What a crypto signal channel's win rate counts

How it works

A call is a post. It names a pair, a side, an entry zone, a ladder of targets, a stop and often a leverage, and it goes out to every member of a Telegram or Discord channel at once. Later the same channel posts the outcome: a target reached, a stop hit, or nothing at all. The win rate in the channel's bio is a tally of those outcome posts, kept by the person who sells the subscription.

Nothing in that chain is tied to an order. The call is text, the outcome is text, and whether anyone, the seller included, traded it is not recorded anywhere a subscriber can see.

The channel itself is an editable archive. Telegram's Channels FAQ says the owner "can broadcast messages, delete any messages", and that "if a message is deleted in a channel, it will disappear for all subscribers." Posts carry the channel's name, not the author's, and "if the signatures are disabled, there is no way for users to know whether a message was posted by the owner or one of the administrators." So the history a prospective buyer scrolls through before paying is the set of posts its owner chose to leave, written by people the buyer cannot identify.

Four definitions that move the number

None of these needs a false post. Each is a choice about what to count, and a channel rarely says which choice it made.

1. Which target is a hit. Take a long call with three targets at +2%, +5% and +10% and a stop at −8%. This example is ours. Price touches the first target, turns, and runs down to the stop. A channel that scores a call as won when the first target is touched records a win. A member who held the whole position for the last target lost 8% of it, and one who closed a third at each target still lost on the remainder. The same price path is a win or a loss depending on a rule the buyer was never shown.

2. Counted by trade, not by size. Where a venue does publish its win-rate formula, it counts positions. Binance's copy-trading FAQ defines it as "Number of Profitable Closed Positions / Total Number of Positions", and adds that a position closed only in part "doesn't count as a closed position". Count-based means size-blind: nine calls that each make 1% and one that loses 20% are a 90% win rate and a loss of about 11% on an equal stake per call. In a channel, a call with no stop that is still underwater is not a loss until somebody posts that it is.

3. Leverage inside the percentage. Binance's futures calculator gives "ROI% = PnL / Initial Margin", with the initial margin rate "IMR = 1 / leverage". The return is measured against the margin, not the position, so a 1% move in the price of a perpetual at 20x reads as 20%, and "+300%" can be a 15% move at 20x. The same multiplier works the other way: a 5% move against that position is the whole margin, and the venue's liquidation rules close it before it gets there. A percentage with no leverage beside it cannot be compared with anything.

4. Adding instead of compounding. Returns from separate periods multiply. A period of +25% then one of −20% leaves an account where it started, since 1.25 × 0.80 = 1.00; added together they read as +5%. Exchanges do not all chain periods the same way either. Bitget's rule for copy-trading ROI, dated 15 August 2023, is "Total ROI = current ROI + sum of historical ROI per period", where a period restarts at each transfer, and beginning assets under 200 USDT are treated as 200 USDT. Bitget changed how position-level ROI is computed in a notice of 24 May 2024, which does not say it replaces the account-level rule, and both pages describe themselves as time-sensitive. Binance, by contrast, divides PnL by a maximum base balance of deposits net of withdrawals. A "total ROI" from one venue is not the same quantity as one from another.

What a venue-computed profile shows

A copy-trading profile on an exchange is a different kind of evidence. The trades are fills in a real account and the venue, not the trader, computes the figures under a formula it publishes. On Binance's futures copy trading, as its FAQ describes it:

• PnL is "Realized + Unrealized Profit or Loss", so a losing position held open shows up here even though it is not yet in the win rate.
• ROI is PnL divided by MaxBaseBalance, where base balance is the initial deposit plus net deposits; the page says all existing lead portfolios were recomputed this way from 12 November 2025.
• Maximum drawdown is measured on ROI from its historical peak.
• The Sharpe ratio is annualised with a 0% risk-free rate and "will not be displayed if the portfolio has been trading for less than 30 days".
• ROI and PnL are updated hourly.

What a profile cannot show is everything outside that one account. The venue does not see the trader's other accounts or other exchanges. It does not show whether the calls in a paid channel are the trades in the profile, made at the same time, or a selection from them. And a copier's own orders follow the leader's, so their fills differ by whatever the price did in between: the slippage that a leader's ROI never contains.

Some sellers offer a third-party tracker that reads their exchange account through an API key. That proves the trades in the keyed account and nothing about the others; what such a key can and cannot reach on each venue is set out in what your exchange API key lets a trading bot do.

Where the call sends your money

A signal can earn its seller money in ways that have nothing to do with whether it is right. An exchange sign-up link in the channel description is the plainest: the referral mechanics are in who pays for your free trading bot.

