Copy 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 22 tweets on this scheme¹. View counts are as collected on 6 October 2026. Dates are worked out from each tweet’s ID number. The page judges the scheme, not the people who tweeted.

“WHY NOBODY IS TALKING ABOUT THE BEST WALLET FOR COPY TRADING? Copied three of his trades manually before adding to the bot. All winners. This guy joined just 2 weeks ago and is already sitting on $183,000.”

@DextersSolab, 3 October 2026, 3,605 views¹. The tweet goes on with a wallet link and is cut off in our copy.

“THIS GUY MADE JUST 23 TRADES ON CRYPTO MARKETS And he’s already sitting on $42,058. The best part? His win rate is almost 100%.”

@DextersSolab, 5 October 2026, 1,579 views¹. Also cut off in our copy.

“WANT TO COPY TOP CRYPTO TRADERS? You don’t need to spend hours analyzing charts to get started with copy trading. With the FOMO app, you can discover experienced traders, explore their strategies, and copy their trades directly from your phone.”

@cryptodylnews, 2 October 2026, 1,669 views¹. Cut off in our copy after this point.

“September finished +$1.1K before brokerage fees. Today was a $159 red day. The month still closed green. Here’s my full @UsePeakBot copy trading review the wins, losses, and results since launch:”

@ActOfProfits, 30 September 2026, 755 views¹. We did not open the linked review.

The collection is small and lopsided. Eight of the 22 tweets are one sentence in slightly different wordings: the writer says a copy trading bot made them $14,000 the week before. They were posted as replies by two accounts with about 500 and 1,000 followers, all within about two and a half hours on 6 October 2026, and they had 27 views between them when collected¹. We do not name small accounts, so those replies are described here, not quoted. That sentence is the claim the page tests most closely, because it is the one with a number.

One tweet from August 2024, by an account with 544,967 followers, holds 549,083 of the roughly 624,000 views in the whole collection (our sum)¹. It offers to hire followers and “share my calls”, and asks for a comment and a follow¹. It is about following someone’s trades, but it states no income from copying.

None of the 22 tweet texts gives the money put in, the percentage return, the fees paid, or how the copier did compared with the trader copied¹. We did not view the images, videos or threads attached to the tweets, which may say more.

What the scheme is

Copy trading means your account repeats someone else’s trades automatically. You pick a person or an account to follow, set an amount, and software places the same buys and sells for you. It comes in four forms.

• A feature of a broker or exchange. eToro, Binance, Bybit, OKX and Bitget each run one. You stay inside your own account, and the platform does the copying. The person copied is often called the lead trader.
• Signal copying into a trading account. A provider’s trades are sent to your account on MetaTrader (MT4 or MT5, common software for trading currencies and indices) or at a futures broker, usually for a monthly fee.
• Wallet copying. In crypto, every trade a “wallet” (an account on a public blockchain) makes can be seen by anyone. A bot, often run through the Telegram chat app, watches a wallet you choose and repeats its trades with your money seconds later. The same idea is applied to Polymarket, a site where people bet on the outcome of events.
• Vaults. You deposit money into a pool and its leader trades the whole pool. Hyperliquid, a crypto exchange, has thousands of these.

The mechanism is real and, on licensed platforms, legal for the copier. The questions are what it costs, how often copiers end ahead, and how much money it takes to produce the sums in the tweets.

The arithmetic

The sum behind every claim is:

profit = money put in x return x (1 minus the lead trader’s share) minus fees

The tweets give the profit and leave out everything on the right-hand side¹.

Step 1: what the copied trader takes

On exchanges, the lead trader is paid a “profit share”: a cut of whatever the copier gains.

• Binance futures: “Up to 30% profit share from their copy traders’ earnings. A 10% commission on their copy traders’ trading fees.”² The landing page advertises 10% as the profit share³. Futures are bets on a price, usually made with borrowed money; “spot” means buying the coin itself.
• Bybit: 10%, 10%, 12% or 15% of the follower’s net profit, depending on the lead trader’s rank, settled weekly⁴. Its copy trading for non-crypto markets charges 15% to 30%⁴.
• OKX: up to 30%, settled weekly. Bitget: 10% to 20% by tier, on a page from 2023 that may be out of date⁵⁷.
• eToro: “There is no additional charge for copying another trader or traders.” The minimum is $200 per trader copied⁵. Normal trading costs still apply, and eToro charges $5 to withdraw from a US-dollar account; withdrawals from local-currency accounts are free⁶.

The share is taken on gains. On Bybit the lead trader is paid only when the follower’s closed positions show a net profit for the period, and is not charged when they show a loss⁴.

Step 2: what the tools charge

Wallet-copying bots take a cut of every trade. Trojan: “Trojan charges 1% per successful trade.”⁷ GMGN: “GMGN only charges a 1% handling fee for a single transcation” (its spelling)⁸. Axiom: 0.75% to 0.95% depending on volume⁹. A buy and a later sell are two trades, so a round trip costs about 2% before anything else⁸.

Our estimate: a copier who makes 100 round trips of $500 pays about $1,000 in bot fees (100 x $500 x 2 x 1%). Network fees are extra. A priority fee or tip is a payment to the network to process a trade faster. GMGN says these are not collected by GMGN and can be set up to 2 SOL, the coin used on the Solana network⁸.

The copier gets a worse price. The copier buys after the copied wallet has already pushed the price up. “Slippage” is the gap between the price you expected and the price you got. The bots’ own instructions tell copiers to allow a lot of it. Trojan: “Copy Trading will require a higher number, especially if it’s a heavily copied wallet. … Start low (10-15%) and adjust.”⁷ GMGN: “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”⁸. These are ceilings a user sets, not measured losses.

We found no measurement on real copier accounts. One academic backtest (a test on past data, not with real money) estimates the gap. Using pump.fun data on 6,000 tokens, it reports that the wallets its filter picked “achieve an average return of 14%, while the corresponding estimated copier return is 3% per meme coin investment under realistic market frictions”⁵⁴. That is about 11 points lost per coin by copying, even with the paper’s best filter. For simpler ways of picking wallets it found “copier returns remaining negative despite accurate wallet identification”⁵⁴. So picking a wallet that really is profitable was not enough. The paper’s description of pump.fun includes that site’s own 1% fee; a Telegram bot’s fee of about 1% per side would be on top (our reading)⁸⁵⁴.

Polymarket charges the copier a fee. Someone who takes an existing offer pays; someone who posts an offer does not: “Makers are never charged fees. Only takers pay fees.”¹⁷ On crypto markets the formula is shares x 0.07 x price x (1 minus price)¹⁷. As a share of the money traded, that is about 7% for shares priced near 0 cents, 3.5% at 50 cents ($1.75 on a $50 trade) and close to zero near 100 cents (our division from Polymarket’s table)¹⁷. In dollars per share it is highest at 50 cents. Geopolitics markets are free¹⁷. A copier entering after the copied wallet would normally be the taker. That is our reading, not Polymarket’s statement.

Subscription tools cost a fixed sum each month.

