Memecoin and speculative token trading

The research was done by AI agents that open web pages. Some sites refuse them; where that happened we say so. The small numbers point to the source list at the end. Where a figure is our own sum or guess, it is marked “our estimate” and the basis is given.

The claim

These are examples of the claim, copied from our collection of 8 tweets on this scheme¹. View counts are as collected. Dates are worked out from each tweet’s ID number. The page judges the scheme, not the people who tweeted.

“made $16,740 trading memecoins today

took some time off but I’m back”

@treysocial, 3 October 2026, 9,544 views¹. The tweet goes on; we quote its opening lines.

“My first ever 5 fig win from trading. Bought early and just diamond handed it for a few days, ignoring all the noise of paper hands.”

@EvilPanda, 30 September 2026, 6,034 views¹. The stored text is cut off after a few more lines. “Diamond handed” means held without selling; “paper hands” means people who sell early.

“Every tool you’ll need this memecoin bullrun:

Trading apps & terminals:
@Pumpfun - Launch & trade new tokens
@AxiomExchange - Advanced trading
@gmgnai - Trading & smart money
@fomo - Social trading & community
@tradewithPhoton - Fast trading & sniping
@trojanonsolana - trading”

@FabianoSolana, 30 September 2026, 25,628 views¹. The stored text ends here with a link we did not open; the post may continue.

The collection holds only 8 tweets, so the page rests mostly on how the method works and on public trading data. The other five, described without quoting: a post giving a coin’s contract address (the string of characters that identifies a coin); a post saying the author has bought a coin with a market value under $8,000 and will send its address to anyone who comments; a post from 5 September 2025 advertising a paid group at $30 a month; a post from 26 December 2024 promoting a podcast episode about crypto AI agents; and a post saying the author left a job after trading memecoins¹.

Three of the 8 state a result. None gives the money put in, the number of trades, the losing trades, the fees paid, a longer record, or a wallet address that anyone could check¹. We read the tweets as stored text and did not view their images, videos or threads, which may show more.

What the scheme is

A memecoin is a crypto token with no product or income behind it. Its price is whatever the next buyer will pay. The US securities regulator’s staff put it this way: meme coins “typically have limited or no use or functionality” and their value “is driven primarily by market demand and speculation”¹⁸.

Most of the coins in these tweets start on a launchpad, a website where anyone can create a token in a minute. The largest is Pump.fun, on the Solana blockchain. A new coin first trades on a “bonding curve”: a formula that raises the price as people buy. If enough money comes in, the coin “graduates” and trades on an ordinary crypto exchange run by software. “Market cap” (mcap) is the price times the number of coins; a coin with a market cap under $8,000 is at the very start of the curve.

The scheme is to buy such coins early and sell them for several times the price (“10x”, “30x”), many times over. Traders do this through:

• Trading terminals (Axiom, GMGN, Photon): websites that show new coins and let you buy in one click.
• Telegram bots (Trojan, BONKbot): the same thing inside a chat app.
• Sniping: using software to buy a coin in the first seconds after it appears.
• Wallet tracking and “KOLs” (key opinion leaders): watching what well-known traders buy and copying it.
• Call groups: paid or free chat groups where someone names coins to buy.

“AI agent tokens” are a variant: coins tied to a chatbot or automated account. They trade the same way. This is different from letting an AI program do the trading for you, which is covered briefly further down.

Every trade is recorded on a public ledger, so the results of millions of wallets can be counted. A wallet is an account, not a person. One person can run many wallets, and many wallets are run by software.

The arithmetic

A trading profit is a stake times a multiple, minus fees, minus the losing trades. The tweets give none of the four. The fees are the part that can be pinned down exactly.

Step 1: the fee on every trade

Pump.fun’s fee page, last updated on 8 October 2026, charges 1.25% of every buy and every sell while a coin is on the bonding curve: 0.95% goes to Pump.fun and 0.30% to whoever created the coin². The same 1.25% applies on its exchange, PumpSwap, while the coin is small (up to 420 SOL in market value)². SOL is Solana’s currency; it stood at $109.76 when we read the price¹⁷, so 420 SOL is about $46,000 (our sum). The fee steps down as a coin grows and reaches 0.30% only above 98,240 SOL, about $10.8 million (our sum)².

The terminals charge on top of that:

GMGN’s page still says Pump.fun charges 1%; Pump.fun’s own page now says 1.25%, so we use that²⁴. We could not read Photon’s or fomo’s fee pages.

Pump.fun’s own apps add nothing on top. Its fee page says: “none of the pump.fun frontend services (the pump.fun web app, pump.fun/advanced, and the pump.fun mobile app) charge any fees in addition to those above on Solana trades.”²

A “round trip” is one buy and one later sell. Our sums, from the pages above:

• Small Pump.fun coin through Axiom at the entry tier: 2 x (1.25% + 0.95%) = 4.4%
• Small Pump.fun coin through GMGN, Trojan or BONKbot: 2 x (1.25% + 1%) = 4.5%
• Small Pump.fun coin through Pump.fun’s own app: 2 x 1.25% = 2.5%
• One of the largest coins through Axiom at the entry tier: 2 x (0.30% + 0.95%) = 2.5%

These are our estimates. They assume the terminal’s fee is added to the launchpad’s fee, which is how GMGN’s page describes it⁴. So on a small coin, the average trade must gain more than about 4.4% through a terminal, or about 2.5% through Pump.fun’s own app, just to break even. A heavy trader on Axiom’s top tier, trading larger coins, pays less than a newcomer does²³.

Step 2: the costs that are not on the fee page

Slippage. Slippage tolerance is the share of the expected coins that the trader agrees in advance they might not receive, because the price moved before the order went through. GMGN’s own help page says: “the recommended slippage for automatic buy/sell and other pending order operations is 30%-35%; for new and popular tokens, it is recommended to be 50% or more”⁴. Those are recommended settings, not measured losses. We found no measurement of what traders actually lose this way.

Front-running. A “sandwich” is when a bot buys just before your order and sells just after it, taking the difference. A report by Helius, a Solana infrastructure company, covering 7 December 2024 to 5 January 2025, says of one program: “Over 30 days (Dec 7th to Jan 5th), the program executed 1.55 million sandwich transactions for a profit of 65,880 SOL ($13.43 million). The average profitability per attack was 0.0425 SOL ($8.67).”¹⁴ The same report says: “Among the top 20 sandwiched tokens, 16 were created on Pump Fun”¹⁴. A conference talk put the total taken from Solana users at “$370 million to $500 million” over about 16 months to May 2025³⁰. Both figures are old and the second is second-hand.

