Why Prop Firms Turned Against Third-Party EAs

A marketplace EA sells to three hundred traders. Each of them buys a $100,000 challenge and runs it. To each trader it's one account and one bet.

To the firm's risk desk it's one trader with thirty million dollars.

Prop firms restrict third-party EAs because hundreds of accounts running the same EA behave like a single account. They enter on the same tick, stop out at the same price and ask for payouts in the same week. Per-strategy capital caps, "distinct strategy" rules and source-code requirements all do one job: make the firm size a strategy once, rather than once for every account it has been copied to.

None of this means firms dislike automation. Several allow it outright. What they've stopped accepting is the same automation, many times over, sold as if it were many traders.

What does a risk desk actually look at?

Not your account. The book.

A desk's main screen shows net exposure per symbol: every open position across every account, added up into one live number per instrument. Individual traders barely move it. When news hits, hundreds of people buy EURUSD, but their orders arrive spread over seconds and minutes, at different sizes, with stops all over the chart. The number drifts.

Three hundred copies of one EA don't drift. Most EAs make their decision on the close of a bar, so every copy acts on the first price of the next one. Orders from three hundred logins land within the same second or two, and the net position in that symbol jumps in a single step. On the screen it looks exactly like one client sending one order for the sum of everyone's lots.

How does a firm tell copies from coincidence?

By the combination, repeated. Plenty of people buy the same pair at the same moment, so no single sign gives it away.

What does give it away is the same set of things happening trade after trade:

• Entries bunched within a second or two, on the same symbol, in the same direction

• Identical stop and target distances, to the point

• Lot sizes that are the same fraction of each balance. EAs usually size by risk percent, and prop accounts come in standard sizes, so a 10k, a 50k and a 100k account produce lots that are exact multiples of each other

• The same EA identifier, the magic number an EA stamps on its orders, and often the same order comment

The broker's server stores the magic number and comment with every order. It also records how each order was placed: by hand on the desktop terminal, from the mobile app, from the web terminal, or by an Expert Advisor.

The software firms buy to watch for this compares accounts in pairs, counting how often their trades match inside a short window. Past a threshold, the matched accounts are treated as one cluster, and often as one trader. The vendors describe it openly in their own marketing. PropForge lists copy-trade detection by trade timing and entry and exit pattern matching across accounts. FXPropTech's published guidance says matching symbols and timing alone isn't enough evidence, and that stronger detection also compares stop and target placement, position-size ratios and whole trading histories.

Why the exit is where it costs the firm

The stop or target sits at the same price on all three hundred accounts, so they all close on the same tick.

A losing day is three hundred breaches at once. A winning month is three hundred payout requests at once, all generated by one strategy.

That second case is the one the rules are written for. A firm prices its payouts on the assumption that funded traders' results are independent: some up, some down, and over a month it averages out. Copies don't average. They're one bet placed three hundred times, and when it wins, it wins everywhere.

A per-strategy capital cap measures exactly that. The firm is sizing a strategy, not a trader, however many accounts the strategy has been spread across.

How the rules changed

Firm by firm the wording differs. The target is the same.

Each is a reasonable answer to the same problem, and a firm that restricts automation isn't making a worse product. Its payout model depends on independent outcomes, and third-party automation is the quickest way to lose them. Those firms are simply not compatible with it.

Alpha Capital's definition of a strategy is the most revealing line in the set. On its help centre, the same trader may trade EURUSD and GBPUSD on two accounts "as long as the trades are opened and closed at different times, essentially making them different trade ideas." A strategy is defined by when it trades, not by what it's called.

Why do firms ask for source code?

Because a compiled EA proves nothing about who wrote it.

FundingPips says it directly: a compiled file on its own is not proof of ownership, and it asks for the uncompiled source instead. The5ers requires that the trader own the source code of any EA they run. A source file is evidence that the logic exists in one place, yours, rather than in three hundred downloads from the same marketplace page.

If you built your own strategy, the practical point is to keep the record of building it. That's what the firm will ask for, and it's much easier to produce on the day than to reconstruct afterwards.

What actually makes a strategy distinct?

Its behaviour. Not its labels.

Changing the lot size, the magic number, the order comment or the name changes nothing a risk desk looks at. The entries, exits and sizing logic are exactly what they were, and the account still lands in the same cluster.

What does change it:

• Entry logic and the conditions behind it

• Timeframe

• Filters

• The session you trade

• How the stop and target are placed and managed

There's a simple test. Run your version and the original side by side on demo for a few weeks. If they take different trades at different times, you have a different strategy. If they enter and exit together, you have the same one with a new name.

Adding a random delay or varying the lot size to hide a shared EA doesn't pass that test either. It's the same strategy, smeared slightly, and surveillance compares whole histories rather than single trades. It's also exactly what these rules exist to catch.

Building from a template

On Vectorhelm, every template opens fully editable in the strategy builder: entry and exit conditions, timeframes, filters, size, stops and targets, with AI-assisted building to help turn it into your own version. If a strategy is still identical to its template when you launch it, it's flagged as an unmodified template, and you're asked to make it your own before it trades, with an explanation of why that matters on a funded account.

The flag can tell that something changed. It can't tell that the strategy now trades differently. That judgement is the one in the section above, and it's yours.

Building your own logic also doesn't settle what your firm's EA policy says about the software that executes it. That's a separate question, and the firm is the one to answer it, in writing. The FTMO version of that question is covered in its own article.

The honest limitation

Independent strategies can still collide.

Popular logic, a moving-average cross on the hourly close with a round-number stop, is used by thousands of traders who've never heard of each other. Their entries bunch on the same bar for the same reason three hundred copies of one EA do. At least one vendor's own published guidance says shared timing alone shouldn't count as proof of copying, and that a review score shouldn't fail an account automatically. But no firm publishes where its line sits, so there's no way to know in advance how close is too close.

The more common your logic, the more your account looks like other people's. That's worth knowing before you build a strategy on the first setup every YouTube channel teaches.

Rules change, too. The policies above were checked on each firm's own pages on 8 October 2026, and EA policies are among the most frequently revised rules in this industry.

Before you put an EA on a funded account

If you bought it, assume other people are running it, and read your firm's per-strategy cap and EA rules with that in mind.

Change the logic, not the labels.

Keep proof that you built it.

Ask the firm in writing whether your setup is allowed, and keep the reply with a date on it.

We're building risk enforcement for funded accounts with strategies you build yourself, starting from templates that are meant to be changed. It isn't open yet. Early access is here if you want it when it is.

Vectorhelm is execution software. It runs the rules you define on accounts you own and control. Nothing here is trading advice, and no configuration of risk limits prevents losses. Prop firm rules change; verify against each firm's own published terms before relying on anything here.