• PeakBot, the tool in one example tweet: “PeakBot Pro Subscription $199/month”, and a brokerage account approved for borrowing and for combined options trades is required¹⁰. Its page for traders says subscribers pay at least $150 a month to copy a trader¹⁰. We did not establish whether a subscriber pays one of these or both.
• Alertsify: $99 a month for one-tap copying, $249 a month or $1,997 a year for automatic copying, with subscriptions to individual traders sold separately; its example is a “separate $189/month VIP membership” to copy one trader¹³.
• MetaTrader’s own signal market: most listings we saw were $30 to $50 a month, some $100¹⁵.
• US30 Trading Lab, which posted one of the 22 tweets: $499 for a year of “VIP Copy Trading”, $99 a month for signals¹⁶.

Our estimate: $199 a month is $2,388 a year. On a $10,000 account that is 23.9% a year that must be earned before the first dollar of profit.

That puts the ”+$1.1K” tweet in context. It reports $1,100 for September before brokerage fees and does not state the account size¹. PeakBot’s page for this trader says its figures are “based on 1x unit sizing, gross of broker fees, since launch. Pulled daily from actual broker fills”, and the page’s code sets one unit at $3,000 of buying power¹¹. We do not know whether the tweet’s $1,100 is on one unit. A subscriber with the same gross result would keep about $900 to $950 after a $150 to $199 subscription, before brokerage fees; a $199 subscription is about 18% of a $1,100 month (our estimates).

Step 3: the money needed

Rearranged, the sum is: money needed = profit wanted ÷ net return for the period.

We found no measured typical return for people who copy. So the table uses four rates, and the rates are our assumptions:

• 2% a month net. That is about 27% a year, kept up every month. For scale, the best of five traders eToro shows on its own copy page returned 54.89% over two years, which is about 1.8% a month (our sum)⁵.
• 10% a month before a 30% profit share, so 7% net.
• 10% a month before a 10% profit share, so 9% net. A single month at 10% happens. A promise of it every month is another matter: the US derivatives regulator’s advisory on trading bots describes a bot programme that “guaranteed at least a 10 percent monthly return (or more than 200 percent per year)” and says “very little money was actually traded”³³. So these two rows show one good month, not a rate to plan on.
• 11.74% a month before a 10% share, so about 10.6% net. This is the last 30 days of the largest account among those featured on Binance’s futures copy page when we fetched it on 10 October 2026: “30 Days PNL (USD) 685,728.11 ROI 11.74% AUM 6,583,116.08 30 Days MDD 14.30%”³. PNL is profit and loss in dollars. ROI is the percentage return. AUM is, in Binance’s words, the “Lead Trader’s Portfolio Investment Amount + Total Copy Trading Investment Amount”, so it includes the lead trader’s own money³. MDD is the largest fall from a peak in the period.

All figures in the table are our estimates. A week is taken as 7/30 of a month. For the last cell: 14,000 ÷ (0.1174 x 7/30 x 0.9) = about $570,000. At the 30% share Binance allows, the same cell is about $730,000².

That row is one account in one month. Four featured accounts in our copy of the page had AUM above $1 million. Their 30-day returns were 9.46%, 11.74%, 21.49% and 84.54%³. On the same sum, $14,000 in a week takes about $705,000, $570,000, $310,000 and $79,000 (our estimates). The two higher-return accounts had 30-day falls of 53.42% and 26.76% from their peaks³.

With a subscription tool, add its monthly price to the profit wanted before dividing.

Four cautions about the Binance figures. They cover one month. They are the lead trader’s own return, not what copiers received. The page is a showcase that changes: a review agent’s fetch of the same address later the same day returned a different list, so we kept a copy of what we read. And a showcase shows current winners only. The same page had rows with far higher returns and far deeper falls, for example “ROI 7,499.13% AUM 351,062.60 30 Days MDD 71.91%”³.

So $14,000 in a week is mostly a statement about how much money was at risk, and that is the figure the replies leave out¹. A single week can also land far above or below 7/30 of a month. Read the other way: $14,000 in a week on $14,000 would be a 100% return in seven days, and on $140,000 it would be 10% in seven days (our sums). Weeks like that do occur on small accounts, as the rows with returns in the thousands of percent show. They come with falls like the 71.91% beside them³.

Step 4: how many reach those sums

No source measures income for copiers alone. The nearest figures cover everyone trading on the places the tweets point to.

Notes on each:

• Polymarket. The figures are from an independent analyst using public blockchain data, not a peer-reviewed study²⁰. They count addresses, not people, and realised profit only. The author notes that another analyst, using a different formula in December 2025, found 30% of 1.7 million addresses in profit²⁰. Two of our example tweets link to Polymarket trader profiles (see “Who makes money from it”).
• Fomo. The figures are from a write-up by DWF Ventures, an investment firm, which says “only 6.16% of 292k wallets analyzed on the platform were profitable in the last 3 months (based on realized profits). Even among profitable wallets, net profits exceeding $10k were only limited to 25 wallets.”²¹ That is 1 wallet in about 11,700 (our division). The data comes from a public query by an X user that we could not open, so the method is unchecked²¹. A second report of the same data says 88% of the profitable wallets made under $100 and the median wallet lost $120²². The dataset is every wallet on the app, not only people copying automatically. Neither report carries a reply from Fomo²²²³.
• pump.fun is a site for launching new tokens; a token “graduates” when it moves to an open exchange. One tweet in our collection lists wallets of token creators to track, with “Graduation rate: 100%”¹. The 0.4% figure was reported by Decrypt in January 2025²⁴. The site’s co-founder disputed the method. He replied that “most of the gains are made AFTER a coin [graduates]” and that “It’s likely that the true number of profitable and extremely profitable wallets is an order of magnitude LARGER than exhibited”²⁴. Decrypt added his comments after first publishing the story²⁴.

Two more Polymarket figures bear directly on the example tweets, which point to a wallet with a two-week record and one with 23 trades¹. 0.98% of addresses earned over $5,000 in any single month. “Two months in a row drops to 0.1%.” “Three months in a row is 0.03%.”²⁰ And “Only 172 addresses out of 6,600 with average profit above $5,000 were active for more than a year.”²⁰ A short hot record is the kind that rarely continues.

What people who tried it report

We found no account from an individual copier with proof of income, such as payment records checked by a third party. Forum threads were not opened. Everything below is counted across many accounts.

On exchanges, about half of follower outcomes were positive. YieldFund, a company with its own investing product, copied public data from Binance, Bybit and MEXC by hand in October 2025: 1,486 lead traders and their followers over 90 days²⁵. “In our analysis of 100,236 outcomes, we found that 48,592 trades were winning, while 51,644 trades resulted in losses.”²⁵ That is 48.48%. The study’s unit is an “outcome”; it also calls them trades and events, and says its result is “event-weighted”²⁵. So it is not established that each one is a single person’s result over the full 90 days. By exchange: Binance 66.5%, MEXC 57.79%, Bybit 43.65%²⁵.