Network costs. These are small. Solana’s base fee is 5,000 lamports per signature, a fraction of a cent (a lamport is a billionth of a SOL; a signature is the approval attached to a transaction)¹⁵. For the optional tip that speeds up a trade, the median tip paid was about 0.000003 SOL and the 75th percentile about 0.00001 SOL when we looked, at 10:30 UTC on 10 October 2026; the feed is live and moves¹⁶. GMGN suggests 0.002 to 0.005 SOL per trade⁴, which is $0.22 to $0.55 (our sum¹⁷). On a $50 trade that adds 1% to 2% to a round trip; on a $2,000 trade it hardly matters (our estimate).

Step 3: what “$16,740 today” requires

Profit equals the stake times (the multiple minus 1). Working backwards from $16,740, before fees (our sums):

The tweet gives neither the stake nor the multiple¹.

Step 4: what $1,000 and $10,000 a month require

The amount of trading needed depends on how much of each dollar traded the trader keeps after fees and losing trades. A public leaderboard of well-known traders gives a top-end rate. We opened the 30-day pages of the six wallets at the top of its daily board on 10 October 2026³⁶³⁷. All profit figures there are “realised”, meaning they count only coins already sold.

Source:³⁷. Five of the six were ahead for the month. The middle of the six kept about 1% of volume (our sum). The first wallet’s page also showed $16,928 of unrealised gains, meaning coins still held and valued at the current price, so its month was close to flat³⁷. These are six wallets from the top of one day’s board of chosen traders, not an average. The page does not say whether “volume” counts buys only or buys and sells together.

Trading volume needed (our estimates: target divided by the rate):

Now the fees for a newcomer using a terminal. Take a trader who makes ten round trips a day for 30 days. The trade count and the position sizes are our assumptions; we found no published typical trade size. We use 4.45% per round trip, the midpoint of the 4.4% and 4.5% in Step 1, charged on the amount bought. Counted the way a leaderboard might count volume, buys plus sells, the trading is about twice the amount bought.

• $1,000 a month on $500 positions. 10 x $500 x 30 = $150,000 bought. Fees at 4.45% are about $6,700. To keep $1,000, the trades must gain about $7,700 before fees, an average of 5.1% per trade.
• $10,000 a month on $2,000 positions. 10 x $2,000 x 30 = $600,000 bought. Fees are about $26,700. To keep $10,000, the trades must gain about $36,700 before fees, an average of 6.1% per trade.

In both cases the platforms, coin creators and referrers together collect more than the trader keeps (our estimate). Through Pump.fun’s own app, at 2.5% per round trip, the fees would be about $3,750 and $15,000 (our sums). A trader who makes a few large bets and holds them for days pays far less; one quoted tweet describes that style¹.

Step 5: how many wallets get there

CoinGecko, a crypto data company, counted every wallet that sold a coin on Pump.fun or PumpSwap each month from April 2024 to April 2026⁶. April 2026 was the best month for traders in that record:

Source:⁶. So in the best month on record, about 1 wallet in 19 showed a realised gain over $1,000 (our sum). CoinGecko’s words: “At the very top, 169K wallets (5.4%) recorded profits exceeding $1,000. This is a meaningful cohort in absolute terms, but still a small fraction of total active wallets.”⁶

These are counts of money out minus money in for each wallet, across all its coins, and count only coins that were sold. CoinGecko lists the limits itself: “A wallet that bought Token A and sold an unrelated Token B in the same month would appear profitable even if Token A is still held.”⁶ It adds that dollar prices for thinly traded coins “can be missing, stale, or distorted”⁶. So 5.37% is better read as a ceiling on the share of wallets that truly made $1,000 that month (our reading).

The table has no $10,000 line, so the share of wallets making $10,000 in a month is not published⁶. The $10,000 figures we found cover other periods and were all read second-hand:

• August 2024: “only 0.8%” of Pump.fun traders were up more than $10,000, as reported by The Defiant⁵³.
• January 2025, whole life of each wallet: 55,012 of “over 13.4 million” Pump.fun wallets (0.4%) had taken $10,000 or more in profit, according to a count reported by Decrypt⁷. Pump.fun’s co-founder disputed that count; his reply is in the next section.
• August 2026, 90 days: a chart of wallets using one trading app, fomo, showed 25 wallets with realised gains over $10,000, as reported by Cryptopolitan⁶¹. The number of wallets on that chart was not given.

Step 6: the odds on a single new coin

Most coins never leave the launchpad. A study of all 15.2 million Pump.fun coins found: “In our sample, 1.02% of coins graduate.”⁴¹ An earlier paper found that “fewer than 2% of tokens successfully transitioned to major decentralized exchanges”¹¹. The rate moves: The Block reported that it “reached 6.7% last Friday, roughly 8x higher than the average throughout June” in late July 2026, after Pump.fun changed how launches work¹⁰. We did not check the rate for October 2026.

So a buyer of a coin with a market value under $8,000 is picking from a pool where, on these counts, between 93 and 99 of every 100 coins do not graduate (our sum from¹⁰¹¹⁴¹). A coin does not have to graduate for an early buyer to sell at a profit, and a graduated coin can still lose its buyers money. The graduation rate shows how few coins keep attracting money. It is not the share of buyers who profit, which we could not find.

What people who tried it report

The best evidence here is counts of wallets, not stories. None of the counts removes wallets run by software.

Most months, most wallets lost. CoinGecko’s summary: “From April 2024 through late 2025, the majority of active traders on Pump.fun walked away from each month in the red, with the share of profitable wallets rarely exceeding 50% and bottoming out at just 30.1% in June 2025.”⁶

In early 2026 that reversed. The profitable share was 50.08% in January 2026, 56.83% in February, 70.00% in March and 73.28% in April⁶. CoinGecko says it cannot explain this for certain and suggests that losing traders left: monthly wallets fell from 5,262,050 in May 2025 to 1,795,474 in December 2025⁶. Roughly two thirds of the monthly participants were gone within seven months (our sum).

These figures flatter the traders, by CoinGecko’s own account. It counts only coins that were sold: “This study only accounts for Realized PnL; this means that it excludes bagholders who never sold their tokens even if it crashes to zero. We understand that this understates losses”⁶. (“PnL” is profit and loss; “realized” means only coins already sold are counted; a “bagholder” is someone left holding a coin that fell.) Its second limit points the same way: “Buy/sell flows across all tokens are netted at the wallet level”, so a wallet that sold one coin and still holds another it paid for can show as profitable⁶.

