EAs and copy trading: what passes, what costs you an account

The general rule fits in one sentence: a robot is accepted if it belongs to you and executes your strategy, and refused as soon as it exploits the broker’s infrastructure rather than the market. Copy trading between your own accounts at one firm is often tolerated, sometimes capped, almost always declarable. Copying a third-party signal followed by dozens of traders is the case most frequently penalised, and it is penalised at payout time. Firms monitor correlation between accounts because it is the heart of their risk: a hundred accounts taking the same position in the same second turn individual payouts into a portfolio exposure.

Three practices the rules treat differently

The phrase “copy trading” is used for very different situations. Telling them apart is the first step.

The personal EA. An algorithm you developed, bought or configured, running on your account and executing your logic. This is the most widely accepted case. The great majority of firms allow it explicitly, provided it does not fall into the prohibited categories described below.

Copying between your own accounts. You hold several accounts at the same firm and a master replicates its orders onto the others. The firm’s position varies here: some allow it without reservation, some limit it to a number of accounts or a total capital, some prohibit it outright. The motive is not moral, it is accounting: the firm calculates its exposure per trader, not per account.

Following an external signal. A signal provider, a Telegram channel or a copying service broadcasts orders that several clients replicate on prop accounts, often at the same firm. This is the most penalised situation, and often the one the trader considers the most innocent. Whether you are dealing with a long-standing firm like FTMO or a more recent one like FundingPips, the question to ask is the same: do the rules explicitly allow this configuration, or do they merely fail to prohibit it?

What is almost always prohibited

One family of strategies recurs in every rulebook, whatever the firm, because it exploits how the platform works rather than the market.

• Latency arbitrage. Exploiting the lag between the firm’s price feed and a faster one. Prohibited everywhere without exception, and often described as fraud in the terms and conditions.
• Tick scalping and HFT. Positions of a few seconds aiming to capture micro quote differences. Many firms set a minimum holding time, sometimes expressed in seconds, to exclude this behaviour.
• Exploiting wicks and quote gaps. Orders placed to be filled on an aberrant price in the feed.
• Hedging across accounts. Opposing positions of equivalent size spread over several accounts at the same firm, or between a firm and a live account. The strategy is not seeking a market profit but a statistical pass. It is the industry’s most documented ground for closure.
• Grid and martingale without a risk limit. Not always prohibited, but frequently framed by a limit on open positions or total exposure.

One point deserves emphasis: a trader can fall under these clauses without intent. An EA bought on a marketplace that opens thirty positions an hour with two-second holding times breaches a minimum-duration rule even if its user never read one.

Why copying between accounts is the real subject

A firm manages a risk portfolio. As long as its traders take independent decisions, some traders’ gains statistically offset others’ losses and the model works. The day three hundred funded accounts place the same order in the same second because they follow the same signal, the firm finds itself with a massive directional position it neither chose nor hedged.

That is why rulebooks rarely distinguish the “good” copier from the bad. From a risk standpoint, a high-quality free signal followed by five hundred people is more dangerous than a poor signal followed by three. The penalty does not measure performance, it measures correlation.

How firms detect it

The means are finer than many traders imagine, and they apply retroactively across the account’s whole history.

• Millisecond timestamps. Two accounts opening the same pair in the same direction hundredths of a second apart, repeatedly, are statistically linked. A manual delay never produces that profile.
• Constant lot ratio. A proportional copier leaves an obvious signature: sizes always vary in the same ratio between accounts.
• Correlation across trade series. The analysis does not cover one isolated order but the full distribution of entries and exits, losing trades included.
• Technical fingerprint. IP address, device identifier, platform serial number, time zone, connection data. A VPN does not suffice when the rest lines up.
• KYC identity. The final verification links accounts to a natural person. An account opened in a relative’s name to get round a limit is treated as documentary fraud, which is more serious than a trading rule breach.

These checks are almost always triggered at the withdrawal request, not during the evaluation. A trader can therefore go several months without an alert before the whole history is examined at once.

What that looks like on a payout

The sequence is mechanical. The trader requests their withdrawal. The compliance review analyses the history, identifies a correlation with other accounts or an execution profile incompatible with manual trading, and suspends the request. Depending on the terms and conditions, several outcomes follow.

The payout is refused and the account closed, the challenge fees staying with the firm. The trader receives a standard notification citing a general clause, with no detail on the evidence relied upon — most firms do not disclose their detection methods.

In some cases the terms and conditions provide for the clawback of payouts already made where an earlier breach is established. That clause exists more often than people think, and it turns an account problem into a financial one.

Recourse is limited. The contract generally provides for internal arbitration, a distant jurisdiction and a short window to contest. Against a statistical analysis, the burden of proof is hard to reverse.

What to do

• Read the EA section of the rules before buying, not after the first trade. Look for three things: the prohibited categories, the minimum holding time, and the treatment of multiple accounts.
• Declare what must be declared. Several firms require prior declaration for using an algorithm or holding several accounts. The declaration protects you; silence is taken as concealment.
• Never share signals with other traders at the same firm. Even without an automatic copying tool, simultaneous manual execution produces a detectable correlation.
• Avoid mass-marketed EAs. An algorithm sold to hundreds of buyers creates exactly the correlation profile the firm is looking for, even if each user acts independently.
• Do not open an account in someone else’s name. It is the only decision on this list that moves you from a commercial dispute to a serious problem.
• Check what happens to the permission between phases. An EA tolerated during the evaluation is not necessarily accepted on the funded account, where the firm sometimes genuinely hedges positions.

The useful question is not “will I be detected”, but “will my configuration look like a breach when someone examines a hundred trades at once”. That is the gaze that decides the payout.