Added together, the followers were up about 709,000 USDT, a crypto coin pegged to the dollar²⁵. That total is gains minus losses: +572,979.94 on Binance, +346,106.01 on Bybit and -210,040.51 on MEXC²⁵. It works out at about $7 per outcome (our division), but the study gives no sizes of individual gains or losses. On MEXC a majority of wins still summed to a loss, which the authors put down to “occasional significant losses”²⁵. They also say the profits “were concentrated in a few leaders”²⁵.

Limits: this is a count of wins and losses, not of money per person. It is one 90-day window. Lead traders who were removed after heavy losses are not in it. It does not separate out fees or profit share²⁵.

The leaderboard does not predict the follower’s result. In the same data: “97.04% of the leaders in our sample recorded a positive PnL, while only 43.61% produced a positive follower PnL.”²⁵ Nearly every lead trader showed a gain on their own money, and fewer than half made money for the people copying them. The authors point to later entry prices, fees and profit share, without measuring each²⁵.

Platforms’ own figures are the most favourable. Bitget said that in the first half of 2023, “93% of those involved in futures trading and 82% involved in spot trading achieved profitability, with cumulative gains surpassing 74 million USDT”, across more than 109,000 profitable followers²⁶. It gives no method and no count of losing followers, and prices were rising in that period²⁶. Taken at face value, the average winner made about 680 USDT in six months (our division).

Followme, a social trading platform, published a report on 17,727 active accounts in the first half of 2025: “copy trading proved the most stable, with 47.27% posting net profits and an average net gain of USD 322. In contrast, self-directed traders faced average losses of USD 3,236, while signal providers also posted negative returns, averaging a loss of USD 1,297.”²⁷ We read this in a trade-press section that carries company material. On that platform, copying did less badly than trading alone, and the traders being copied lost on average.

Copying whoever is on top has been tested, and it lost. A 2018 study in a peer-reviewed journal found: “simply investing in those traders with the highest accumulated returns leads to high losses, while taking the Sharpe ratio into account improves the achieved returns. Positive returns, however, are only possible for sophisticated strategies that consider information on the risk of the signal providers’ portfolios.”²⁸ The Sharpe ratio is a measure of return for the risk taken. A summary by an investors’ association gives the size: copying the top-ranked portfolios lost 92% over 36 months on one German platform and nearly 79% over 29 months on another; the risk-based approach still lost 36% and 13%²⁹. The data is from before 2018, on platforms that are not the ones promoted on X today. We read the abstract, not the full paper.

A copy button changes behaviour. In a laboratory experiment, 17 of 48 people chose to copy and 12 of those 17 copied the single top earner³⁰. The authors write: “We conclude that copy trading reduces ex-ante welfare, and leads to excessive risk taking.”³⁰ This is an experiment, not a measure of real losses. It suggests why leaderboards are risky to choose from: whoever is on top often got there by taking the most risk.

On wallet-copying bots, a vendor’s own guide describes the outcome. GMGN’s copy-trade page says: “a lot of users, although their copy trades profit, but the gas priority fee is too high, resulting in the end of the still lose money”⁸. That is the seller’s statement, not a measured rate.

An older study found copying beat trading alone. A 2012 study by researchers at MIT used data from an online currency broker with a copy feature. Its abstract does not name the broker; it is identified elsewhere as eToro, which we did not confirm. It concluded: “generally social trades outperform individual trades”⁵⁵. We read the abstract only. Beating a retail currency trader’s own trades is a low bar. And a comparison of trades says little about money won or lost: the MEXC result above shows a majority of winning outcomes adding up to a loss²⁵.

What the rules allow now

For the copier

Copying inside a licensed firm’s feature is allowed, and the copier needs no licence. In the UK the firm does: “We classify copy trading as portfolio or investment management where no manual input is clear from the account holder.”³² The EU markets regulator, ESMA, takes the same view where “the order execution is automatic and does not require any further action from the client”³¹. Its 2023 briefing is guidance to national regulators and is not binding.

Even there, the copier carries the losses. eToro’s page says: “Copy Trading does not amount to investment advice. The value of your investments may go up or down. Your capital is at risk.” It also states: “51% of retail investor accounts lose money when trading CFDs with this provider.”⁵ A CFD (contract for difference) is a bet on a price move, usually made with borrowed money. The 51% covers all such accounts at eToro, not copiers alone.

A decided UK complaint shows that the copier is expected to understand what is being copied. A customer was copying a trader whose account the platform then closed. A “stop” is a price at which a position is closed automatically; “margin” is the money put up to keep a position open. The trader had put in extra money so his positions could survive a bigger fall. The copier’s account had the same exit price but not the extra money. In the Financial Ombudsman’s words: “because G had extended the stop rate by adding margin to its own positions, Mr A had the same stop loss rate, but without the increased margin.”⁴⁵ The complaint against eToro was not upheld; the Ombudsman found the notice given was reasonable and the choice to copy was the customer’s⁴⁵.

MetaTrader’s signal market says the same in its rules: “the Subscriber understands all risks involved and takes full responsibility for copying trading operations of others in his own trading account.”¹⁵

For people who sell signals or run copy groups

• EU. ESMA says “the trades that are being shared on social media could still qualify as investment advice provided by the copied trader who share their trades”³¹.
• US. Anyone who is paid to advise others on trading futures or currencies must register as a commodity trading advisor unless an exemption applies. One exemption: “Advice was provided to 15 or fewer persons during the past 12 months and the entity does not generally hold itself out to the public as a CTA”³⁶. The securities regulator’s guidance from 2009 says: “‘Auto-trading,’ like any other arrangement that allows someone else to trade in your account without first asking your permission, can be highly risky.”³⁷
• A settled US case. In 2023 a court entered a consent order (an agreed settlement) under which BareIt Media, which traded as SignalPush from 2013 to 2017, and its owner pay a $100,000 penalty. The firm had offered trade signals with automatic execution on binary options, a kind of yes-or-no bet on a price, without registering³⁵. One CFTC commissioner dissented, saying the agency “seems to be once again changing its interpretation of the definition of a ‘commodity trading advisor’” through an enforcement case³⁵. The order is about registration and makes no finding about customers’ results.
• A pending UK case. The UK regulator has charged three people who, in its words, “are alleged to have encouraged social media followers to invest in foreign exchange (forex or FX) trading through high-risk products known as contracts for difference, without having the authorisation to promote these investments”³⁹. They pleaded not guilty and trials are set for 2027. These are allegations and the case is pending. It concerns promoting trading, not copy trading by name.

We did not check whether any account or product in our collection is registered anywhere. PeakBot’s own disclaimer says it is not registered as a broker-dealer, investment adviser or commodity trading advisor and relies on an exclusion for publishers¹².

What regulators say about the marketing

ESMA tells supervisors that “the information on past performance should not be the most prominent element of the communication, and it should include performance information which covers the preceding 5 years”³¹. That applies to EU-regulated firms. The example tweets show two weeks, 23 trades or one week¹.