Other counts for 2026 are much less favourable, and we could not reconcile them. Cryptopolitan, a crypto news site, reported on 19 August 2026 on a dashboard covering 90 days of Solana memecoin trading: “a total of 304,161 traders were active on Solana DEX. Of those traders, only 19,003 were in the green, but the median trader lost $120.”⁶¹ (A DEX is an exchange run by software.) That is 6.25% in profit (our sum). The article adds: “Of all traders in the green, 88% made under $100.”⁶¹ It says the count uses realised results. Separately, it reports a chart for the fomo app showing 25 wallets with realised gains over $10,000; that figure may describe that app’s users only⁶¹. Other sites repeated these figures⁵¹. A March 2026 report said over half of Pump.fun wallets lost that month⁵². These use different time windows and methods. The August article links to a query on Dune, a public data site, which we did not open, and 304,161 wallets is about a tenth of CoinGecko’s monthly count, so the two do not cover the same group. We prefer CoinGecko’s figures, which we read at the source with the full table, and treat the others as unconfirmed.

Older counts agree on the shape.

• August 2024: 3% of Pump.fun traders had made more than $1,000, 0.8% more than $10,000, and over 60% had lost money, as reported by The Defiant⁵³.
• January 2025 (whole life of each wallet): 0.4% of wallets had made $10,000 or more; 294 wallets had made $1 million or more⁷. Decrypt notes that the count “excludes the purchasing of Pump.fun tokens after they graduated from the platform”⁷. Pump.fun’s co-founder replied that most gains are made after a coin graduates, and that around 30% of wallets had placed only a single order, a sell order, and so were unlikely to be run by people. In his words: “It’s likely that the true number of profitable and extremely profitable wallets is an order of magnitude LARGER than exhibited.” The article reports no figures behind the “order of magnitude” estimate⁷. Ten times 0.4% would be 4% (our sum).
• Six months to June 2025: “of 4.257 million addresses that traded more than 10 Pump.fun tokens in the past six months, around 2.4 million addresses (56.6%) posted cumulative losses between $0 and $1,000”; about 5,000 addresses made more than $100,000 and 311 more than $1 million, as reported by BeInCrypto⁸. The same article carries a second set of numbers that does not match the first, so we use it with care.

What happened to the coins. Solidus Labs, a company that sells market-monitoring software, studied Pump.fun coins from January 2024 to March 2025 and wrote: “Over 7 million tokens deployed with at least five trades, but only 97,000 tokens maintain liquidity above $1,000.”¹² (“Liquidity” is the money available to buy the coin back from you.) That is about 1.4% (our sum). Solidus has a commercial interest in the finding, and its page gives the share that collapsed as 98.6% in one place and 98.7% in another¹². Press coverage says Pump.fun rejected the report; we saw that only in a search summary and did not open Pump.fun’s reply. Chainalysis, another data company, used a much stricter test across all blockchains and flagged 3.59% of the tokens launched in 2024 (74,037 of 2,063,519); it also found that only 1.7% of that year’s tokens were still actively traded²⁸. The two firms measure different things. They agree that most coins stop trading.

Coins that reach big exchanges often fall too: Coin Metrics, as reported by Cointribune on 9 October 2026, found 81% of 150 exchange-listed memecoins down at least 90% from their peak⁵⁰.

Speed. Galaxy Research, as reported by Cointelegraph, found that the typical Solana memecoin was held for about 100 seconds, down from about 300 a year earlier¹³. A person clicking by hand is trading against software.

Signals can be manufactured. A September 2026 study from Carnegie Mellon and EPFL researchers covering all Pump.fun coins states: “We identify five classes of manipulation strategies”, including trading with yourself to make a coin look busy, and describes paid tools that do this for people with no technical skill⁴¹. The volume and “trending” lists that terminals show are what gets worked on.

Following promoted coins. CoinWire tracked 1,567 memecoins promoted by 377 X accounts with 10,000 or more followers and reported that 86% of the promoted coins were down 90% after three months, and that “76% of Twitter influencers have promoted meme coins that are now dead.”⁴⁷ The study is from November 2024 and says nothing about the accounts in our collection.

One practitioner’s account. A full-time trader interviewed by the journalist David Z. Morris said he got into crypto around 2020 to 2021 and “as a beginner got burned for the first two or three years”, before he began trading memecoins in late 2023⁴⁹. That is his own statement, and it describes his years in crypto before memecoins.

AI agent tokens. The only evidence we found is a price fall: Cointelegraph reported in February 2025 that leading AI agent tokens were down as much as 90% from their highs⁴⁵. We found no wallet-by-wallet results for these tokens and no checkable record of sustained profit from trading them.

Letting an AI agent trade for you. This is a different thing, and the nearest evidence we found. A September 2026 study followed AI programs that trade on their own: 3,505 user-funded accounts trading with real money in memecoin markets on the Base blockchain for 21 days in February and March 2026, and a second group trading other instruments. Its summary says: “neither fleet shows a directional edge.”⁴⁴ A “fleet” is one group of these programs. In plain words, neither group did better than chance at picking which way prices would go. We read the summary only.

What the rules allow now

It is legal to trade in the US, without the usual protections. A joint interpretation by the US Securities and Exchange Commission (SEC) and the Commodity Futures Trading Commission, effective 23 March 2026, classes meme coins as “digital collectibles” that are not securities: “Meme coins typically are acquired for artistic, entertainment, social, and cultural purposes, and their value is driven by supply and demand, rather than any essential managerial efforts of others.”¹⁹ The cost of that status was spelled out by SEC staff in February 2025: “neither meme coin purchasers nor holders are protected by the federal securities laws.”¹⁸ The 2026 release replaces earlier staff statements, so the 2025 line is background.

The UK is different. The Financial Conduct Authority (FCA) has had Pump.fun on its warning list since 3 December 2024: “This firm may be providing or promoting financial services or products without our permission.”²² That is the FCA’s standard notice for a firm it has not authorised; it is not a finding about how Pump.fun treats customers. Pump.fun’s own terms, updated 8 October 2026, list the United Kingdom first among the places where its service may not be used²¹. The FCA’s general advice on crypto is that “you should be prepared to lose all the money you have invested”²³.

The platform’s terms leave little recourse. Pump.fun’s terms require users to be 18 or over, forbid users from “market manipulation”, cap Pump.fun’s liability at the fees on the trade or $100, whichever is larger, set a one-year limit on claims, and send disputes to individual arbitration in the British Virgin Islands²¹. In capitals: “YOU IRREVOCABLY WAIVE YOUR RIGHT TO PARTICIPATE IN A CLASS ACTION”²¹. Two terms cut the other way. A user may opt out of arbitration by email within 30 days of first accepting the terms; disputes then go to the courts of the British Virgin Islands and the bar on class actions stays²¹. And the terms say that consumer rights which the law does not allow to be excluded, naming the UK, the European Economic Area and Australia, “are not affected by these Terms”; the one-year limit applies “To the extent permitted by Applicable Law”²¹. We could not open the terms of Axiom or Photon.

