Passing a Prop Firm Challenge on an ICT EA: The Full Path

What Passing a Challenge With an ICT EA Actually Requires

Four things have to be true at the same time. Miss any one of them and the other three stop mattering.

1. Permission — the firm allows automated trading, and allows your kind of automation.
2. Fit — the firm's platform, execution and rule numbers suit a system that trades unsupervised.
3. Compliance — the EA respects the daily loss limit, the consistency rule and the news blackout while you are asleep.
4. A second gear — once funded, the objective changes from "hit the target" to "keep the account", and the configuration has to change with it.

The order is not decorative. Permission and fit are decided before you spend a challenge fee. Compliance is decided in the EA's inputs before you attach it to a chart. The funded-phase change is decided in the week you pass, when you are least inclined to touch anything that just worked.

Is Your EA Even Allowed at This Prop Firm?

Three separate permissions hide behind the phrase "we allow EAs", and a firm can grant one while quietly refusing another.

The first is plain algo-trading permission: may a program place orders on this account at all, on both the evaluation and the funded phase? Some firms permit automation only after funding. Some permit it only as a trade-management assistant, not as an entry engine.

The second is strategy-family permission. Even firms that welcome EAs usually publish a banned list: latency arbitrage, tick-level order spam, exploiting a delayed feed, and high-risk recovery schemes such as martingale and grid averaging. Most ICT logic sits comfortably outside that list — it is structural, low-frequency and discretionary in spirit — but read the wording anyway, because a "no scalping inside X seconds" clause can catch a sweep-entry system that you would never think of as high frequency.

The third permission is the one that catches ICT traders specifically, because ICT and Smart Money EAs are heavily sold as finished commercial products. A growing number of firms require that the strategy be yours — not one that hundreds of other funded accounts are running on the same instruments in the same minute — and they cap how much aggregate capital they will allocate to any single identifiable strategy. A marketplace EA with a large customer base is therefore a different compliance question than one you wrote, commissioned, or materially modified. Nothing about that rule is visible in a backtest, and no amount of tuning changes it.

Read the actual rule document rather than the sales page, and read it before you buy the evaluation. The full survey of which prop firms allow expert advisors, and the exact language each one uses, is a separate piece of work from this one — and so is the single most-asked version of the question, whether you can use an EA on FTMO.

Choosing a Prop Firm Built for Running an EA

Permission is a yes/no gate. Fit is the part traders skip, and it is where an automated account quietly bleeds.

A discretionary trader can absorb a bad fill by simply not taking the next trade. An EA cannot. The things that matter to an unsupervised system are unglamorous:

• Execution quality. An ICT entry is often placed at the exact moment liquidity thins out — right after a sweep, into displacement. That is where slippage and requotes live. A firm whose fills are perfectly fine for a swing trader can be actively hostile to an entry model that fires on the candle that took out the high.
• Symbol naming. Firms hand you server-specific instrument names with suffixes and prefixes. An EA with a hardcoded symbol list silently trades nothing, or trades the wrong thing. Handle symbol mapping explicitly rather than assuming the demo server's names carry over.
• Hedging or netting. If your logic can hold two positions in opposite directions, or scales into a zone in layers, the account model decides whether that behaviour is even possible.
• Holding rules. Overnight and weekend holding, and how swap is treated, decide whether a Friday-afternoon setup is a trade or a rule breach.
• Platform and hosting. Whichever terminal the firm runs, the EA has to be awake when the setup prints, which in practice means hosting it on a server rather than on your laptop.

Cost belongs in the same decision. An evaluation you have to retake twice costs three fees, not one, so put the real number in front of yourself with the challenge cost calculator before you commit. Working out the best prop firm for running an MT5 EA is a ranking exercise with its own criteria and its own comparison table — treat it as a deliberate choice rather than defaulting to whichever firm advertised at you most recently.

Surviving the Evaluation's Daily Drawdown Limit

The daily loss limit is the single most common way an otherwise-profitable EA gets disqualified. It is a hard ceiling on how much the account may lose in one trading day, and breaching it usually ends the evaluation on the spot — no warning, no appeal, no partial credit for the fourteen good days before it.

