Your EA waits for a London sweep, a displacement candle and a return into the gap it left behind. It places maybe two orders a day. Nothing about it is fast, and nothing about it is exotic. Yet the FTMO rules page open in your other tab never mentions liquidity, structure or order blocks — it talks about ultra-high-speed tools, mass data entry, hyperactive robots and simulated trades that do not resemble real markets.

That vocabulary gap is the entire problem. FTMO's rules are written in the language of the abuse cases they exist to stop, not in the language of the model you actually trade. So this page translates them: what FTMO genuinely permits, the specific behaviors it restricts, and where an ICT-style expert advisor sits against each one — close enough to the rule text that you can audit your own bot instead of guessing.

Key Takeaways
  • FTMO permits expert advisors on the evaluation and the funded account; there is no strategy whitelist and no ICT-specific clause.
  • The restrictions describe behaviours — stale-price exploitation, extreme speed, hyperactive server traffic, offsetting positions, disproportionate sizing — which a closed-candle structure model clears by construction.
  • Three rules only switch on once you are funded: the two-minute news window, the weekend-close requirement, and the Standard-versus-Swing split.
  • The most common EA-specific traps are tick-by-tick stop modification, a resting stop inside a news window, and one bought EA stacked across accounts past the $400,000 capital cap.
Table of Contents (22 min read)Contents

Does FTMO Allow Expert Advisors?

Yes. FTMO permits algorithmic trading and expert advisors — on the evaluation phases and on the funded account alike. There is no strategy whitelist to get on, no application to file, and nothing in the rule set that treats a Smart Money or ICT model differently from any other automated system. The algo-trading permission question, which is a genuine blocker at some firms, is simply not the question at FTMO.

Essential answer

Can you use an expert advisor on FTMO?

Yes — FTMO permits algorithmic trading and EAs on both the evaluation and the funded account. What it restricts is a list of behaviors, not the fact that a machine sends the orders.

Evaluation phases EAs allowed
Funded FTMO Account EAs allowed
Strategy whitelist None - behavior-based rules
Third-party or bought EA Allowed, capital-capped
Open orders at once 200 platform ceiling
Server requests per day 2,000 hyperactivity line

Paraphrased from FTMO's published FAQ, trading objectives and forbidden-practices pages - re-check the current wording before you rely on it.

The answer to the headline question, before any of the conditions attached to it.

What "allowed" does not mean is "unexamined". Five separate things can still catch an automated account, and only one of them is about strategy at all:

  1. Behaviors on the forbidden-practices list — written as descriptions of conduct, not as strategy names.
  2. Hard platform ceilings — a cap on how many orders you can hold and how much traffic your EA may generate.
  3. The news-release window — which applies to some phases and account types and not others.
  4. The loss limits — measured on equity, and reset on a clock your EA probably ignores.
  5. The capital-allocation cap — the one that quietly punishes running a popular bought EA.

Work through those five and you have your answer for your specific bot. The rest of this page is that walk-through, with a self-audit checklist at the end that maps each rule back to a behavior you can actually check in your code.

The Behaviors FTMO Restricts — Not the Tool Itself

Two identical sealed trading machines at a glass scanning arch: the calm one passes through, the one throwing off chaotic sparks is stopped.
Identical machines, opposite outcomes — the restriction lands on behaviour, not on automation itself.

FTMO's forbidden-practices list describes conduct. Read in EA terms, it prohibits:

  • Trading off broken or stale prices — strategies that exploit display errors or delays in price updates. This is the latency-arbitrage and stale-quote family.
  • Opposite positions used as a device — entering offsetting trades, on one account or across several accounts and providers, to manufacture a result rather than to trade.
  • Ultra-high-speed tools and mass data entry — software, automation or speed advantages of a kind a normal market participant could not have.
  • Gap trading around scheduled events — deliberately opening into a scheduled release, an earnings print, or immediately before a market closes for a long break.
  • Hyperactivity — an EA that drives the account past roughly two thousand server requests a day.
  • Manufacturing an even profit curve without spreading the risk — hedging or holding opposing positions on the same or highly correlated instruments so profit lands on more days than the risk did.
  • Disproportionate sizing and concentrated exposure — one trade, or a run of trades, that is far larger or far more numerous than the rest of your activity.
  • Third-party access — someone else trading your account, or you trading someone else's.