The costlier pattern is a group that chooses the venue for you. In its complaint against Morocoin Tech Corp. and others, filed on 22 December 2025 (LR-26453), the SEC alleges that defendants operated "so-called investment clubs using WhatsApp" with "supposedly AI-generated investment tips", steered members to trading platforms where "no trading took place", demanded "advance fees" from victims, and took at least $14 million. These are allegations in a pending case. The CFTC's messaging-app advisory of 31 October 2024 describes the same setting from the other end: groups that talk of "trading crypto futures with leverage" and "cooperative trading projects", which it calls "also called pump-and-dump schemes", with "testimonials from other group members" that are "all fake".

Neither is a case about a paid signal channel that simply sells calls. We found no US or UK enforcement action that charges one as such. That absence is itself the point for a buyer: nobody outside the channel has checked the numbers in its bio.

What a regulator says about the claims

The CFTC's advisory on AI trading scams, issued on 25 January 2024, names "algorithms that provide buy and sell signals to subscribers" among the products scammers promote, and claims that they "yield 100 percent 'win' rates" among the signs. Its case study is Mirror Trading International, where the CFTC says the operator "created fake customer accounts and balances using MetaTrader demo accounts".

For futures advisers inside its rules, the CFTC prescribes what a hypothetical record has to say. Under 17 CFR 4.41 the statement begins "These results are based on simulated or hypothetical performance results that have certain inherent limitations", says the results "do not represent actual trading", and that such programs are "designed with the benefit of hindsight". It must sit "in immediate proximity" to the figures, and the rule applies to material on the internet and "regardless of whether the commodity pool operator or commodity trading advisor is exempt from registration". NFA Rule 2-29 goes further for members: actual results must be "representative of the actual performance for the same time period of all reasonably comparable accounts" and net of all fees, and a member may not show hypothetical results for a system once it "has three months of actual trading results for that system".

Whether any of that reaches a given channel depends on what it advises on and where it operates. The CFTC's registration check page says venues offering spot trading in virtual currencies "are not required to be registered with the CFTC", while the agency keeps "general anti-fraud and manipulation enforcement authority" over those markets. The rules are useful to a buyer as a yardstick either way: a seller who would fail them is showing a record that a supervised adviser could not.

Warning lists help less than they look. The CFTC says inclusion on its RED List "does not mean that the CFTC or a court has concluded that a violation" occurred, and tells readers to check registration "even if the entity you are considering is not on the RED List". In the UK, the FCA said on 24 April 2026 that of 1,267 illegal financial adverts it found on Meta platforms during a week of action, 66% came "from firms or individuals already on the" FCA's Warning List. Being listed did not stop them advertising, and not being listed shows nothing.

What it costs

The subscription price is printed; whether you can get it back depends on how you paid it.

Sources: Whop's Resolution Center and disputes pages, and Telegram's Stars terms. Whop's rules for sellers also require that "any direct or implied earnings claims must be accurate, substantiated, and not misleading" and that testimonials "disclose atypical results", which gives a buyer something specific to report a channel's bio against.

The larger cost is not the fee. It is the position size you put behind calls whose record you could not check, at the leverage the call suggested.

What you can do about it

Save every call when it arrives. A screenshot or a forwarded copy with the time on it, in a place the channel cannot reach. Deletion in a channel is complete, so this archive is the only record of what was actually sent, and it is the one thing a later dispute will need.

Ask for the definitions before you pay. Which target counts as a win? Are stopped-out and still open calls in the count? What leverage is the percentage at? Are period returns added or compounded? A channel that cannot answer in writing is quoting a number it has not defined.

Recompute without leverage. Divide the advertised percentage by the leverage in the call, then count each call at the last target with the stop as posted. That is the move in the market, which is the only figure two channels can be compared on.

Prefer a profile the venue computes, and read its formula. If the seller has a copy-trading profile, check how long it has existed, its maximum drawdown and its leverage, and read that venue's definitions rather than assuming they match another's. Remember it is one account.

Check a tracker's key before you trust it, and before you hand one over. A read-only key proves less than a tracker's badge suggests, and a key with trading rights proves nothing more while risking more.

Execute on a venue you chose. A group that names the exchange, the app or the wallet you must use is the pattern in the SEC's Morocoin complaint, and a demand for a fee before a withdrawal is its last step.

Pay in a way that can be reversed. On Whop, a card keeps the 120-day case window and the chargeback; crypto keeps neither. Stars are final unless the developer refunds.

Keep your own fills. If you run a channel's calls through a bot or a terminal, its order history is the record of what you earned. Altrady has a trade journal, and signal bots such as WunderTrading and 3Commas turn an alert into an order on your own exchange account, where the exchange's history records the fill.

Report it where a rule applies. An earnings claim that breaks Whop's rule goes to Whop. Money taken on a platform where nothing traded is fraud to report to the police and, in the US, the FBI's IC3; in the UK, to the FCA.

Expect a second approach. Someone who lost money to a group is the target of the next pitch: an offer to recover it for a fee. How that works, and what US and UK authorities say about it, is in why a crypto recovery offer asks for a fee first.