The US derivatives regulator, the CFTC, published an advisory on trading bots, announced on 25 January 2024. It begins: “Fraudsters are exploiting public interest in artificial intelligence (AI) to tout automated trading algorithms, trade signal strategies, and crypto-asset trading schemes that promise unreasonably high or guaranteed returns.”³³ The advisory, which is about bots marketed as using AI, lists very high returns and near-perfect win rates among the claims to distrust³³. Those words are the CFTC’s, about the cases it cites. The advisory names no account or product in our collection, the tweets in our collection do not mention AI, and we do not apply its words to any of them.

A different product that uses the same words

Some offers ask you to send money to an operator whose “bot” trades a pool. That is not copy trading in your own account. In the largest such case, a US court entered a default judgment against the founder of Mirror Trading International and ordered $1,733,838,372 in restitution³⁴. The CFTC says the defendants “purportedly traded off-exchange, retail forex through what they falsely claimed was a proprietary ‘bot’ or software program”, taking at least 29,421 bitcoin from at least 23,000 people in the US³⁴.

The cases in this section concern only the companies and people named in them. We found no regulator or court action against any account quoted or named on this page, and we do not suggest any connection.

Regulatory actions involving eToro that are not about copying

Two regulator actions involve the largest copy platform but concern other products, so they are noted only briefly. In the US, eToro USA LLC paid $1.5 million in 2024 to settle charges about crypto trading, without admitting or denying the findings³⁸. In Australia, the regulator sued eToro Aus Capital Limited in 2023 over who it allowed to trade CFDs; these are allegations, the case is pending, and the regulator’s page lists hearing dates up to 28 April 2026 and shows no judgment as of 10 October 2026⁴⁰. We did not read a reply from eToro at source. Both concern only the companies named. eToro’s 2025 annual report says CopyTrader is available in the US only “on a limited basis”⁴⁸.

Platform rules

• X. Its rules on platform manipulation prohibit “repeatedly posting identical or nearly identical posts in a duplicative manner popularly known as ‘Copypasta’” and bulk unsolicited replies⁴¹. Whether any reply in our collection breaks that rule is for X to decide. We read X’s rules in an archived copy of 7 October 2026; the live page refused our tools.
• Polymarket. Its documentation lists more than 30 countries and regions (37 entries, with five more restricted on the website only), including the United States, United Kingdom, France, Germany, Italy, Australia and Singapore, as “close-only”: “Close-only means users can close existing positions but cannot open new ones.”¹⁹ A reader there cannot open new positions on the main platform, with or without a bot. We did not check its separate US offering.
• Prop firms. These are companies that let traders use the firm’s money after passing a test. Some restrict trade copying. One copier vendor tells its customers: “Tradecopia provides the copy trading tool. Ensuring your firm permits its use is your responsibility.”⁴⁷

Risks that have nothing to do with the trader you copy

• The bot itself can be attacked. A security firm reported: “In late October two of the most popular Telegram trading bots were exploited, leading to a total loss of $1.1 million.”⁴² The bots were Maestro and Unibot, in 2023. A commentary site puts an incident on 19 September 2024 at another bot, Banana Gun, at about $3 million across 11 users, and reports that Banana Gun promised full refunds from its own treasury⁴³. We did not confirm that the refunds were paid. Whether Maestro and Unibot users were repaid is not stated in the report we read⁴².
• Free bots can carry code that takes the wallet. StepSecurity, a security firm, found more than 20 code repositories presented as Polymarket trading bots in one account on GitHub, a code-sharing site, the lead one named for copy trading. It reports that “hidden inside its npm dependencies are two typosquatted packages that steal wallet private keys, exfiltrate sensitive files, and open an SSH backdoor on the victim’s machine.”⁴⁴ In plain words, the download contained look-alike code that copies the wallet’s secret key and lets the attacker into the computer.
• Tax. In the US: “If you have digital asset transactions, you must report them whether or not they result in a taxable gain or loss.”⁴⁶ A bot that makes hundreds of trades makes hundreds of reportable events. We did not check other countries.

Who makes money from it

Three groups are paid whether or not the copier ends ahead.

The platform or bot. A bot takes about 1% of each buy and each sell⁷⁸. An exchange keeps its trading fees on every copied order. Polymarket charges the taker¹⁷.

The trader being copied. Binance and OKX lead traders can take up to 30% of followers’ gains, and at Binance also 10% of followers’ trading fees²⁵⁷. eToro pays its copied traders itself: “Receive a payment of up to 1.5% of assets under copy (AUC), paid out monthly (Champion tier and above)”⁵⁰. Champion is the second of the programme’s four tiers; the first is unpaid⁵⁰. PeakBot’s page for traders is headed “Earn $10,000+ Per Month Letting Others Copy Your Trades” and says: “A minimum of $150/month is required, but you can set the pricing to whatever fee per month you would like to charge your subscribers!”¹⁰ Alertsify says lead traders “keep up to 95% of subscription revenue”¹³. The earnings headlines are the companies’ own marketing and are not audited.

ESMA names this as a conflict of interest: “Such conflicts of interest, in the case of copy trading services, can arise in particular from the structure of the payment made to the traders whose trades are being copied”³¹.

Copying is also concentrated. A 2018 paper found “The top 5% of copied traders accounts for 61.1% (ZuluTrade) to 92.8% (eToro) of copier relationships.”³⁰

The person who refers you. Bots and platforms pay promoters a share of the fees their referred users generate.

• GMGN’s referral page: “Invite your TG group members, Twitter followers, and friends to trade and easily earn over 40 Sol ($8000+) monthly!”, with 10% to 30% of referred users’ fees⁸. The $8,000 is GMGN’s marketing figure with no data behind it.
• Trojan’s FAQ says “Receive close to 50% of revenue share from your referrals.” Its referral page describes five layers with totals of 27.5% to 32.5% at its first two levels⁷. We did not reconcile the two.
• Polymarket’s terms, effective 28 May 2026, pay a referrer 10% of net trading fees from direct referrals and 5% from indirect ones, for the first 30 days¹⁸. The referrer needs “$10,000 in lifetime trading volume”, and rewards end earlier if the referred user reaches a high fee tier¹⁸. An article from April 2026, before those terms, reported 30%²⁰.

So a promoter can be paid for each trade a reader makes. That is a fact about the pay structure, not about any person.

What the 22 tweets lead to

Our tally of the tweet texts and the collector’s labels¹:

No tweet links to a course. We do not know which bot, if any, the eight replies lead to: their text shows no link¹.

