Skip to content
MIND

Pump.fun Rug Pull Statistics: How Many Tokens Fail?

MIND Research· · 9 min read

Liquidity pulled
On this page9 sections
  1. How many Pump.fun tokens are rug pulls?
  2. What is a rug pull and what is a pump-and-dump?
  3. How do launchpad rugs work?
  4. What do launch buyers lose?
  5. What are the warning signs of a rug pull?
  6. How does a funded launch change the risk?
  7. What do these statistics not show?
  8. Is Pump.fun the problem?
  9. What should you do before buying a new token?

Almost every token launched on Pump.fun dies, and most of them are rugged on the way down. Solidus Labs found that 98.6% of more than 7 million Pump.fun tokens created between January 2024 and March 2025 collapsed into rug pulls or pump-and-dumps, and Chainalysis found that when a pool was flagged as a pump-and-dump, the creator removed the liquidity 94% of the time.

This article lays out the main Pump.fun rug pull statistics, explains how the rugs work, lists the warning signs, and shows how a funding gate such as MindPad changes who carries the risk. It is not financial advice.

How many Pump.fun tokens are rug pulls?

The largest study answers bluntly: almost all of them. The table sets out what each source measured.

Source Period Finding What it measures
Solidus Labs Jan 2024 to Mar 2025 98.6% of more than 7 million Pump.fun tokens with at least five trades fell below $1,000 in liquidity. Only about 97,000 held more. Tokens that collapsed into rug pulls or pump-and-dumps shortly after launch, measured as liquidity under $1,000. Tokens that completed the curve and moved to Raydium were excluded.
Solidus Labs Period not stated About 93% of 388,000 Raydium V4 pools, roughly 361,000, showed a soft rug pull. Median loss was about $2,832. Pools where at least 90% of liquidity was withdrawn.
Chainalysis 2024 74,037 of 2,063,519 new tokens, or 3.59%, showed patterns linked to pump-and-dump schemes. In 94% of the suspect pools, the pool creator removed the liquidity. All tokens across chains, not Pump.fun alone.
arXiv, The Memecoin Phenomenon Q4 2024 Fewer than 2% of Pump.fun tokens graduated to a major decentralized exchange. Graduation from the bonding curve.
arXiv, 832,941 launches 8 May to 10 June 2026 A pooled graduation rate of 0.198%, which the authors call a lower bound. Only graduations within about six minutes were visible.

Why is the Chainalysis share so much lower? It applies a strict pattern test across every chain and every launchpad, while Solidus looks at Pump.fun alone. Both point the same way. On an open launchpad the default outcome for a buyer is a dead token, and the usual way it dies is a rug or a dump.

The graduation figures also move with the platform's rules. A July 2026 report from The Block put the graduation rate at 6.7% on one Friday after Pump.fun introduced a mechanism called BOOST, about eight times the June average. The same report said the rate averaged 4.7% over the previous four days. A single day is not a trend, and the article does not name its data source.

What is a rug pull and what is a pump-and-dump?

A rug pull is an exit where the people who control a token take value out and leave buyers holding a token that no longer trades near its old price. A pump-and-dump is a coordinated rise in price followed by a sale into the people who bought the rise. The two overlap, and launchpad data rarely separates them.

Solidus Labs uses "soft rug pull" for the most common form on Solana. Liquidity is abruptly withdrawn, the price collapses, and no hack or contract exploit is involved. In the Raydium sample, a liquidity sweep meant removing at least 90% of a pool's liquidity. The median loss in that sample was about $2,832. A quarter of the rug pulls involved less than $732, and the largest was $1.9 million.

Those sizes matter. Rugs on launchpads are mostly small and frequent, which is why they escape headlines. A large number of small losses adds up to a large number, and few buyers report them.

How do launchpad rugs work?

Four patterns recur, and they often appear together.

  1. Liquidity pull. The creator seeds a pool, waits for buyers and removes the liquidity. Chainalysis found the pool creator was the remover in 94% of the pump-and-dump pools it flagged.
  2. Creator dump. The creator or a linked wallet holds a large share of supply and sells into buyers as the price rises.
  3. Sniping and bundling. Bots or wallets tied to the creator buy in the first block, before ordinary buyers can react, then sell to the later arrivals. MindPad's own description of open launchpads is blunt about this, and says they "reward speed. Bots buy the first block, creators sell into retail, and nobody asks whether the token should exist."
  4. Manufactured hype. Paid promotion, fake volume and coordinated posting create the appearance of demand while insiders prepare to sell.