Two hard ceilings sit alongside that list: the platform accepts around two hundred orders open at any one time, and it stops accepting order and modification messages beyond the hyperactivity threshold. Neither is a judgement call — they are limits your EA either respects or hits, which is why a max open trades limit belongs in the code rather than in your intentions.

Here is the part worth saying plainly, because no rules page will say it for you: a structure-based ICT EA clears almost all of this by construction. It waits for a candle to close before it decides anything, so it cannot exploit a stale quote. It trades a handful of setups inside a killzone, so it is nowhere near a high-frequency profile. It has no reason to hold both sides of a pair, and no reason to touch anyone else's account. The bans are aimed at a different animal.

One behavior does bite, and it is the one nobody expects: stop management. An EA that nudges a trailing stop on every tick, or re-writes its take-profit each time price moves a pip, generates modification messages continuously. Twenty positions doing that will produce more traffic than a hundred clean entries — and once the terminal starts queuing those requests you also start seeing trade context busy errors, which is your own platform telling you the same thing FTMO's rule does. Move stops on bar close, not on tick, and the problem disappears.

Audit your own bot

Will your EA trip the 2,000-request-a-day ceiling?

Set how your EA actually behaves - symbols, trades, and how often it moves a stop - and read the daily request load that behaviour produces.

Symbols the EA trades
Trades per symbol per day
Stop or target moves per trade
Pendings placed or cancelled per trade
Trades opened per day
Share of the daily ceiling
Requests left before the cap
Push the stop-moves slider toward tick-by-tick trailing and watch a two-trade-a-day EA turn into a hyperactive one.

The Two-Minute News Window Your EA Can Breach Without Opening a Trade

On a funded FTMO Account of the Standard type, trades on instruments tied to a scheduled high-impact release may not be opened or closed inside a window that begins two minutes before the release and ends two minutes after it. Instruments unrelated to that release are unaffected — a euro-sterling position is not restricted by a US employment print, while dollar and sterling pairs are.

Three details decide whether this rule ever touches your EA:

  • It does not apply during the evaluation. The Challenge and Verification phases carry no news-window restriction. Passing with news exposure and then keeping the same settings once funded is exactly how automated accounts get caught.
  • The Swing account type is exempt. It carries no news-release restriction at all.
  • A resting stop counts as a trade. If a stop-loss or take-profit triggers inside the window, that is treated as closing a trade inside the window. Your EA does not have to do anything for this to happen. A position opened two hours earlier, with its protective stop sitting exactly where the release will spike, is the breach.

That third point is the one that matters most for an ICT model, because ICT models leave stops in the obvious places — under the low that was swept, above the high that was raised — which is precisely where a release-driven wick goes hunting. A compliant setup therefore needs a news filter that does two jobs, not one: block entries in the window, and clear or widen exposure that is already open before the window arrives. Building that mechanism properly is its own subject; the point here is only that the rule reaches a passive position, not just an active order.

Separately, and independently of the two-minute window, the forbidden-practices list rules out strategies whose edge is the scheduled event — opening into a release to capture the gap. An EA trading confirmed structure that happens to fire near an event is a different thing from an EA that waits for the event. Keep them distinguishable in your logs and that distinction stays obvious to a reviewer.

The Loss Limits Your EA Is Measured Against

FTMO's evaluation and funded accounts run on two loss limits, both checked continuously. On the two-step route the daily limit is five per cent of the initial capital and the overall limit is ten per cent, calculated once and fixed. On the one-step route the daily limit is three per cent and the overall limit trails upward with your best end-of-day balance instead of staying static. Which route you bought therefore changes the arithmetic your EA has to respect, even though the marketing describes both as "the FTMO Challenge".

Two mechanics matter more to an automated system than the headline percentages:

They measure equity, not balance. The check includes floating profit and loss on open positions, plus swaps and commissions. A drawdown that exists only on paper still breaches — which is the practical difference between balance and equity that catches every trader who assumed an unclosed trade is not yet a loss.