What we saw on the sellers’ own pages:

• The two wallet tweets. Both short links in the @DextersSolab tweets resolved on 10 October 2026 to Polymarket trader profiles, and both carried the tag ?via=dexter-molu¹⁸. Polymarket’s referral programme works through shared market and profile links; its page does not describe the tag itself¹⁸. Rewards are paid only if the sharer has $10,000 of lifetime trading volume and a new user signs up through the link¹⁸. We do not know whether this account qualifies or has earned anything. We did not check the two wallets’ balances or records.
• The PeakBot review. PeakBot’s site lists “Act of Profits” as a trader whose futures and options trades subscribers can copy, live since 10 September 2026, and says: “PeakBot mirrors Act of Profits trades automatically, so entries and exits are executed directly in your connected brokerage account.”¹¹ The page says its results are taken daily from broker fills, for one $3,000 unit, before broker fees¹¹. Those live results did not load for us, so the +$1.1K is the account’s own statement. PeakBot says it pays such traders a share of subscription revenue; the split is not published¹⁰.
• US30 Trading Lab. Its site sells VIP signals at $99 a month, $399 for six months, $499 a year and $999 for life, “VIP Copy Trading 1 Year Access” at $499, and a course at $2,000¹⁶. Its refund policy says there are no refunds on monthly, six-month, yearly or lifetime plans, and “If you profit from any of our signals there will be no refund!”¹⁶ The same page opens with “a 30-day return policy”, so the two statements conflict; we report both¹⁶. We found no checked track record on the pages we read.
• Alertsify. One tweet, by @EnhancedMarket, says “Copy trading across multiple different platforms” with a link we did not open¹. Alertsify’s home page says that “copying Enhancedmarket requires a separate $189/month VIP membership”¹³. Alertsify’s policy says: “all payments are final and non-refundable”, and lists losing money as not a reason for a refund¹⁴. That policy page lists different prices from the home page and mentions per-trade fees, while the home page says there are none; we did not establish which is current¹³¹⁴. The home page says it publishes broker-verified leaderboards with losses shown¹³. We did not audit them.
• PeakBot’s refund terms. A 30-day guarantee that “is immediately void if any coupon, promotional code, or discounted offer was applied”¹².

The upside

Copy trading does pay some people well. The best-documented cases are on the side of the person being copied, and they are a small minority of those who are copied.

Being the trader others copy

eToro’s annual report, filed with the US securities regulator, says: “As of December 31, 2025, we had over 4,750 members of the Pro Investor program. 17 of these Pro Investors had over $10 million in Assets Under Copy, and 125 had over $1 million in Assets Under Copy. 65% of participants in the Pro Investor program have a tenure of five years or more”⁴⁸. The filing does not define tenure. A year earlier eToro reported over 3,300 members, 12 above $10 million and “over 900” with five years or more⁴⁹. A jump from about 900 to about 3,090 in a year (our sum) suggests the 2025 figure counts years as an eToro user, not years of being copied.

eToro’s programme page sets out four tiers⁵⁰. Our estimates of yearly pay from its rates:

The counts are from a regulatory filing. The pay is our multiplication from the programme page, which says: “Payments begin at the Champion tier. Cadet-level participants do not receive payments”⁵⁰. How many of the 4,750 members are at Champion or above is not published. The 4,750 members are about 0.12% of eToro’s 3.81 million funded accounts (our division)⁴⁸. Entry needs two months of history and risk limits⁵⁰.

Running a vault

Hyperliquid publishes data on every user-run vault. One of our research agents downloaded the file on 10 October 2026 and counted⁵¹. All of these are our estimates from that file. A second run reproduced them, apart from a difference of two in the number of vaults that ever traded, because the file changes continuously:

• 9,469 user vaults, created since February 2023. 2,155 show a profit over their whole life. That is 22.8% of all, and 26.8% of the 8,032 that ever traded.
• 716 made over $1,000, 273 over $10,000, 70 over $100,000 and 11 over $1 million.
• Added together, the user vaults lost about $70 million: about $54 million of gains against about $124 million of losses. One vault accounts for about a third of the net loss. The median vault that traded lost $142.
• Among the 73 open vaults holding $100,000 or more, 52 (71%) are in profit. That figure flatters the picture, because vaults that lose money shrink or close and leave the group.

This is profit at the level of the vault, not what each depositor got, which depends on when they joined.

A vault leader is paid: “Vault leaders receive a 10% profit share for managing the vault.” The leader must keep at least 5% of the vault in their own money, and “Creating a vault requires a 10k USDC gas fee.”⁵² USDC is a crypto coin pegged to the dollar, and the share is on profit only. Hyperliquid now marks that documentation page “legacy” and describes a newer kind of vault that builders set up themselves; we did not find leader terms for the newer kind⁵².

Large winners exist on the venues the tweets point to

On Polymarket: “More than $100,000 was earned by 840 addresses, or 0.033% of all traders.”²⁰ About 6,600 addresses average over $5,000 a month, and 172 of those were active for more than a year²⁰. Whether those are skilled people, bots, or firms whose business is quoting prices to both buyers and sellers is not known.

Slow copying of public filings

One regulated fund copies a different kind of record: the share purchases that Democratic members of the US Congress must disclose. Its page shows a return of 112.39% from February 2023 to 30 September 2026, or 22.94% a year, with a yearly cost of 0.72%⁵³. The same page gives its benchmark, the S&P 500 index with dividends, as 95.74% and 20.22% a year over the same period⁵³. That is one fund in one rising market, weighted towards large technology companies, so the lead may come from that weighting and not from copying. It is a fact about a method, not a suggestion to buy the fund.

What the top tenth and the top hundredth earn

No source gives this for copiers. The nearest figures:

• Polymarket, all traders. “Only 2% of 2.5 million traders have made more than $1,000 over their trading history.”²⁰ About one in 80 (1.25%) averages over $1,000 a month, and about one in 770 (0.13%) averages over $10,000²⁰.
• Fomo, all wallets. About one in 16 (6.16%) ended three months ahead, and most of those by under $100²¹²².
• Hyperliquid vaults. About 2.9% of vaults made over $10,000 across their life and about 0.12% made over $1 million (our division from the counts above)⁵¹.
• Exchange followers. About half of outcomes in one 90-day study were positive; the size of the gains is not published²⁵.

A realistic good result in the first year or two (our estimate)

As a copier on a licensed platform. The five traders eToro chose to show on its own page had two-year returns of -9.49%, 1.30%, 31.82%, 48.43% and 54.89%⁵. For comparison, the S&P 500 index with dividends returned 15.74% in the year to 30 September 2026 and 22.89% a year over three years⁵³. Two years at those rates is roughly 35% to 50% (our estimate; the dates do not match eToro’s exactly). Two of the five showcased traders were in or above that range and three were well below it. So a good result is about what the share market returned in the period, with a real chance of less. On $1,000 to $5,000, a 15% to 25% year is roughly $150 to $1,250. Other figures point the same way: an average gain of $322 in six months on one platform, and about 680 USDT in six months for winners on another²⁶²⁷.

As a copier of wallets through a bot. We found no evidence of a typical positive result. The backtest in Step 2 found a 3% average gain per coin for copiers with its best filter and losses with simpler ones⁵⁴. A memecoin is a token with no business behind it, traded on attention. Against a 3% average gain, a bot’s fee of about 2% per round trip takes most of it (our sum)⁸.

As a copied trader. Most start unpaid. On eToro the first paid step needs $50,000 of other people’s money copying you and $10,000 of your own, and pays $3,000 to $6,000 a year⁵⁰. About 2.6% of programme members reach the level that pays roughly $15,000 a year⁴⁸. Elsewhere the copied side often loses: on Followme, signal providers lost $1,297 on average, and 77% of Hyperliquid user vaults show no profit²⁷⁵¹.