Pump.fun's design removes some of these. The Memecoin Phenomenon study describes a fair-launch design with no presale and no team allocation, where buyers trade on a bonding curve and can sell at any time. The study says the design helps prevent classic rug pulls, and it does not measure rug pulls itself. The design removes the presale dump. It does not stop a creator from buying early through several wallets, or from promoting a token hard and selling into the rise.

A federal class action in New York goes further. The amended complaint calls Pump.fun a "front-facing slot machine cabinet" and alleges it extracted more than $5.5 billion from users. It also names Solana-linked entities. These are allegations, and Cointelegraph reported no response from the defendants at publication. We cite it to show how seriously the harm is being litigated, not to claim the outcome.

What do launch buyers lose?

Most buyers lose their stake, and many lose it fast. If more than 98% of tokens end below $1,000 in liquidity, a random buyer is far more likely to hold a dead token than a winner. The loss per event looks small, with a median of about $2,832 in the Solidus Raydium sample, and the total is large because there are so many events.

There is a second cost that does not show up in a loss figure. When a market is full of dead tokens, honest projects have to compete with them for attention. A founder who wants to build a real product on a launchpad is judged against thousands of exits. That is the case for a more controlled onchain market, one where launching requires someone to commit capital first.

What are the warning signs of a rug pull?

No checklist catches everything, and some rugs look clean until the exit. These signs raise the odds.

  • Concentrated supply. A few wallets hold a large share, or the top holders were funded from the same source.
  • First-block buying. Many wallets bought in the same block as the launch, which points to bundling or sniping.
  • Unlocked liquidity. The creator can still remove the pool. Prefer liquidity that is burned or locked on a schedule you can verify.
  • Retained powers. The creator can still mint new tokens or freeze accounts. Check the token's authorities in a block explorer.
  • Anonymous team, loud promotion. No verifiable people, and heavy paid promotion that arrives before a product does.
  • Guaranteed returns. Any promise of a fixed return or a price floor from an unknown party is a red flag.
  • Volume without depth. High reported trading volume next to a thin pool suggests wash trading.
  • Approval requests. A site that asks for unlimited token approvals is asking for more than it needs. MetaMask advises checking and revoking old allowances regularly.

These signs reduce risk. They do not remove it, and a token can pass every check and still fall 90% in a day.

How does a funded launch change the risk?

A funded launch moves the decision before the market opens. On an open launchpad, the creator launches and the market decides afterward, at the buyers' expense. In a funding model, backers decide first and the market opens only if they do.

MindPad is a Solana launchpad in development that works this way. Its lifecycle has five steps.

  1. Propose. A creator sets the token, venue, raise target and deadline. The terms are hashed and published.
  2. Back. Backers commit SOL or USDC into program-governed escrow. Every contribution is recorded per wallet.
  3. Funded or refunded. If the target is hit by the deadline, the raise locks. If it is missed, the launch never happens.
  4. Launch on a venue. The token goes live through Pump.fun, Raydium or Meteora, with the funded buys disclosed.
  5. Graduate and trade. The venue creates the pool. Agents then manage liquidity under signed mandates.

If the target is missed by the deadline, or the creator cancels while that is allowed, backers claim a full refund. That is the main change. The first buyers of a launch are backers who chose to fund it, not bots racing a block.

Be clear about the limits. A funded token can still fall in price after it launches, and a refund covers a missed goal, not a falling market. MindPad does not guarantee returns, and it is not available yet. What it changes is the structure. Capital goes in before the launch, the terms are public, and a token that fails to raise never reaches the market.

For how agents manage liquidity after a launch, see agentic trading and market making. For the company behind MindPad, see What is MIND?.

What do these statistics not show?

Three limits apply to every number above.

First, none of the studies proves intent. A token can fall below $1,000 in liquidity because its creator left, because nobody wanted it, or because it was never meant to be more than a joke. Solidus counts all three the same way. Chainalysis tries to isolate deliberate exits, and it still labels its results patterns that "may be linked" to pump-and-dump schemes, not convictions.