A lawsuit against Pump.fun is pending. In a class action in federal court in New York (case 25-cv-880), the judge ruled on 31 August 2026 that the racketeering claims of two of the three named plaintiffs may go forward against Baton Corporation, which operates Pump.fun, and three people the order describes as Pump.fun’s co-founders and officers²⁰. The claims are brought under RICO, the US racketeering law. The order’s words are that the two plaintiffs “have adequately pleaded substantive RICO and RICO conspiracy claims” against those four defendants²⁰. The same ruling dismissed the third plaintiff’s racketeering claims, the securities claims, and every claim against Solana Labs, the Solana Foundation and the other defendants the order groups with them²⁰. This is a ruling that the case may proceed, not a finding. The allegations are unproven and the case is pending. The plaintiffs allege that insiders bought before the public and that “retail traders collectively lost between $4 billion and $5.5 billion trading Pump.fun tokens”; that is their estimate²⁰. Pump.fun asked the court to dismiss the case and succeeded on three of five counts²⁰. The judge also noted that the plaintiffs allege about sixty percent of wallets ended down: “That necessarily means that approximately forty percent did not.”²⁰ We did not find Pump.fun’s public statement on the ruling. The case concerns only the companies and people named in it.

Paying for artificial trading volume has been charged. On 9 October 2024 the SEC announced a case titled “SEC Charges Three So-Called Market Makers and Nine Individuals in Crackdown on Manipulation of Crypto Assets Offered and Sold as Securities”²⁹. We did not check the outcomes. The charges concern only the firms and people named, and tokens the SEC said were sold as securities, not Pump.fun memecoins.

Promoting is riskier than trading. In the UK, FCA guidance says influencers who promote financial products without approval “may be committing a criminal offence”³³. A law firm’s summary gives the penalty as “a fine and/or up to 2 years’ imprisonment” and reports seven influencers sentenced in February 2026, in a case about currency trading, not crypto³⁴. We found no prosecution of an influencer over a memecoin. EU regulators warned in October 2025 about “aggressive promotion on social media by finfluencers”, without naming memecoins³⁵.

Tax. In the US, every swap of one token for another must be reported: “If you have digital asset transactions, you must report them whether or not they result in a taxable gain or loss.”³¹ In the UK, swapping one token for another counts as a disposal for Capital Gains Tax, and “You must keep your own records of your transactions.”³² Hundreds of trades a month means hundreds of entries.

X’s own rules. We could not read them. X’s policy pages refused us both directly and through a reader, so this page says nothing about what X allows.

Who makes money from it

The launchpad. Traders have paid about $1.27 billion in fees on Pump.fun’s launch product since March 2024, and about $942 million on PumpSwap since February 2025, according to DefiLlama, a site that adds up fees from the public ledger⁹. DefiLlama is a third party, not Pump.fun’s accounts.

The terminals and bots. Fees on DefiLlama, read on 10 October 2026⁹. “Fees” is what traders paid. “Kept by the platform” is what DefiLlama counts as the platform’s own revenue after payments to coin creators, to liquidity suppliers (people who supply coins to the exchange), to referrers and back to traders as cashback.

The all-time and “kept” columns cover every blockchain a tool works on. GMGN and Axiom also serve other blockchains: only $7.0 million of GMGN’s $30.5 million came from Solana⁹.

The seven rows with meaningful 30-day activity add up to about $244 million in a month on all blockchains, and about $217 million on Solana alone (our sums)⁹. That money went to platforms, coin creators, liquidity suppliers and referrers together. The platforms kept about $100 million of the $244 million (our sum)⁹. On PumpSwap the platform kept about 13% of fees in those 30 days, so about 87% went elsewhere (our sum); Pump.fun’s fee table sends that share to coin creators and liquidity suppliers²⁹.

Trading between participants creates no new money, so traders as a group are behind by about the fees, less whatever comes back to them as cashback. Divide the $217 million of Solana fees by about 3.1 million monthly wallets and the result is roughly $69 per wallet per month. This is a rough estimate of ours: the top number is fees at all seven tools, the bottom number counts only wallets that sold on Pump.fun or PumpSwap, and it comes from a different month (April 2026)⁶. Some PumpSwap trading is not memecoins⁹.

The tools themselves rise and fade: Photon and BullX took hundreds of millions and now take very little⁹. Galaxy Research, as reported by Cointelegraph in October 2025, said Axiom had taken “over $200 million in fees with fewer than 10 employees”¹³.

Coin creators. The creator receives 0.30% of every trade while the coin is on the bonding curve, and 0.95% once it has graduated and is worth 420 to 1,470 SOL, falling in steps to 0.05% for the largest coins². Pump.fun’s page says: “On every trade the creator receives some portion of the total fees.”² Decrypt documented one wallet that “has profited more than $840,000 by launching 17,794 tokens over the last three months”, selling its own launches within minutes⁴⁰. Pine Analytics traced wallets that were funded by a coin’s creator and bought in the first moment of trading: over one month, more than 4,600 such wallets took over 15,000 SOL, and “87% of snipes were profitable”⁴². This is the other side of the trade the tweets promote, and not a method this page suggests.

People who refer traders. Each tool pays whoever brings in a trader:

• Axiom: “You earn 30% of Axiom’s net fee on every trade made by your Level 1 referrals”, plus 3% and 2% on two further levels²⁴. DefiLlama shows Axiom keeping $464.4 million of $768.5 million in fees, so about $304 million went out as referral payments and cashback (our subtraction)⁹.
• GMGN: “Earn 10% to 30% in commissions”, advertised with “easily earn over 40 Sol ($8000+) monthly!”²⁵ To earn $8,000 a month at 30% of a 1% fee, the people referred would need to trade about $2.7 million a month (our estimate). GMGN’s page prices 40 SOL at $8,000; at today’s price 40 SOL is about $4,400, which would need about $1.5 million of referred trading a month (our estimate)¹⁷.
• Trojan: 27.5% to 47.5% of referred traders’ fees, across five levels of referrals²⁶.

A referrer is paid on volume, whether the people referred win or lose. Referral pay gives anyone who lists trading tools a possible income from the list. The tool-list tweet in our collection contains @-mentions only, with no referral links in the stored text; we did not check the author’s profile¹.

