled launch.
Human traders in the first-buyer list: The ratio of wallet addresses that show human trading patterns (variable trade sizes, diverse token history, consistent but not mechanically precise timing) versus bot patterns (uniform sizing, sub-second timing, high transaction failure rates) in a token's first 50 buyers is a strong predictor of whether the token will sustain. Research on Solana memecoin data shows tokens dominated by bot activity in their first 50 buyers have systematically lower graduation probability.
Deployer track record: The wallet that launched the token has a history. If that wallet has deployed 40 previous tokens that all dumped within 30 minutes, the current token is almost certainly the same. If the deployer has 3 previous tokens — two of which graduated and one that failed — the base rate is meaningfully different.
Graduated market cap trajectory: Tokens that graduate (roughly 85 SOL raised on the curve) from pump.fun's bonding curve show consistent buy-pressure in their chart — multiple price discovery cycles, not a single spike. Failed tokens show one steep spike and immediate collapse. The shape of the bonding curve in the first 20 minutes distinguishes these before graduation.
Reading Deployer History
The deployer wallet is the most predictive single variable. A deployer who has launched 60 tokens with a 2% graduation rate is a fundamentally different counterparty than a deployer with 8 prior launches and 5 graduates.
What to check:
- Number of prior token launches from this wallet
- Graduation rate (tokens that reached Raydium)
- Average time-to-death on non-graduates (did they dump in minutes or fade over days?)
- Whether the deployer wallet received tokens back from prior launches (insider allocation pattern)
- Funding source of the deployer wallet (fresh wallet funded from exchange = anonymous operation; aged wallet = more accountability)
A deployer wallet with 0 prior launches is not automatically bad — everyone has a first token. But a wallet with 50 prior launches and a 0% graduation rate is a documented serial dumper.
This check takes 2 minutes on Solscan. Filter the wallet's transaction history for token creation events, check how many prior tokens still have liquidity on DexScreener.