Three details do nearly all the damage to automated accounts. First, the line is almost always measured on equity, not balance — the distinction is exactly the one in balance vs equity — so an open position sitting at an unrealised loss counts against you before it closes. An ICT EA that holds through a retracement into a deeper order block can breach the limit without ever booking a losing trade. Second, the "day" is the firm's server day, and its reset hour rarely matches your local midnight or your session filter; a system built around the London open and a firm whose day rolls at the New York close disagree about which trades belong to which day. Third, some firms measure the limit from the day's starting balance and others from the day's highest equity, which turns a good morning into a smaller afternoon budget rather than a larger one.

Note the fourth row. Minimum-trading-day requirements catch automated accounts more often than they catch humans, because an EA that finds three clean setups in a week can reach the profit target before the day count is satisfied, then keep trading — fully exposed, for no additional reward. There is far more to managing the daily drawdown limit than one setting: where the guard lives in the code, how it halts new entries versus how it treats positions already open, and how to align its clock with the firm's are all their own subject, and they deserve the full treatment.

Staying Inside the Consistency Rule While the EA Runs

The consistency rule is not a loss limit at all, which is exactly why traders miss it. It caps how much of your total profit may come from a single day, or sometimes from a single trade. Break it and the account can finish the evaluation green, inside every drawdown limit, and still be refused — or the payout is held back until the profit distribution flattens out.

An ICT EA is structurally exposed here. Smart-money logic produces lumpy output on purpose: most sessions give nothing, and then one displacement day after a clean sweep delivers a large multiple of an average day. That is the strategy working as designed, and it is also the precise shape the rule penalises. You cannot code your way out of the tension by trading more; you manage it by capping what any one day is allowed to contribute, usually by reducing size once the day's profit passes a threshold. Check where your own equity curve sits against that threshold with the consistency rule calculator rather than assuming you are clear. How the rule is defined firm to firm, and how an EA stays inside it without throttling itself into missing the profit target, is a mechanism worth studying on its own.

Does Your EA Handle News-Event Blackouts?

Most firms restrict trading around high-impact economic releases. The restriction takes at least three shapes, and they are not interchangeable: some firms forbid opening a position inside the window, some forbid holding one through the release, and some simply void the profit from any trade that touched it.

This rule bites ICT systems harder than most, because ICT logic is drawn toward the news window rather than away from it. Displacement, stop runs and the clean sweeps your EA is built to detect cluster around exactly the minutes the firm has fenced off. A slow mean-reversion system may sit quiet through a release by accident; a sweep-entry EA will walk straight into it.

Two settings do the work. A news filter built on the firm's own definition of "high impact" and the firm's own window — not a generic calendar plugin's defaults — and a trading session filter that keeps the EA inside the hours you actually validated. Both need to handle positions already open, not just new entries, if the firm's wording bans holding through a release. The edge cases in staying compliant with news-trading restrictions are real, too: scheduled releases get moved, and an unscheduled central-bank statement appears on no calendar plugin at all.

The Smart-Money Risk Framework Behind Every Setting

Here is what turns the previous three sections into one problem instead of three. The daily loss limit, the total drawdown and the consistency rule are all downstream of a single decision: how much the EA risks on each trade, and how that number is derived.

An ICT entry defines its own invalidation. The stop sits below the low that was swept, or beyond the far edge of the order block, and that distance changes trade to trade — sometimes by a factor of several. An EA using a fixed lot size therefore takes a wildly variable risk per trade, and it is precisely the widest-stop trades, on the messiest structure, that tend to arrive in clusters. That is how a system with a respectable win rate produces a breach: not because it was wrong, but because it was wrong three times in a row at three times the intended size.

The fix is a position sizing rule that computes lots from the stop distance and a fixed percentage of equity, so every trade risks the same amount no matter how far away the invalidation level sits. Once that holds, the rule limits become arithmetic you can check in advance: a given risk percentage and a plausible run of consecutive losses either fit under the daily line, or they do not.

Two honest notes on that arithmetic. It assumes each loser takes its full stop, which is optimistic in fast conditions and pessimistic when partial exits are in play; and it assumes concurrent positions are genuinely independent, when three correlated pairs held long at once are really one position wearing three names. To turn a chosen risk percentage into an actual lot for a given stop distance, use the max lot calculator. The wider framework — how a favourable reward-to-risk ratio, structural stop placement and Smart Money risk management for the funded account fit together into one methodology — is the backbone this whole sequence rests on, and it repays reading in full rather than in summary.

Getting Funded: Protecting the Payout After You Pass

Passing is a milestone, not an ending, and the day it happens the incentives invert. The evaluation paid you for reaching a target, so a degree of aggression was rational. The funded account pays you for staying funded, so that same aggression becomes the main threat to your income.