None of the copier results we could check is near $14,000 in a week. The capital table in Step 3 shows what such a week takes.

What it takes to compete

What the people who do well have

1. Capital. At 2% a month, $1,000 a month takes $50,000 (Step 3). With a few hundred dollars, a good year is tens of dollars (our estimate).
2. Selection on risk, not on the top of the list. The only strategies with positive returns in the 2018 study used information on how much risk the copied trader took²⁸. The plain approach, copying the highest past return, “leads to high losses”²⁸. A later study of real signal providers found that “underperforming signal providers increase their net exposure towards lottery-like stocks, in turn exposing signal followers to a lottery-like return structure”⁵⁶. In plain words, traders who were behind took long-shot bets, and their followers got those bets too. We read that one as an abstract only.
3. A long record to judge. ESMA’s yardstick is five years³¹. Exchange leaderboards show 30 to 90 days³²⁵.
4. A leader with their own money at stake and a limit on risk. Hyperliquid requires leaders to hold 5% of the vault⁵². eToro’s programme has risk limits⁵⁰. A leader paid only on gains, with no share of losses, is rewarded for taking risk with followers’ money⁴³¹.
5. Understanding of what is being copied. The Ombudsman decision above turned on the copier not having what the trader had⁴⁵.
6. For wallet copying, your own tooling. The 2026 paper says of plain copying that adversaries “deploy manipulative bots to front-run trades, conceal positions, and fabricate sentiment, systematically extracting value from naive copiers at scale”⁵⁴. Front-running means trading just ahead of someone whose order you can see coming. The paper adds: “Even when a copier immediately replicates the same trades as a smart money wallet, the copier necessarily trades at a worse point on the curve.”⁵⁴ The curve is pump.fun’s pricing rule, under which each buy raises the price for the next buyer. Its positive result needed a purpose-built filter.
7. To be copied: your own money, an audience and time. On eToro the first paid tier needs $10,000 of your own money and five copiers with $50,000 between them, and 65% of members have a tenure of five years or more⁴⁸⁵⁰. On Hyperliquid’s older vaults it meant 10,000 USDC to open a vault and 5% of it in your own money⁵². On PeakBot or Alertsify it means subscribers willing to pay a monthly fee¹⁰¹³. Exchange leaderboards need no long record, which is why they are weak evidence³.

What it costs

Costs are set out in Steps 1 and 2: 10% to 30% of gains on Binance, Bybit and OKX, about 2% per round trip on wallet bots plus network fees and the worse price, and $30 to $249 a month for subscription tools^247810131557. eToro charges no copy fee and needs $200 per trader copied⁵. The refund policies we read exclude losses, exclude most plans or are void with a discount code¹²¹⁴¹⁶.

How to tell early which side you are on

These checks are our judgement, built from the figures above.

• Do you know the percentage, not just the dollars? Divide the profit claimed by the money needed to make it. If nobody will tell you the money, the claim cannot be checked.
• Is the fixed cost under a few percent of your account each year? A $199 monthly tool on $10,000 is 23.9% a year (Step 2).
• How long is the record, and what was its worst fall? Two weeks or 23 trades is short. A 30-day fall of 71.91% sat next to a return of several thousand percent on the same Binance page³.
• Does the leader’s record show what followers got? 97.04% of leaders were up on their own money and 43.61% made money for followers²⁵.
• After three months, are you ahead after every fee? About half of exchange follower outcomes were positive in one study²⁵. On the Fomo app about one wallet in 16 was²¹.
• Does the person recommending it get paid when you trade? Tags naming the sharer are often visible in a link. It does not make the advice wrong, but it means their income does not depend on yours.
• Does the offer ask you to send money to someone, or to paste a wallet key into software? Copy trading on a platform does neither. See the pool case and the bot incidents above³⁴⁴⁴.

What we did not verify

• What copiers earn. No platform, regulator or study we found gives a median return or a loss rate for copiers alone. The capital table rests on assumed rates and on one saved copy of a Binance showcase that changes continuously; a later fetch of the same address showed other rows, which we did not use. eToro’s annual report has no figure for what copiers earned.
• The claims in the tweets. We did not check either wallet’s balance on the blockchain. We did not open the PeakBot review, and PeakBot’s live results page did not load. We do not know which bot the eight templated replies promote, whether it exists, or whether those accounts are automated. Tweet images, videos, threads and profiles were not viewed, and several tweet texts are cut off in our copy. Tweet dates are worked out from ID numbers. That the tweet about “crypto markets” wallets refers to Polymarket rests on where its link resolved. Whether the ?via= tag on those links earns its sharer anything is not known. Whether the PeakBot tweet’s $1,100 is on one $3,000 unit is not known.
• The Fomo figures. The public query behind them was not opened. One report says the wallets together lost $1.26 billion, which looks large beside a median loss of $120; we did not use it²². Whether those wallets copied automatically or bought by hand is unknown. Fomo’s own site does not state its fee; review sites give about 0.5% per trade.
• The exchange study. YieldFund’s spreadsheet was not opened. It sells an investing product. It does not say how the lead traders were chosen, or whether an “outcome” is one person’s result. Its date is taken from the article text; the page’s own data shows 2022.
• The Hyperliquid figures. These are our sums from a file that changes continuously. They assume the last value in each series is lifetime profit in dollars. Terms for Hyperliquid’s newer vaults were not found.
• The two-year index comparison. It is our estimate from one-year and three-year figures on a fund’s page. Price data for the exact two years could not be opened; one review run put the index at about 40%.
• Academic papers. The 2018, 2012 and 2022 studies were read as abstracts only; publisher pages refused us. The loss figures for the 2018 study come from a summary. The 2026 paper’s tables were not checked in detail, and its result is a backtest. That the 2012 study’s broker is eToro is not in the abstract we read.
• Exchange terms. Binance’s and Bybit’s minimum copy amounts, Binance’s spot copy terms and the dates on Bybit’s help pages were not read. The OKX, Bitget and Binance pages were read through a tool that extracts text. How many eToro programme members are in each tier is not published. ZuluTrade, Darwinex and other platforms were not examined.
• Measured slippage. Only the bots’ recommended settings and one backtest were found, with no measurement on real copier accounts. No dataset was found on wallets set up to attract copiers and then sell to them.
• Prices. PeakBot’s share to lead traders, how its $199 and $150 fees combine, which Alertsify price list is current, and the terms of the Telegram and private groups in the collection.
• Refunds after bot incidents. Banana Gun’s promise of refunds is reported in the commentary we read; payment was not confirmed. For Maestro and Unibot we read nothing at source.
• Legal status. Which crypto coins eToro offers in the US now, the outcome of the Australian case, and any reply from eToro to it. The registration or licence status of any account or product in the collection. Rules outside the UK, EU and US. Tax outside the US. Prop firms’ own rules on copying: one firm’s page refused us and another’s does not mention it.
• X’s rules. Read in an archived copy from 7 October 2026, not on the live page. We found no count of accounts suspended for this kind of reply.
• Left out. A news report of followers of one trader losing their money in 48 hours was seen second-hand and concerns private individuals. A survey of copiers’ results was seen only in a search summary. Neither is used.
• Refused pages. Many vendor pages were read through a reader service (r.jina.ai) after direct fetches timed out or returned errors: Binance, Bybit, the bot documentation, PeakBot, eToro, the FCA and CFTC advisory pages, DWF, YieldFund and others. X’s help pages, x.com posts, SSRN and several publisher pages could not be opened at all.
• Whether the people named have seen this page. The people and companies named here were not contacted before publication.