What the accounts in our collection point to. Our collection records a live stream in the X bio of the author of the “$16,740” tweet¹. A page of links at beacons.ai/trey links to that X account and to a Twitch channel, which is what ties them together⁵⁶. The Twitch bio says in part: “since then I’ve made 1m+ profit trading memecoins. Please don’t copy trade me. None of what I say is financial advice.”⁵⁶ The page of links includes one to a trading terminal, described as “The trading platform I use for fast fills and most cashback”, along with a free Discord group and a Telegram channel⁵⁶. The link has the form of a referral link; we did not see its terms. No paid course was seen on these pages⁵⁶. The profit figures are his own statements.

The paid group advertised in one tweet, from September 2025, has a storefront on Whop, a marketplace for paid communities, under the name Jellycubes. The storefront links to the X account the tweet mentions, which is what ties the two together¹⁵⁷. Its description begins: “I have been trading crypto full-time since 2020 and creating content on YouTube for the past 4 years”⁵⁷. The storefront still exists, but on 10 October 2026 its products tab said: “Unfortunately, there are no products currently available to the public.”⁵⁷ It showed three reviews. No price was visible; the $30 a month is as tweeted in September 2025 and is not confirmed on the page¹⁵⁷.

A second Whop listing shows the kind of thing sold around memecoins; it is not connected to any tweet in our collection. It is a $99-a-month add-on for buyers of the seller’s course, about launching coins and not about trading them, headed “Work Directly With Me Until You Make Your First $5,000+”⁵⁸. When we read it, it offered a waiting list and no buy button, and said “Cancel anytime”⁵⁸.

Paid promotion. An investigator known as ZachXBT published a price sheet in September 2025 covering crypto accounts approached for paid posts. As reported by ForkLog, he wrote, of more than 200 accounts that were contacted: “From 160+ accounts who accepted the deal I only saw <5 accounts actually disclose the promotional posts as an advertisement.”⁴⁸ This is one investigator’s count, read second-hand. It does not concern the accounts in our collection.

The upside

Real winners exist, and here they can be checked, because the trades are public.

The largest documented results.

• One Pump.fun wallet had an accumulated profit of $39,826,672 by March 2025, according to a public dashboard reported by The Block³⁸. Nobody knows who runs it or how. Second-hand; on-chain data (read from the public ledger).
• 294 Pump.fun wallets had taken more than $1 million each by January 2025, and 55,012 more than $10,000⁷. Second-hand; on-chain data. Five-figure winners number in the tens of thousands, and they are about 1 wallet in 250 (our sum). That count leaves out purchases made after a coin graduated and includes wallets that only ever placed one sell order, so the true share is probably higher⁷.
• In the six months to June 2025, about 5,000 addresses made more than $100,000, roughly 0.1% of the 4.257 million counted (our sum)⁸. The same count found nearly 1,700 addresses down more than $100,000⁸.
• One wallet bought $16.20 of a coin while it was still on Pump.fun and sold over three days for about $3 million, as reported by Decrypt with the purchase linked³⁹. This was a single held position, which Decrypt itself describes as luck.
• Nansen, which sells a wallet-tracking product, lists ten memecoin wallets with results such as $9.65 million over 1,681 trades and $3 million turned into $35 million; most came from being early in a handful of well-known coins⁴⁶. It is a list of survivors, chosen by a company selling the tool to follow them.
• On single famous coins the gains were as concentrated. For one coin launched in January 2025, NBC News reported Chainalysis data showing 58 wallets making more than $10 million each while about 764,000 wallets lost money⁵⁴. For another, The Block reported Nansen data that 86% of traders with more than $1,000 at stake lost⁵⁵. Both are second-hand.

It is still happening. Kolscan is a public leaderboard of well-known traders’ wallets with each trade linked to the ledger. It has belonged to Pump.fun since July 2025: “Pump.fun has acquired Kolscan.”⁶⁰ Its page carries an “Apply as KOL” button, so traders ask to be listed³⁶. It is a board of chosen, well-known traders, run by the platform that earns the fees. On 10 October 2026, the top 50 wallets each took between $1,232 and $17,832 for the day; the middle of the 50 was $3,128 (our calculation)³⁶. So the tweet’s “$16,740 today” is about the size of the single best daily result on that board. It is possible, and it is not a typical day even for traders famous enough to be tracked.

What a strong month looks like. The two strongest of the six top daily wallets we opened (Step 4) showed realised gains of $312,613 and $133,791 in 30 days. The first won 33.9% of its trades with an average holding time of 2 hours; the second won 48.2% with an average of 3 hours³⁷. A seventh wallet we opened, 11th on the day, showed $261,862 on $3.574 million with 52.7% of trades won³⁷. We read these figures on the leaderboard and did not re-add the trades ourselves.

A big day is not always a good month. Five of the six top daily wallets were ahead over 30 days, by realised amounts from $2,603 to $312,613³⁷. The wallet at the top of the daily board, up $17,832 that day, showed a realised 30-day result of minus $19,947 on $1.17 million of trading, with $16,928 of unrealised gains still held, so its month was close to flat³⁷. The second-placed wallet, up $12,108 for the day, showed $10,222 for 30 days with 15.3% of trades winning, so without that one day its month was a small loss (our sum)³⁷.

The top tenth and top hundredth. The data does not give these exactly. In the best month on record the top 5.4% of wallets made more than $1,000⁶. Over a wallet’s whole life to January 2025, the top 0.4% had made more than $10,000 and the top 0.048% more than $100,000, on a count that leaves out purchases after graduation⁷.

A realistic good result. In the best month on record, two thirds of all wallets made between $1 and $500, and about 1 in 20 cleared $1,000⁶. In most earlier months, most wallets lost, and usually less than $1,000⁶⁸. So for a careful newcomer, a small loss is the usual result, and a good result is a gain under $500, often far under: about 8% of wallets showed more than $500 in the best month⁶, and in the 90-day count from August 2026, 88% of the wallets in profit made under $100⁶¹. This is our estimate from those sources. Nothing we found measures how long it takes to become profitable, or shows that practice reliably improves results.

What it takes to compete

The public records of winning wallets show what they have that most who try do not.

Capital and volume. The six top daily wallets we opened traded between $1.2 million and $4.2 million in a month³⁷. At the two strongest wallets’ rates, a $16,740 day needs roughly $223,000 to $418,500 of trading that day (our estimate, Step 4). Not every big day comes from volume: one quoted tweet describes a single position held for days, and the $16.20 purchase above was one trade¹³⁹.