The daily limit resets on a clock, and the clock is not your broker's. The daily allowance is recalculated at midnight Central European time from the balance recorded at that moment. An ICT position held through the roll therefore carries its floating loss into the new day without having reduced the balance it is measured from — so a position three per cent underwater at midnight starts the next day with only two per cent of daily room left, on a limit that has just "reset". Any EA that holds overnight needs that arithmetic in its risk check, or it will breach a limit it believes it just refreshed. If you want to see the shape of it before you code it, run your own numbers through the prop-firm drawdown calculator.

Automating the enforcement of those limits — flattening at a threshold, sizing back after a losing sequence — is a whole discipline of its own and is not what this page is for. Treat the numbers above as the compliance facts your EA must be built around, and note that every figure here describes rules, not results; trading a leveraged account carries substantial risk, as our risk warning sets out in full.

The Rule Set Changes With the Phase and the Account Type

Most "FTMO EA rules" write-ups flatten the firm into a single rule set. It is not one. The same EA can be fully compliant during the evaluation and in breach two weeks later on the funded account, without a single line of code changing, because three of the rules only switch on once you are funded — and one of them switches off again if you hold a Swing account.

Same EA, different rule set
RuleChallenge & VerificationFTMO Account - StandardFTMO Account - Swing
Expert advisors Allowed Allowed Allowed
Max daily loss 5% two-step route, 3% one-step route Same Same
Max overall loss 10%, static on the two-step route, trailing on the one-step route Same Same
News-release window No restriction No open or close 2 min either side, targeted symbols only No restriction
Weekend and long-rollover holding Allowed Positions must be closed Allowed
Forbidden-practices list Applies Applies Applies
Two of the four rules that can end a funded account do not exist during the evaluation - which is why passing proves less about compliance than traders assume.

The weekend row deserves its own sentence, because it decides whether a whole category of ICT model fits the Standard account at all. On a funded Standard account, positions have to be closed before the market closes for the weekend and before any rollover break long enough to matter. A model that takes a Tuesday setup targeting the weekly high, or one that holds a daily-bias trade across Friday, is not a rule-breaker on the Challenge and becomes one the moment it is funded. The Swing account exists precisely for that trader, and choosing it is a configuration decision about your EA's holding period, not a judgement about the firm.

One more phase-dependent rule to file away: the one-step route adds a best-day condition, requiring that no single day contributes more than half of your total profitable-day gains. That is a consistency constraint rather than an EA-permission constraint, and it is a subject in its own right — flagged here only so you know which route it attaches to.

The $400,000 Cap and the Shared-EA Problem

FTMO limits the total simulated capital allocated to one trader — or to one strategy — to four hundred thousand dollars across all accounts, before any scaling, with equivalents in the other base currencies. You may hold as many accounts as you like; it is the combined capital that is capped, so four accounts at one hundred thousand each reach the ceiling exactly as two at two hundred thousand do.

Four identical glass account panels showing the same trade, fed by one cartridge, with a chrome ceiling bar hanging just above them.
The cap follows the strategy, not the account — which is what makes a popular bought EA a capital-allocation risk.

The words or one strategy are what make this an EA rule rather than an accounting rule. If identical trading is detected across multiple accounts and the combined capital exceeds the cap, those accounts can be suspended — and FTMO's own guidance warns that traders using a third-party EA risk being denied a funded account for exactly this reason. Two independent buyers of the same popular ICT expert advisor, running default inputs on the same pairs in the same killzone, produce trade timing so similar that it reads as one strategy operating above the cap.

This is the one restriction where a well-behaved, entirely compliant EA can still cost you an account, so treat it as a portfolio question:

  • Count your own capital first. Every open FTMO account, evaluation or funded, counts toward the same ceiling.
  • Do not stack the same EA to get past the ceiling. Spreading one strategy across extra accounts to trade more capital than the cap allows is the specific pattern the rule targets.
  • Resist the urge to "differentiate" artificially. Randomising entries or shuffling parameters purely to look different is closer to circumvention than to compliance, and it degrades the model you spent months building.
  • Use the scaling route for more capital. Growing an account through FTMO's own scaling plan is the sanctioned path past the starting ceiling.