Sources

“Read at source” means we opened the page itself on 10 October 2026. “Secondary” means we read an article or summary about the source, not the source itself. “Estimate” means our own sum. “Through a reader service” means a service fetched the page for us because the site refused our tools directly.

1. Does It Pay collection of 22 tweets on this scheme, collected 2026-10-06; individual tweets linked above. Read from our collected copy on 2026-10-10, not from the live tweets. Counts by label are estimates.
2. Binance Support, “Lead Trader Benefits in Binance Futures Copy Trading”. https://www.binance.com/en/support/faq/lead-trader-benefits-in-binance-futures-copy-trading-ea9bacf82b9e4ddfae50ebc98565241b . Published 2023-08-22, updated 2026-09-22. Read at source on 2026-10-10, through a reader service.
3. Binance, copy trading page (Futures tab) and its featured lead traders. https://www.binance.com/en/copy-trading . Snapshot of 2026-10-10, of which we kept a copy. The list changes continuously and the address may show the Spot tab, so the rows quoted may no longer be there. Read at source on 2026-10-10, through a reader service. Capital figures derived from it are estimates.
4. Bybit Help Center, “Copy Trading Profit Sharing Explained”. https://www.bybit.com/en/help-center/article/Copy-Trading-Profit-Sharing-Explained . Undated page. Read at source on 2026-10-10, through a reader service. Non-crypto rates: https://www.bybit.com/en/help-center/article/TradFi-Copy-Trading-Profit-Sharing-Explained , same route.
5. eToro, CopyTrader page. https://www.etoro.com/copytrader/ . Undated page. Read at source on 2026-10-10. Risk warnings: https://www.etoro.com/copytrader/risk-warnings/ . The five showcased returns were read through a tool that extracts text.
6. eToro, fees page. https://www.etoro.com/trading/fees/ . Undated page. Read at source on 2026-10-10.
7. Trojan, FAQ. https://docs.trojan.com/faqs . Undated page. Read at source on 2026-10-10, through a reader service. Referral system: https://docs.trojan.com/trojan-arena/referral-system.md , read at source on 2026-10-10.
8. GMGN documentation. Fees: https://docs.gmgn.ai/index/gmgn-fees-settings . Copy trade: https://docs.gmgn.ai/index/copy-trade-copy-smart-money-automatically-earn-sol.md . Referral: https://docs.gmgn.ai/index/cooperation-referral-refer-friends-to-earn-rebate-30-rebates-easily-earn-over-usd8000-monthly.md . Undated pages. Read at source on 2026-10-10. Fee totals derived from them are estimates.
9. Axiom, fees. https://docs.axiom.trade/getting-started/fees/axiom-fees.md . Undated page. Read at source on 2026-10-10, through a reader service.
10. PeakBot, pricing: https://peakbot.com/pricing . Page for traders: https://peakbot.com/copy-trading . Home: https://peakbot.com/ . Undated pages. Read at source on 2026-10-10, pricing through a reader service.
11. PeakBot, “Act of Profits” page. https://peakbot.com/act-of-profits/ . Page modified 2026-10-08; strategy start (2026-09-10) and unit size ($3,000) are set in the page’s code, and the live figures did not load. Read at source on 2026-10-10.
12. PeakBot, refund policy: https://peakbot.com/risk-free-policy/ . Disclaimer: https://peakbot.com/disclaimer/ . Undated pages. Read at source on 2026-10-10.
13. Alertsify, home page. https://alertsify.com/ . Undated page. Read at source on 2026-10-10.
14. Alertsify, no-refund policy, version 2.0. https://alertsify.com/no-refund-policy . Effective 2026-03-17. Read at source on 2026-10-10.
15. MQL5, signals list: https://www.mql5.com/en/signals . Rules: https://www.mql5.com/en/signals/rules . Undated pages. Read at source on 2026-10-10.
16. US30 Trading Lab, shop: https://us30tradinglab.com/ . Refund policy: https://us30tradinglab.com/policies/refund-policy . Undated pages. Read at source on 2026-10-10.
17. Polymarket documentation, fees. https://docs.polymarket.com/trading/fees.md and https://docs.polymarket.com/polymarket-learn/trading/fees . Undated pages. Read at source on 2026-10-10.
18. Polymarket documentation, referral programme. https://docs.polymarket.com/programs/referral-program.md . Terms effective 2026-05-28. Read at source on 2026-10-10. The tweets’ short links were resolved the same day. The page does not describe the ?via= tag.
19. Polymarket documentation, geographic restrictions. https://docs.polymarket.com/polymarket-learn/FAQ/geoblocking . Undated page. Read at source on 2026-10-10.
20. Sergeenkov, Polymarket profitability analysis. https://sergeenkov.com/polymarket-profitability/ . 2026-04-06, data to 2026-04-01. Read at source on 2026-10-10. Its statement on Polymarket’s referral rate is secondary and predates the terms in source 18.
21. DWF Ventures, “The thesis on social trading”. https://www.dwf-labs.com/research/the-thesis-on-social-trading . 2026-08-28. Read at source on 2026-10-10, through a reader service. The data it cites was not opened.
22. Four Pillars, “The dawn of public PnL”. https://research.4pillars.io/en/research/the-dawn-of-public-pnl . 2026-09-09. Secondary, through a reader service.
23. Yellow.com, report on the Fomo data. https://yellow.com/news/fomo-copy-trading-94-percent-wallet-losses . 2026-08-28. Secondary, through a reader service.
24. Decrypt, report on pump.fun wallets. https://decrypt.co/300403/pump-fun-traders-millionaires . 2025-01-10, updated after publication with the co-founder’s comments. Secondary.
25. YieldFund, “Is copy trading profitable? A 90-day multi-exchange study”. https://yieldfund.com/is-copy-trading-profitable-a-90-day-multi-exchange-study/ . Dated November 2025 in the article text (the page’s own data shows 2022); data gathered October 2025. Read at source on 2026-10-10, through a reader service.
26. Bitget, copy trading report post. https://www.bitget.com/blog/articles/bitget-copy-trading-report . 2023-08-03. Read at source on 2026-10-10.
27. Finance Magnates, Followme mid-year report 2025. https://www.financemagnates.com/thought-leadership/followmes-2025-mid-year-report-reveals-global-retail-trends/ . 2025-07-17. Secondary.
28. Dorfleitner and others, Quarterly Review of Economics and Finance, volume 70, 2018. https://ideas.repec.org/a/eee/quaeco/v70y2018icp160-171.html . Abstract read at source on 2026-10-10, through a reader service; full paper not read.