A stomach for many losing trades. Among the seven wallets we opened, the share of trades won ranged from 15.3% to 52.7%; the one with the largest gain won 33.9%³⁷. A published 15-day test of an automated memecoin trader, using pretend money, won 40.5% of 190 trades and ended well ahead, yet “Removing the top three trades (1.6 percent of sample) flips cumulative return unprofitable.”⁴³ An older analysis of hand-picked winning wallets found profit just as concentrated even where most trades won: its best ten wallets won 63.55% of their trades on average, yet the three largest wins were about 91% of total profit⁵⁹. A few large wins carry the result, so a newcomer cannot tell skill from luck in their first hundred trades.

Speed and tools. With the typical coin held for about 100 seconds¹³ and front-running bots at work¹⁴, the competition is software. Faster connections, paid tracking tools and paid groups exist; we did not price them beyond the two listings above.

Information. Over half of launches in one study were bought in the very first moment; in about 1.75% of launches the study could prove the buyer had been funded by the creator⁴². A buyer who arrives through a post on X is later than those wallets by definition. Reading the ledger well enough to spot this is a skill the winners need.

Time. One full-time trader said he had about three years in crypto, the first two or three of them losing, before he started on memecoins⁴⁹. Pump.fun’s co-founder, asked whether people do this full time, said: “Are there people trading memecoins full-time? Yes. Many of them also create coins”⁴⁰.

An audience. Accounts with followers have income the trade itself does not provide: 30% of referred traders’ fees at Axiom²⁴, up to 47.5% at Trojan²⁶. We found no measurement of how much of well-known traders’ profit depends on followers copying their purchases.

What it costs. For a newcomer on a small coin, about 4.4% of every round trip through a terminal, or about 2.5% through Pump.fun’s own app (our estimates, Step 1). Heavy traders on larger coins pay less. Add slippage and tips, and the stake, all of which can be lost.

How to tell early which side you are on.

• Keep your own record of every trade after fees. Judge the total over a few hundred trades, not the best day.
• Take out your three best trades. If what remains is a steady loss, the result so far rests on luck⁴³.
• Add up the fees you have paid. If they are larger than your profit, the platforms and coin creators are earning more from your trading than you are (Step 4).
• If your trades come from a call group or a post, check what the caller is paid for: a subscription, a referral share, or a position in the coin. Those are paid whether you win or not²⁴²⁵²⁶.
• A profit claim with no wallet address cannot be checked. A claim with one can, by anyone, for free.

What we did not verify

• The claims in the tweets. No tweet gives a wallet address, so no profit figure could be checked. We read the tweets from our stored file and did not open them on X or view their images, videos or threads¹. Apart from the pages at⁵⁶ and⁵⁷, we did not open the authors’ profiles, and did not confirm that any link pays its poster.
• Typical trade size and trade count. No published figure was found, so Step 4 uses our own assumptions.
• The share of wallets making $10,000 a month, and any wallet-level results after April 2026⁶. The Block reported in June 2026 that Pump.fun activity had fallen sharply; we saw only the headline.
• The conflict between counts for 2026. CoinGecko’s 73% in April⁶ and the 6.25% over 90 days reported in August⁶¹ are not reconciled. The Dune query and the fomo-app chart behind the August figures were not opened, and we do not know how many wallets the fomo chart covers.
• How many wallets truly profit. CoinGecko’s counts are money out minus money in per wallet, on sold coins only⁶. Nobody we found has counted results coin by coin, or the share of early buyers of a new coin who sell at a profit.
• Figures read second-hand. The wallet counts in⁷,⁸,³⁸,⁵¹,⁵²,⁵³ and⁶¹ come from news reports of public dashboards we did not open. Galaxy Research¹³, Coin Metrics⁵⁰, the single-coin figures⁵⁴⁵⁵ and the price sheet⁴⁸ were also read through news reports.
• Exact wording. Several pages were read through a tool that summarises, including¹²,¹⁸,²² and²⁹. Their quotes should be matched against the pages. The Twitch bio in⁵⁶ was read by one of our agents from Twitch’s data service and not re-read in the final check.
• Pump.fun’s replies. Its response to the Solidus report was seen only in a search summary. Its statement on the August 2026 ruling was not found, and we did not check the case since then.
• Fees. Photon’s and fomo’s fee pages could not be read. One of our agents could not open Axiom’s fee page; two others read it through a reader, and it says it was last updated a year ago³. Our readings of Pump.fun’s fee table differ on the lower tiers for larger coins; all read 1.25% for small coins². Whether fees add up exactly as in Step 1 on every route was not tested.
• Front-running and slippage today. No 2026 figures were found¹⁴³⁰.
• The Kolscan figures. Not re-added from the ledger. The 30-day window is inferred from the page address. All profit figures are realised only. We do not know how Pump.fun decides which applicants to list, which tools or fee tiers the listed wallets use, or whether “volume” counts both buys and sells³⁶³⁷. Seven wallets on one day are not a sample.
• Risks left out. Reports of security incidents at trading tools were not checked and are not used.
• The paid group. Its current price, and whether it still takes members, could not be seen⁵⁷.
• Law. X’s rules could not be read. Tax outside the US and UK, countries that restrict crypto trading, the US market-structure bill after July 2026 and rules for AI agent tokens on other blockchains were not checked.
• Other blockchains. Almost all the data is about Solana and Pump.fun. One tweet in our collection gives an address on another blockchain¹; we have no outcome data for that market.
• Time to profit. No source measures how long it takes, how many give up, or whether the same wallets stay profitable from one period to the next.