Most configurations should carry less risk after the pass than during it. The account is worth something now — a fee, weeks of runtime, and a passing result you would have to reproduce from scratch to replace. A Smart Money EA on a funded account still running evaluation-phase sizing is optimising hard for a goal it has already achieved, at the expense of protecting the payout it now exists to produce.

Three practical things change. Payout cycles introduce a new clock, with minimum days and request windows that decide when profit becomes money rather than a number on a dashboard — model your side of it with the payout split calculator. Drawdown rules often change shape after funding, sometimes trailing your equity high instead of sitting static. And a manual kill switch you can reach from your phone stops being a nice-to-have, because the account is now an asset rather than an experiment. The work of protecting your payout once funded — how far to scale risk down, how to time activity around a payout window, and what to do after the first withdrawal clears — is a distinct discipline from passing the evaluation in the first place.

Pre-Flight Checklist: Matching EA Settings to Prop-Firm Rules

Everything above collapses into one table. Each row is a knob in your EA's inputs, the rule it exists to protect, and what a missing setting actually costs — because most challenge failures are not strategy failures. They are one unset parameter.

Two of those rows deserve a line of their own. A max open trades limit is not the same thing as a correlation cap — an EA can honour the first while running three EUR-weighted longs that behave as a single oversized position. And a distinct magic number per EA is what lets your own guard code tell its trades apart from anything you opened by hand, which matters the moment you want to intervene without confusing the system.

Then validate the whole configuration where it will actually run — on the VPS that will host it, not on your desktop. A pass in the strategy tester proves the logic; a forward test on the firm's own demo server, with their symbol names, their server clock and their spreads, proves the configuration.

The Full Path, Start to Finish

The reason so many capable ICT systems fail evaluations is not that the strategy is weak. It is that the strategy was tuned against a P&L curve while the evaluation grades four other things at the same time. Everything above, in the order you actually need it:

If you want to pressure-test the plan before committing money, the challenge simulator lets you play a profit target, a daily limit and a risk percentage against each other until the combination stops looking comfortable. That is usually the fastest way to discover that a configuration which felt conservative was carrying twice the risk the rule set allows.

FAQ

Can an EA actually pass a prop firm challenge?

Yes, at firms that permit automated trading — automation is not itself a disqualifier at most of them. What decides the outcome is whether the EA's risk behaviour fits inside the firm's rule set, not whether a human or a program pressed the button. A system with fixed-percentage sizing, a hard stop on every trade and a working daily-loss guard has a structural advantage over a discretionary trader here, because it never negotiates with its own rules at 3 a.m.

Do prop firms detect that you are running an EA?

Assume yes. Firms review trade timing, execution latency, order patterns and how closely your account's activity resembles other accounts, and that review typically happens at payout time rather than during the evaluation. This is why the eligibility question belongs at the start rather than the end: discovering that a rule prohibited your setup is far cheaper on day zero than at the first withdrawal request.

Should I run my ICT EA with the same settings I use on my own account?

Almost never. Your own account has no daily loss ceiling, no consistency rule and no news blackout, so a configuration tuned there is optimised against a completely different constraint set. At minimum the risk percentage usually needs to come down, and three guards need to be added — daily loss, per-day profit cap and news filter — before the same logic is fit for an evaluation.

What is the most common reason an ICT EA fails an evaluation?

An unrealised loss on a held position crossing the daily equity line. ICT entries frequently sit through a retracement toward a deeper level before working, and the firm's daily limit is measured on equity in real time, so the account can breach while every closed trade is still a winner. The second most common cause is variable position size: a fixed lot setting combined with the widely varying structural stops that ICT setups naturally produce.

Can I run the same ICT EA on several challenge accounts at once?

Only where the firm explicitly allows it, and the answer varies more than any other rule in this article. Some firms permit the same strategy across multiple accounts you personally own; others treat identical trade copies as a violation; and many cap the total capital they will allocate to any single identifiable strategy across their whole customer base. If your EA is a commercial product with many users, that cap can be reached by other people's accounts before yours is affected at all.

Sources & Further Reading

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

• MQL5 Reference — AccountInfoDouble, the equity and margin properties a drawdown guard reads
• MetaTrader 5 Help — Expert Advisors and algorithmic trading permissions
• Federal Reserve — FOMC meeting calendar and statement release times