29. American Association of Individual Investors, summary of the same study. https://www.aaii.com/journal/article/returns-for-social-trading-platforms-are-negative . 2018-06-01. Secondary, through a reader service.
30. Apesteguia, Oechssler and Weidenholzer, “Copy trading”, discussion paper. https://d-nb.info/1192450213/34 . 2018-06-27. Read at source on 2026-10-10. Later published in Management Science, 2020, https://doi.org/10.1287/mnsc.2019.3508 (abstract only, secondary).
31. ESMA, “Supervisory briefing on copy trading”, ESMA35-42-1428. https://www.esma.europa.eu/sites/default/files/2023-03/ESMA35-42-1428_Supervisory_Briefing_on_Copy_Trading.pdf . 2023-03-30. Read at source on 2026-10-10.
32. Financial Conduct Authority, “Copy trading”. https://www.fca.org.uk/firms/copy-trading . First published 2015-05-12, last updated 2026-07-27. Read at source on 2026-10-10.
33. CFTC, customer advisory on AI trading bots. https://www.cftc.gov/LearnAndProtect/AdvisoriesAndArticles/AITradingBots.html . Announced 2024-01-25; the page itself shows no date. Read at source on 2026-10-10.
34. CFTC, press release 8696-23 (Mirror Trading International). https://www.cftc.gov/PressRoom/PressReleases/8696-23 . 2023-04-27. Read at source on 2026-10-10.
35. CFTC, press release 8770-23 (SignalPush). https://www.cftc.gov/PressRoom/PressReleases/8770-23 . 2023-08-29. Read at source on 2026-10-10. Commissioner Pham’s dissenting statement of the same date: https://www.cftc.gov/PressRoom/SpeechesTestimony/phamstatement082923 , read at source on 2026-10-10.
36. National Futures Association, who has to register as a commodity trading advisor. https://www.nfa.futures.org/registration-membership/who-has-to-register/cta.html . Undated page. Read at source on 2026-10-10.
37. SEC, investor publication on auto-trading. https://www.sec.gov/investor/pubs/autotrading.htm . 2009-06-08. Read at source on 2026-10-10.
38. SEC, press release 2024-125 (eToro). https://www.sec.gov/newsroom/press-releases/2024-125 . 2024-09-12. Read at source on 2026-10-10, through a reader service.
39. Financial Conduct Authority, press release on three people charged. https://www.fca.org.uk/news/press-releases/first-court-appearance-three-finfluencers-charged-fca-led-global-crackdown-illegal-promotions . 2025-09-10, updated 2025-10-13. Read at source on 2026-10-10.
40. ASIC, media release 23-204MR (eToro). https://asic.gov.au/about-asic/news-centre/find-a-media-release/2023-releases/23-204mr-asic-sues-etoro-in-its-first-design-and-distribution-action-to-protect-consumers-from-high-risk-cfd-products/ . 2023-08-03, with later editor’s notes on hearing dates up to April 2026. Read at source on 2026-10-10.
41. X, platform manipulation rules, archived copy. https://web.archive.org/web/20261007195718/https://help.x.com/en/rules-and-policies/platform-manipulation . Captured 2026-10-07. Read at source on 2026-10-10 in the archive.
42. CertiK, report on Maestro and Unibot. https://www.certik.com/blog/maestro-and-unibot . 2023-11-02. Read at source on 2026-10-10.
43. Rekt, commentary on Banana Gun. https://rekt.news/bananagun-rekt . 2024-09-25. Secondary: a commentary site, read on 2026-10-10.
44. StepSecurity, report on Polymarket bot repositories. https://www.stepsecurity.io/blog/malicious-polymarket-bot-hides-in-hijacked-dev-protocol-github-org-and-steals-wallet-keys . 2026-03-15. Read at source on 2026-10-10.
45. UK Financial Ombudsman Service, decision DRN-4467717. https://financial-ombudsman.org.uk/decision/DRN-4467717.pdf . 2024. Read at source on 2026-10-10.
46. IRS, digital assets. https://www.irs.gov/filing/digital-assets . Undated page. Read at source on 2026-10-10.
47. Tradecopia, help article on prop firm restrictions. https://docs.tradecopia.com/en/articles/15515313-what-to-do-if-your-prop-firm-flags-or-restricts-your-account-for-copy-trading . 2026-07-22. Read at source on 2026-10-10.
48. eToro Group, annual report on Form 20-F for 2025. https://www.sec.gov/Archives/edgar/data/1493318/000121390026022034/ea0278371-20f_etoro.htm . 2026-03-02. Read at source on 2026-10-10. Shares derived from it are estimates.
49. eToro Group, prospectus. https://www.sec.gov/Archives/edgar/data/1493318/000121390025043558/ea0223534-12.htm . 2025-05-15. Read at source on 2026-10-10.
50. eToro, Pro Investor programme page (formerly Popular Investor). https://www.etoro.com/copytrader/popular-investor/ . Undated page; terms dated 2025-06-04. Read at source on 2026-10-10. Pay figures derived from it are estimates.
51. Hyperliquid, vault data file. https://stats-data.hyperliquid.xyz/Mainnet/vaults . Downloaded 2026-10-10. Read at source on 2026-10-10; all counts and sums are estimates computed from the file.
52. Hyperliquid documentation, for vault leaders. https://hyperliquid.gitbook.io/hyperliquid-docs/hypercore/vaults/for-vault-leaders . Undated page, now titled “For vault leaders (legacy)”. Current vaults page: https://hyperliquid.gitbook.io/hyperliquid-docs/hypercore/vaults . Read at source on 2026-10-10.
53. Subversive ETFs, NANC fund page. https://subversiveetfs.com/nanc/ . Performance to 2026-09-30. Read at source on 2026-10-10, through a reader service. The benchmark figures are from the same page. The two-year index range is an estimate from them.
54. “Resisting Manipulative Bots in Meme Coin Copy Trading: A Multi-Agent Approach with Chain-of-Thought Reasoning”, arXiv 2601.08641, version 3. https://arxiv.org/abs/2601.08641 ; full text https://arxiv.org/html/2601.08641v3 . 2026-02-05. Read at source on 2026-10-10.
55. Pan, Altshuler and Pentland, 2012 conference paper on social trading. https://doi.org/10.1109/socialcom-passat.2012.133 . 2012-09-01. Secondary (abstract only; the abstract does not name the broker).
56. Study of signal providers, Review of Managerial Science. https://doi.org/10.1007/s11846-022-00560-6 . 2022-07-13. Secondary (abstract only).
57. OKX, lead trader profit sharing rules: https://www.okx.com/en-us/help/lead-traders-trader-profit-sharing-rules , updated 2026-08-19. Bitget, profit share tiers: https://www.bitget.com/support/articles/9083311039641 , 2023-12-29. Read at source on 2026-10-10, through a tool that extracts text.

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)