Sources

1. Does It Pay collection of 8 tweets on this scheme, collected 2026-10-06. Tweet dates worked out from tweet IDs. Read from the project’s stored file on 2026-10-10.
2. Pump.fun, “Fees”. https://pump.fun/docs/fees . Page says last updated 8 October 2026. Read at source on 2026-10-10.
3. Axiom, “Axiom fees”. https://docs.axiom.trade/getting-started/fees/axiom-fees . Page says last updated one year ago. Read at source on 2026-10-10 through r.jina.ai.
4. GMGN, “Fees and settings”. https://docs.gmgn.ai/index/gmgn-fees-settings.md . Undated. Read at source on 2026-10-10.
5. Trojan, “FAQs”. https://docs.trojan.com/faqs.md . Undated. Read at source on 2026-10-10.
6. CoinGecko Research, “Pump.fun traders are making a comeback”. https://www.coingecko.com/research/publications/pump-fun-traders-are-making-a-comeback . 7 May 2026. Read at source on 2026-10-10 through r.jina.ai (direct request refused).
7. Decrypt, “Just 0.4% of Pump.fun Traders Have Made More Than $10,000 on Solana Meme Coins”. https://decrypt.co/300403/pump-fun-traders-millionaires . 10 January 2025. Secondary (reports a Dune query we did not open).
8. BeInCrypto, on six months of Pump.fun trading data. https://beincrypto.com/pump-fun-trading-data-majority-lose-money/ . 5 June 2025. Secondary; read through r.jina.ai.
9. DefiLlama fees and revenue data, one request per tool (pump.fun, pumpswap, axiom, gmgn, trojan, photon, bonkbot, bullx). https://api.llama.fi/summary/fees/pump.fun?dataType=dailyFees and the same address ending dataType=dailyRevenue . Data to 9 October 2026. Read at source on 2026-10-10. Sums and subtractions are our estimates.
10. The Block, on Pump.fun’s graduation rate. https://www.theblock.co/post/409815/pump-fun-token-graduation-rate-jumps-boost-changes-launch-incentives . 29 July 2026. Read at source on 2026-10-10 through r.jina.ai.
11. Davide Mancino, preprint on Solana memecoins. https://arxiv.org/abs/2512.11850 . December 2025. Read at source on 2026-10-10.
12. Solidus Labs, “The 2025 Rug Pull Report”. https://www.soliduslabs.com/reports/solana-rug-pulls-pump-dumps-crypto-compliance . May 2025. Read at source on 2026-10-10.
13. Cointelegraph, on a Galaxy Research report about memecoins. https://cointelegraph.com/news/memecoins-attract-users-platforms-reap-profits-report . 1 October 2025. Secondary (the report itself was not found).
14. Helius, “Solana MEV report”. https://www.helius.dev/blog/solana-mev-report . 15 January 2025; data from 7 December 2024 to 5 January 2025. Read at source on 2026-10-10 through r.jina.ai (an earlier reading was through a summarising tool).
15. Solana documentation, “Fees”. https://solana.com/docs/core/fees . Undated. Read at source on 2026-10-10.
16. Jito, live tip data and documentation. https://bundles.jito.wtf/api/v1/bundles/tip_floor and https://docs.jito.wtf/lowlatencytxnsend/ . 10 October 2026. Read at source on 2026-10-10.
17. CoinGecko price data for SOL ($109.76). https://www.coingecko.com/en/coins/solana . 10 October 2026. Read at source on 2026-10-10 through CoinGecko’s public data service.
18. US Securities and Exchange Commission, “Staff Statement on Meme Coins”. https://www.sec.gov/newsroom/speeches-statements/staff-statement-meme-coins . 27 February 2025. Read at source on 2026-10-10.
19. US Securities and Exchange Commission and Commodity Futures Trading Commission, joint interpretation, Release No. 33-11412. https://www.sec.gov/files/rules/interp/2026/33-11412.pdf . 17 March 2026, effective 23 March 2026. Read at source on 2026-10-10.
20. US District Court, Southern District of New York, Aguilar et al. v. Baton Corporation Ltd, case 25-cv-880 (CM), decision and order, document 184. https://storage.courtlistener.com/recap/gov.uscourts.nysd.635992/gov.uscourts.nysd.635992.184.0.pdf . 31 August 2026. Read at source on 2026-10-10.
21. Pump.fun, “Terms and conditions”. https://pump.fun/docs/terms-and-conditions . Page says last updated 8 October 2026. Read at source on 2026-10-10 through a summarising tool and again through r.jina.ai.
22. Financial Conduct Authority, warning notice “Pump.fun”. https://www.fca.org.uk/news/warnings/pumpfun . 3 December 2024. Read at source on 2026-10-10.
23. Financial Conduct Authority, “Crypto: the basics”. https://www.fca.org.uk/investsmart/crypto-basics . Last updated 29 January 2026. Read at source on 2026-10-10.
24. Axiom, “Referral program”. https://docs.axiom.trade/getting-started/referral-program . Undated. Read at source on 2026-10-10 through r.jina.ai.
25. GMGN, referral programme page. https://docs.gmgn.ai/index/cooperation-referral-refer-friends-to-earn-rebate-30-rebates-easily-earn-over-usd8000-monthly.md . Undated. Read at source on 2026-10-10.
26. Trojan, “Referral system”. https://docs.trojan.com/trojan-arena/referral-system.md . Undated. Read at source on 2026-10-10.
27. BONKbot, “Fee structure”. https://docs.bonkbot.io/fee-structure.md . Undated. Read at source on 2026-10-10.
28. Chainalysis, report on crypto market manipulation in 2024. https://www.chainalysis.com/blog/crypto-market-manipulation-wash-trading-pump-and-dump-2025/ . 29 January 2025. Read at source on 2026-10-10 through r.jina.ai.
29. US Securities and Exchange Commission, press release 2024-166. https://www.sec.gov/newsroom/press-releases/2024-166 . 9 October 2024. Read at source on 2026-10-10.
30. Solana Compass, notes on a talk about front-running on Solana at the Accelerate 2025 conference. https://solanacompass.com/learn/accelerate-25/scale-or-die-at-accelerate-2025-the-state-of-solana-mev . 20 May 2025. Secondary (machine-written notes from a video we did not watch).
31. US Internal Revenue Service, “Digital assets”. https://www.irs.gov/filing/digital-assets . Last reviewed 2 September 2026. Read at source on 2026-10-10.
32. GOV.UK, “Check if you need to pay tax when you sell cryptoassets”. https://www.gov.uk/guidance/check-if-you-need-to-pay-tax-when-you-sell-cryptoassets . Page date not captured. Read at source on 2026-10-10 through r.jina.ai.
33. Financial Conduct Authority, finalised guidance FG24/1 on financial promotions on social media. https://www.fca.org.uk/publications/finalised-guidance/fg24-1-finalised-guidance-financial-promotions-social-media . 26 March 2024. Read at source on 2026-10-10.
34. BTO Solicitors, article on the FCA and influencers. https://bto.co.uk/blog/financial-conduct-authority-crackdown-on-influencers-promoting-financial-products . 6 March 2026. Secondary.