Structure-Based Re-Entry vs. Martingale: Where an EA Gets Flagged

Neither martingale nor grid systems appear on FTMO's forbidden list by name. They are judged instead against the standard that governs everything else: trading must be reasonably replicable in a real market and consistent with the risk management a reasonable person would apply to their own money. In practice that standard is written as three observable behaviors — position sizes far larger than your other trades, position counts far higher than your other activity, and repeated exposure stacking into one symbol.

Read those three back and the exposure becomes obvious for ICT models specifically. An EA that re-enters after a failed sweep — second attempt, then a third, each one a "better" price on the same idea — is doing something structurally different from martingale, but it can leave the same fingerprint in a trade log: several positions, one symbol, one direction, opened while the account was losing.

Structure re-entry vs martingale recovery

Winner

Structure re-entry

  • Each entry is a fresh setup: a new sweep, a new displacement, a new invalidation level
  • Risk per trade stays constant - the third entry is the same size as the first
  • A hard cap on concurrent positions bounds exposure in any one symbol
  • Every position carries its own stop, so the model can be wrong and simply stop trading

Reads as one model repeating itself, which is what replicability asks for

Martingale recovery

  • Each entry exists to repair the last loss, not because a new setup formed
  • Size escalates after every loss, so the final trade dwarfs the first
  • Exposure concentrates in one symbol exactly when the thesis is failing
  • Survival depends on the loss limit not arriving before the recovery does

Not banned by name, but it collides with the sizing and exposure standards

The difference a reviewer can see is sizing and invalidation - not what the strategy is called.

Four code-level choices keep the distinction visible from the outside: fixed risk per trade with no post-loss escalation, a hard ceiling on concurrent positions, an exposure cap that treats correlated pairs as one bet, and a distinct magic number per model so the log reads as one system doing one thing. None of those is a compliance trick — they are the same choices that keep the model inside the loss limits anyway.

Flagged for Review, or Hard Breach?

Not every trigger has the same consequence, and no competing page seems willing to say which is which. FTMO's own language separates three tiers, and it is worth knowing which one you are looking at before you panic about a rule.

Tier one — an adjustment request. The softest outcome, and the usual response to server-side friction: FTMO may ask you to modify your EA's parameters or your strategy so it stops overloading the platform. Nothing is lost; you change a setting. Hyperactivity typically lands here first.

Tier two — corrective action on your result. Trades can be removed from the account history, the account can be rebalanced, leverage reduced, or platform access restricted. Your account survives; the numbers on it may not. Practices that produced an unearned result — offsetting positions, exploiting a bad price — sit here.

Tier three — termination-class. Disqualification from the evaluation, forfeiture of any pending reward, and termination of the agreement. This is where deliberate breaches of the forbidden-practices list end up, particularly manipulation, third-party trading and circumventing the capital cap.

Sitting outside all three is the one that is not a review at all: breaching the daily or overall loss limit fails the account automatically. There is no reviewer, no adjustment request, and no appeal to intent — which is why a kill switch that halts the EA at your own threshold, comfortably inside FTMO's, is worth more than any amount of rule-reading.

Is Your ICT EA at Risk? A Self-Audit

Run your own bot down this list. Every item maps to a specific rule above and is answerable from your code, your settings or one look at your trade log — no interpretation of terms required.

Before you start the challenge

Audit your ICT EA against FTMO's rules

0 / 12

Checklist complete — you’re cleared to proceed.

Twelve checks, each answerable from your code or your log - a clean pass here is what 'compliant' actually means.

If you cannot answer one of these from your own settings, that is the item to fix first — an unknown behavior is the same risk as a known bad one, because the rules are enforced against what your account did, not what you believed it would do.

If FTMO's Rules Don't Fit Your EA

Three genuine mismatches exist, and none of them is a failing on your part.