Second, the data covers different periods and different rules. Pump.fun changed how launches work between 2024 and 2026. The Memecoin Phenomenon study describes the late 2024 setup, and the July 2026 BOOST report describes a new default launch mechanism. A statistic from one year does not carry over cleanly to the next.

Third, the 2026 graduation study is explicit about its blind spot. Its collector could only reliably see launches that graduated within about six minutes, so the authors call 0.198% a lower bound. Its own corrections show how easy it is to get this wrong. The authors tried to recover missed graduations and found, after checking 100 sampled tokens against Pump.fun's live state, that none had graduated, so they withdrew the adjustment.

The safest reading is the plain one. Across every method, a small share of launches reach a lasting market, and the rest fade or end in an exit.

Is Pump.fun the problem?

Pump.fun is a venue, and the studies above describe the behavior of a whole category of launches, not one company's intent. The platform lowered the cost of creating a token to almost nothing, and the result was millions of tokens, most of them dead. Any open launchpad with the same economics would see the same pattern.

That is why MindPad treats Pump.fun, Raydium and Meteora as venues where a funded token can go live rather than as competitors. The filter sits before the venue.

What should you do before buying a new token?

Treat every new launch as a high-risk position. Decide the amount you can lose entirely before you open the chart. Check holder concentration, liquidity status and the creator's powers. Look for the same wallet cluster buying at the start. Avoid tokens that rely on urgency, and never connect a wallet that holds more than you intend to risk. A sensible default is to wait and let a token show a track record before you commit money.

Frequently asked questions

What percentage of Pump.fun tokens are rug pulls?

Almost all of them die, and most are rugged or dumped. Solidus Labs found that 98.6% of more than 7 million Pump.fun tokens launched between January 2024 and March 2025 collapsed into rug pulls or pump-and-dumps, leaving only about 97,000 with $1,000 in liquidity. Chainalysis, using a stricter test across every chain, flagged 3.59% of 2024 tokens.

How do I avoid rug pulls on launchpads?

Check who holds the supply and whether wallets bought together in the first block. Confirm liquidity is locked or burned and that creators cannot mint new tokens. Be wary of anonymous teams, guaranteed returns and paid shilling. Size positions you can lose entirely, and never approve unlimited token spending. Nothing removes the risk. This is not financial advice.

What is a soft rug pull?

Solidus Labs uses soft rug pull for liquidity that is abruptly withdrawn so the token price collapses, without any hack or contract exploit. In its Raydium data, a sweep meant removing at least 90% of a pool's liquidity. The creator can keep the token's contract intact while leaving buyers unable to sell at anything close to the old price.

Is there a Pump.fun alternative that protects buyers?

MindPad is being built as one. Backers commit SOL or USDC into program-governed escrow before a token launches. If the target is hit by the deadline, the token goes live on Pump.fun, Raydium or Meteora. If not, nothing launches and backers get a full refund. MindPad is in development, so it cannot be used yet.

Sources

  1. 1.The 2025 Rug Pull Report: Rug Pulls and Pump-and-Dumps on Solana, Solidus Labs, 2025-05-07
  2. 2.Chainalysis report on crypto market manipulation, wash trading and pump-and-dump (2025), Chainalysis, 2025-01-29
  3. 3.The Memecoin Phenomenon, arXiv, 2025-12-12
  4. 4.Pump.fun Graduation Regime Windows: Survival Analysis of 832,941 Token Launches and the Social-Presence Effect, arXiv, 2026-07-01
  5. 5.Pump.fun token graduation rate jumps after BOOST changes launch incentives, The Block, 2026-07-29
  6. 6.Pump.fun hit with amended RICO class action alleging a $5.5B memecoin scheme, Cointelegraph, 2025-07-24
  7. 7.How to revoke smart contract allowances and token approvals, MetaMask Support

Nothing on this site is financial advice.

Share

Signed mandateMid price

· Trading

Agentic Trading and Market Making for Token Projects

Agentic trading for token projects explained: what a market maker does, how agents manage depth, spread and inventory within limits, and where fakery starts.

Get MIND updates in your inbox

One email when we ship. No spam, unsubscribe anytime.