35. European Securities and Markets Authority, joint warning to consumers on crypto-assets. https://www.esma.europa.eu/press-news/esma-news/eu-supervisory-authorities-warn-consumers-risks-and-limited-protection-certain . 6 October 2025. Read at source on 2026-10-10.
36. Kolscan, daily leaderboard. https://kolscan.io/leaderboard . 10 October 2026. Read at source on 2026-10-10 through r.jina.ai. The middle value is our calculation.
37. Kolscan, 30-day pages for seven wallets, in the order of the daily board on 10 October 2026. 1st: https://kolscan.io/account/DrJ6SnDXkEsPeGdmSs93v5rwWumv5QMvAGSZjAyWSd5o?timeframe=30 ; 2nd: https://kolscan.io/account/xyzfhxfy8NhfeNG3Um3WaUvFXzNuHkrhrZMD8dsStB6?timeframe=30 ; 3rd ($312,613 on $4.189 million): https://kolscan.io/account/4vw54BmAogeRV3vPKWyFet5yf8DTLcREzdSzx4rw9Ud9?timeframe=30 ; 4th ($133,791 on $3.324 million): https://kolscan.io/account/BQVz7fQ1WsQmSTMY3umdPEPPTm1sdcBcX9sP7o6kPRmB?timeframe=30 ; 5th: https://kolscan.io/account/5B79fMkcFeRTiwm7ehsZsFiKsC7m7n1Bgv9yLxPp9q2X?timeframe=30 ; 6th: https://kolscan.io/account/Bi4rd5FH5bYEN8scZ7wevxNZyNmKHdaBcvewdPFxYdLt?timeframe=30 ; 11th: https://kolscan.io/account/ardinRsN1mNYVeoJWTBsWeYeXvuR9UUDGMsCDKpb6AT?timeframe=30 . 10 October 2026. Read at source on 2026-10-10 through r.jina.ai. Percentages of volume are our estimates.
38. The Block, on the top Pump.fun trader’s accumulated profit. https://www.theblock.co/post/345046/top-pump-fun-traders-profits-near-40-million-as-solana-memecoin-volumes-shrink . 6 March 2025. Secondary.
39. Decrypt, on a $16.20 purchase sold for about $3 million. https://decrypt.co/291739/lucky-trader-16-3-million-pnut-meme-coin . 14 November 2024. Secondary.
40. Decrypt, on a wallet that launched 17,794 tokens. https://decrypt.co/303439/crypto-trader-840k-17000-meme-coins-three-months . 29 January 2025. Secondary.
41. Preprint from Carnegie Mellon and EPFL researchers on all Pump.fun coins. https://arxiv.org/abs/2609.10246 . 9 September 2026. Read at source on 2026-10-10.
42. Pine Analytics, “Exit liquidity machines”. https://pineanalytics.substack.com/p/exit-liquidity-machines . 21 April 2025. Read at source on 2026-10-10 through r.jina.ai.
43. Preprint on a 15-day test of an automated memecoin trader. https://arxiv.org/abs/2606.08232 . 6 June 2026. Read at source on 2026-10-10 (summary only).
44. Preprint on AI agent accounts trading memecoins with real money. https://arxiv.org/abs/2609.05663 . 4 September 2026. Read at source on 2026-10-10 (summary only).
45. Cointelegraph, on AI agent tokens falling from their highs. https://cointelegraph.com/news/ai-tokens-down-90-from-highs . 6 February 2025. Secondary.
46. Nansen, “Top 10 memecoin wallets to track for 2025”. https://www.nansen.ai/post/top-10-memecoin-wallets-to-track-for-2025 . 19 February 2025. Read at source on 2026-10-10.
47. CoinWire, study of memecoins promoted by X accounts. https://coinwire.com/76-percent-twitter-influencers-promote-dead-memecoins-study/ . 20 November 2024. Read at source on 2026-10-10.
48. ForkLog, on a price sheet for paid crypto promotion. https://forklog.com/en/zachxbt-exposes-hundreds-of-influencers-for-undisclosed-token-promotions/ . 2 September 2025. Secondary.
49. David Z. Morris, interview with a full-time memecoin trader. https://davidzmorris.substack.com/p/how-to-trade-memecoins-interview . 29 June 2025. Read at source on 2026-10-10 (free part only).
50. Cointribune, on Pump.fun’s fees and a Coin Metrics study. https://www.cointribune.com/en/crypto-pump-fun-earns-18-6m-despite-memecoin-slump/ . 9 October 2026. Secondary.
51. Cointribune, on a 90-day count of Solana memecoin traders. https://www.cointribune.com/en/crypto-solana-memecoins-trap-nearly-94-of-traders/ . 20 August 2026. Secondary (a rewrite of source 61). The same figures appear at https://www.diariobitcoin.com/mercados/solo-6-de-los-traders-de-memecoins-de-solana-gano-dinero-en-90-dias/ (19 August 2026).
52. Coinpedia, on Pump.fun traders in March 2026. https://coinpedia.org/news/over-50-of-pump-fun-traders-lost-money-this-month-while-2-wallets-made-over-1m/ . 25 March 2026. Secondary.
53. The Defiant, “Everyone but traders make money from memecoins”. https://thedefiant.io/-everyone-but-traders-make-money-from-memecoins . 21 August 2024. Secondary.
54. NBC News, on wallet results for one memecoin. https://www.nbcnews.com/tech/crypto/58-crypto-wallets-made-millions-trumps-meme-coin-764000-lost-money-dat-rcna205237 . 6 May 2025. Secondary.
55. The Block, on wallet results for another memecoin. https://www.theblock.co/post/342266/traders-of-solana-based-libra-memecoin-lost-251-million-nansen . 19 February 2025. Secondary.
56. Link page that links to @treysocial’s X account and to a Twitch channel. https://beacons.ai/trey . Undated. Read at source on 2026-10-10 through r.jina.ai; the Twitch bio was read from Twitch’s public data service the same day.
57. Whop storefront “Jellycubes Alpha Community”. https://whop.com/jellycubes and https://whop.com/jellycubes/products . Undated. Read at source on 2026-10-10 through r.jina.ai.
58. Whop listing for a memecoin mentorship. https://whop.com/crypto-viper/viper-inner-circle/ . Undated. Read at source on 2026-10-10.
59. Odaily analysis of 541 top Solana wallets, as republished by TechFlow. https://www.techflowpost.com/en-US/article/19306 . 26 July 2024. Secondary.
60. DL News, on Pump.fun buying Kolscan. https://www.dlnews.com/articles/defi/memecoin-kols-bag-millions-as-pump-fun-buys-kolscan-tracker/ . 11 July 2025. Secondary; read on 2026-10-10 through r.jina.ai.
61. Cryptopolitan, “Just 6% of Solana meme traders turned a profit in 90 days”. https://www.cryptopolitan.com/6-solana-meme-traders-profit-in-90-days/ . 19 August 2026. Secondary (reports a Dune query, https://dune.com/queries/8354772/12355632 , and a Dune chart for the fomo app, neither of which we opened); read on 2026-10-10 through r.jina.ai.

Corrections

If you are quoted or named on this page and think something is wrong, want your reply shown beside it, or want to be removed, write to [email protected]. We aim to reply within 14 days, and always within one month. Factual errors are corrected with a dated note here. A sentence that is seriously disputed comes down while we check it.

Change log: (empty)