How Many Solana Traders Get Rugged Every Year?

The measured result and the estimate

Solana's cheap transactions, fast execution and low barrier to token creation support an enormous market for new tokens. They also make it possible to launch, manipulate and abandon tokens at a scale that is difficult to describe with isolated case studies.

The best recent measurement comes from SolRugDetector: Investigating Rug Pulls on Solana. The researchers examined 100,063 tokens newly issued on Orca, Raydium and Meteora during January–June 2025. Their behaviour-guided pipeline identified 76,469 rug-pull tokens. A random manual audit of 382 samples estimated a 0.26% false-positive rate.

That is a measured result within one defined dataset. Everything after it in this article is either arithmetic or scenario modelling. We separate those categories because a detected token is not the same thing as a victim, a wallet is not the same thing as a person, and a failed token is not automatically proof of fraud.

A detected Solana rug roughly every 3.4 minutes

The first half of 2025 contained 181 calendar days. Dividing the study's 76,469 detected rug-pull tokens by that period gives:

The strongest defensible headline is therefore: during the period studied, a Solana token was identified as a rug pull roughly once every 3.4 minutes.

The annual figure is less certain. Doubling a six-month count assumes that issuance and detected-rug activity continue at the same rate. Solana activity changes with market conditions, platform incentives and token-launch volume. The 152,938 figure is useful as an order-of-magnitude estimate, not a forecast.

What the 76.4% study share does and does not mean

Dividing 76,469 by 100,063 produces 76.42%. That is the share classified as rug pulls inside the researchers' defined sample and methodology. It should not be rewritten as “76% of every Solana token is fraudulent” or “76% of Solana traders are rugged”.

The study focused on newly issued tokens across three decentralised exchanges during a particular six-month period. Its detection system looked for three representative behavioural patterns: freeze-authority abuse, liquidity withdrawal and pump-and-dump activity. The result tells us that rug-pull behaviour was extremely common within that measured issuance cohort. It does not establish creator intent in a court, cover every Solana venue or provide a permanent base rate for future launches.

This distinction matters because broad percentages travel faster than methodology. A precise-looking number becomes misleading when the denominator, time period or classification rule disappears.

Independent research supports the speed problem

A separate August 2026 paper, Catching the Rug: Early Prediction of Fraudulent Memecoins on Solana via Machine Learning, assembled a dataset of 6.4 million tokens over seven months. Its authors report that a vast majority of the memecoins exhibiting rug-pull characteristics did so within one hour of launch.

The researchers also tested whether classic machine-learning models could identify potential rugs using only the first five minutes of trading data. This does not prove that every rug can be predicted in five minutes. It does reinforce the operational problem for traders: warning characteristics can emerge before a conventional long-form review would finish.

MemeAssist's own fixed paper cohort points in the same direction without being treated as population evidence. Among 53 strategy-selected, mechanically paper-executed Pump.fun graduation positions recorded from July 25 to August 2, 2026, eight were later classified as rugs and the median modelled time to classification was about 89 minutes. That small, screened cohort cannot estimate how often all Solana meme coins rug. It shows that severe outcomes remained possible after automated checks and could develop quickly inside that workflow.

Why broader industry figures need careful wording

Solidus Labs reports that approximately 98.7% of Pump.fun tokens and 93% of examined Raydium liquidity pools exhibited characteristics associated with pump-and-dump schemes or rug pulls.

Those numbers describe a broader industry surveillance methodology. They should not be translated into “98.7% of Pump.fun creators are proven scammers”. Tokens can collapse, lose liquidity or display manipulation-associated characteristics without public evidence establishing criminal intent. For the event-frequency estimate in this article, we therefore use the narrower SolRugDetector classification and retain its exact sample definition.

How many wallets could be affected?

There is no comprehensive public dataset that connects every detected Solana rug to every outside wallet that lost money. To estimate scale without pretending that missing evidence exists, we use a scenario table.

The starting point is the simple annualisation of 152,938 detected rug events. We then vary one assumption: the average number of affected outside wallets per event.

These are victim-wallet events: an instance in which a wallet is affected by a detected rug under the scenario. They are not measured victims. The table is intentionally mechanical so readers can replace the assumption and see the result.

Many issued tokens attract almost nobody. Others attract hundreds or thousands of wallets. A 3–10 range is not presented as an observed average; it is a conservative modelling band chosen to illustrate what even low participation would imply when multiplied across the study's event count.

Why wallets cannot be converted cleanly into people

Public blockchains record addresses, not identities. One trader can operate several wallets. One wallet can buy several rug-pull tokens. The same person can be counted repeatedly across days, tokens and addresses. Exchange or bot infrastructure can also aggregate activity that does not map neatly to a single retail user.

This produces four categories that must remain separate:

• Failed token ≠ rug pull. A token can fail without deliberate manipulation.
• Rug-pull token ≠ victim. Some detected tokens may attract no outside buyer.
• Victim wallet ≠ unique person. People can use multiple wallets and be affected repeatedly.
• Trading loss ≠ amount stolen. Price decline, failed execution, fees and creator extraction are different economic quantities.

For those reasons, MemeAssist does not publish “X unique people get rugged every year” as a measured statistic. The defensible conclusion is narrower: even conservative participation assumptions produce hundreds of thousands of potential annual victim-wallet events. The number of individual people behind them remains unknown.

What this means for Solana meme coin traders

The studies support three practical conclusions.

Rug-pull behaviour is not an edge case

A six-month study identifying more than 76,000 rug-pull tokens shows that the problem operates at industrial scale inside the observed issuance cohort. Treating safety review as optional is difficult to reconcile with that evidence.

Speed changes the value of evidence

If rug characteristics often emerge in the first hour, old screenshots and launch-time checks can become stale before a trader acts. Timestamps, live liquidity, changing holder behaviour and executable sell routes matter alongside contract configuration.

No single green check proves safety

Revoked mint authority does not rule out hidden concentration. Burned liquidity does not prevent coordinated insiders from dumping supply. High volume can be wash traded. A visible sell route can disappear as liquidity changes. Evidence is more useful when these signals are assessed together and unknown checks remain unknown.

Methodology and limitations

This analysis combines published external research with transparent scenario modelling by MemeAssist. The base count is the 76,469 tokens classified by SolRugDetector among 100,063 tokens newly issued on Orca, Raydium and Meteora in the first half of 2025. Daily, hourly and per-minute figures are arithmetic derived from that six-month count using 181 days.

The 152,938 annual figure simply doubles the observed count. It assumes a constant rate and therefore does not capture changing market activity, token issuance, detection coverage or venue mix. The victim-wallet table multiplies that annualised event count by hypothetical affected-wallet averages. No row is an observed victim count.

The external studies use their own rug definitions and classifiers. Behaviour associated with manipulation does not by itself prove legal fraud or creator intent. The research sample includes newly issued Solana tokens and should not be described as a census of every meme coin, trader or loss. MemeAssist's 53-position paper cohort is cited only as selected timing context; it is not combined with the external data to calculate incidence.

The bottom line

Published research found 76,469 detected Solana rug-pull tokens in six months. Within that period, the count works out to roughly 422 per day, 17.6 per hour, or one every 3.4 minutes.

A simple annualisation produces about 153,000 detected rug events. Under an explicitly modelled 3–10 affected-wallet assumption, that becomes approximately 459,000–1.53 million victim-wallet events. The event range is an estimate, and the number of unique people cannot currently be measured reliably.

The important finding is not a precise victim total. It is that the measured event count is already large enough for conservative assumptions to imply a substantial human problem—and that separate research indicates warning characteristics can emerge within minutes.