The first is a latency-sensitive model — anything whose edge lives in execution speed rather than structure — which runs directly into the tools-and-speed clause no matter how it is configured. The second is a holding-period mismatch you cannot solve with the Swing account. The third is capital: a strategy you want to run above the four-hundred-thousand ceiling has nowhere to go at FTMO, and stacking accounts is the pattern the rule is watching for.

If any of those is you, the question stops being an FTMO question and becomes a survey question — which prop firms allow EAs beyond FTMO, and on what terms — followed by the narrower one of choosing an alternative prop firm for an MT5 EA that suits your specific holding period and capital needs. Both are worth answering with the same rule-text scepticism you have just applied here, because "EAs allowed" means something different at every firm.

Passing the Challenge With Your ICT EA

Compliance is the floor, not the plan. Knowing your EA will not be flagged tells you nothing about whether it will hit a profit target inside a drawdown corridor it has never been tested against — and that is a different body of work: sizing for the daily line, surviving the phase transition, and keeping the account alive once real payouts depend on it. Passing a prop firm challenge with an ICT EA starts from the audit above and continues well past it.

You arrived asking “whether the rules about 'algo trading' were written about your ICT EA” and you leave with five specific behaviors to check, and a clean answer for each.

FTMO never banned your EA - it banned a list of behaviours yours probably does not have

The permission question was the easy half. What actually ends automated accounts is narrower and duller than a ban: a trailing stop that talks to the server too often, a resting stop sitting in a news window that only exists once you are funded, a Friday position on a Standard account, and one popular bought EA running on more capital than the cap allows. Every one of those is a setting, not a strategy - which is why the audit is worth running before the challenge, not after the first warning email.

FAQ

Do I have to declare my EA to FTMO before I start?

No. There is no registration step, no approval queue and no strategy submission — permission to run an expert advisor is blanket across the evaluation and the funded account. What you are accountable for is behaviour after the fact, which is why the rule set is enforced against what your account actually did rather than against anything you told them in advance.

Is an ICT or Smart Money EA treated differently from any other EA?

No. FTMO's rules name behaviors, not strategies, so there is no ICT-specific clause to satisfy. In practice a structure-based model clears most of the restricted behaviors by construction: it acts on closed candles rather than quote latency, trades few times per session, and has no reason to hold both sides of a market.

Can my EA trade during news releases on FTMO?

During the evaluation, yes — no news restriction applies. On a funded Standard account, trades on the affected instruments may not be opened or closed within two minutes either side of a targeted release, and a stop-loss or take-profit triggering inside that window counts as closing a trade. The Swing account type carries no news restriction at all.

Does the 2,000-request limit apply during the Challenge too?

Yes. The forbidden-practices rules, including the hyperactivity threshold, apply throughout the evaluation as well as on a funded account. The most common cause in an otherwise quiet EA is tick-by-tick stop modification rather than the number of trades placed.

Can I run the same ICT EA on several FTMO accounts?

You can hold multiple accounts, but the combined simulated capital is capped at four hundred thousand dollars per trader or per strategy. Identical trading detected across accounts whose total capital exceeds that cap can lead to suspension — a real risk for anyone running a widely sold EA on default settings.

Will FTMO ban my EA for martingale-style re-entries?

Martingale and grid systems are not banned by name. They are assessed against the replicability and risk-management standard, which flags position sizes far larger than your usual trades, unusually high position counts, and stacked exposure in one symbol. Fixed risk per trade and a hard cap on concurrent positions keep a legitimate structure-based re-entry from resembling that pattern.

What happens if my EA breaks a rule by accident?

It depends on which rule. Server overload usually brings a request to adjust the EA's parameters. Practices that produced an unearned result can lead to trades being removed from the history or the account being restricted. Deliberate breaches of the forbidden-practices list can end the agreement. A loss-limit breach is different from all three — it fails the account automatically, with no review step.

Sources & Further Reading

Want to go deeper? These independent, authoritative sources shaped this guide — each one is worth reading in full:

Signalbots Funded Desk

The Funded Desk is the SignalBots editorial team covering prop-firm challenges and funded-account trading. We research and write the guides on evaluation rules, drawdown limits, payout structures and the discipline funded trading demands.

More from this desk

Discussions 0

Leave a comment