EUR/USD expert advisor vetting guide: avoid funded blowups

Why does a good-looking EUR/USD robot fail on a funded account?

You download a robot because the equity curve climbs smoothly from left to right. You attach it to a funded account. The first week looks calm. Then a trend day arrives, the robot adds positions, floating loss grows, and you breach a daily or total loss limit you barely understood. The challenge ends. The problem was not one bad trade. The problem was that nobody taught you how to evaluate an EUR/USD EA before it touched capital that mattered.

This guide fixes that gap. It gives you a risk-first method for judging any EUR/USD expert advisor before it ever touches a live account. You will learn what the robot actually does, why EUR/USD attracts so much automation, how to separate evidence levels, how to read strategy instead of marketing curves, which numbers to demand, how to stress a backtest, how to size positions, and how to run a demo test that tells you something real. If you follow the steps, you will reject most robots quickly and for clear reasons. That is the point. Rejection protects capital.

The direct answer is simple. Do not judge a robot by its curve. Judge it by its losing behavior, its exposure, its costs and its honesty under test. Past performance does not predict future results. Automation does not remove risk. Demo testing comes first, live capital comes last, and only after a written gate says the robot earned it.

What is an EUR/USD EA actually doing when you attach it to a chart?

An EUR/USD EA is a program that runs inside MetaTrader 4 or MetaTrader 5 and sends trade instructions for the EUR/USD pair without asking you each time. When you attach it to a chart and allow automated trading, it watches price, checks its coded conditions, and then opens, modifies or closes positions. It can set entry orders, attach a stop loss and a take profit, move a stop, close part of a position or close everything when a rule triggers. It repeats this on every new tick while it is active.

That sounds powerful, and it is, but it is also narrow. The robot only knows what its developer coded. It does not understand central bank meetings, sudden headlines or a widening spread during a rollover. It follows rules. If the rules fit the current market, results can look orderly. If the market changes character, the same rules can keep trading as if nothing changed. Your job is to learn exactly what those rules are before you trust them.

To evaluate any robot, ask how it decides to enter. Common approaches include trend continuation, mean reversion, breakout, time-based logic and indicator filters. Then ask how it decides to exit. Does it use a fixed stop loss on every position. Does it use a take profit target. Does it trail. Does it close on time, on an opposite signal or only when profit appears. Entry logic gets attention in marketing, but exit logic decides how losses behave, and losses decide survival.

You also need to know how the robot handles multiple positions. Some robots hold one position at a time with a clear stop. Others add to losing positions, hold baskets of trades or keep averaging as price moves against the entry. Those designs change your exposure completely. One position with a defined stop is one kind of risk. Ten correlated EUR/USD positions without a firm stop is a different kind of risk. The chart may not show the difference until stress arrives.

Execution details matter too. In MetaTrader 4 and MetaTrader 5, an EA depends on your broker feed, your spread, slippage between signal and fill, and whether your platform stays connected. A virtual private server, usually called a VPS, keeps the platform running when your computer sleeps. Without stable hosting, a robot that needs to manage open trades can miss its exit logic at the worst moment. That is not a strategy flaw in the code, but it becomes your loss all the same.

Finally, learn what the robot does when conditions are bad. Does it stop trading above a maximum spread. Does it pause around major news. Does it limit trades per day. Does it stop after a daily loss. Does it detect a disconnect. A serious design has clear answers. A weak design trades through everything and hopes. Hope is not a risk control.

If you take one idea from this section, take this. How to evaluate an EUR/USD EA starts with a plain-language description you could explain to another trader. If you cannot say what triggers entries, what limits losses, how many positions can be open and when the robot stands aside, you do not understand it well enough to run it.

Why is EUR/USD the most automated pair, and what does that mean for you?

EUR/USD earns its reputation as the most automated pair for practical reasons. It trades with deep liquidity through much of the day, spreads are usually tight compared with exotic pairs, and many brokers offer continuous quotes with a long price history for backtesting. Developers like that combination because it makes testing easier and transaction costs look lower on paper. That is why you see so many robots built around EUR/USD first, with other pairs added later.

For you, popularity is both an advantage and a trap. The advantage is data and discussion. You can find long histories, session research, broker specification pages and plenty of community threads about behavior during news or rollover. The trap is false confidence. Because EUR/USD often looks smooth and technical, traders assume it is gentle. It is not. Trend days, reversals after data releases and liquidity gaps can move price quickly. An EA that looked calm in a range can face a completely different market in a trend week.

Automation density also changes execution. Many robots watch similar levels, similar session opens and similar indicator values. That does not mean they all trade identically, but it means crowded behavior can appear around obvious zones. Breakouts can extend, pullbacks can be shallow, and slippage can widen when many orders arrive together. Your evaluation should therefore include costs and fills, not only signals. A strategy that needs perfect fills will struggle more in real trading than a strategy that tolerates delay.

Another implication is session dependence. EUR/USD behaves differently across the Asian, London and New York periods. Volatility, spread and direction change through the day. A robot tuned on one session may look strong in its favorite window and weak outside it. Ask the developer which sessions the logic targets. Then check whether the backtest and the demo test cover those sessions honestly, or whether quiet hours are padding the result while active hours carry the risk.

The tight spread story needs the same care. A low average spread helps scalping logic, but averages hide spikes. Rollover, news releases and weekend gaps can widen dealing costs for short stretches. If a robot trades through those moments, its real cost is higher than the average suggests. Evaluation should ask for spread handling, not only spread assumptions. A spread filter, a news pause and a time filter are basic protections you should expect to see documented.

Use this context when you browse vendor pages or community libraries for an EUR/USD EA mt4 or EUR/USD EA mt5 download. Popularity means choice, but choice means filtering. Prefer developers who explain why EUR/USD suits their logic, which sessions they target, how they handle news and what happens when spread widens. Vague claims about the pair being ideal for automation tell you nothing. Session logic, cost logic and pause logic tell you a great deal.

What is the difference between a live result, a forward test, a backtest and a vendor claim?

These four terms sound similar, but they carry very different weight. Mixing them up is how traders pay for lessons they could have learned on paper. Learn the ladder, then demand the right rung before you risk anything.

A vendor claim is marketing. It is a screenshot, a sentence about behavior, a video of a chart or a promise implied by design. Claims cost nothing to make. Treat every claim as a question to verify, not as evidence. When a page says a robot handles trend days well, write down how you will check that. When it says the logic is conservative, ask what mechanism makes it conservative. Language does not limit loss. Code and settings do.

A backtest is a simulation on historical data. The Strategy Tester in MetaTrader replays past prices through the EA rules and reports what would have happened under the chosen settings. Backtesting is useful for learning and for rejecting fragile ideas, but it is not proof. History is fixed, costs are assumed, and execution is modeled. Small changes in spread, slippage, data quality or settings can change the picture. A backtest tells you what the developer chose to show you, under conditions the developer chose.

A forward test runs the EA on current prices as they arrive, usually on demo. It faces real spread movement, real timing and real session behavior, even though fills are still simulated by the demo server. Forward tests are stronger than backtests because the market is unknown in advance. The developer cannot tune entries to future candles that have not printed yet. That is why this guide puts a structured demo test at the center of evaluation.

A live result shows real money with a real broker. Verified tracking on a service such as Myfxbook can add credibility because it ties trades to an account and shows deposits, withdrawals and open exposure. Even then, read carefully. A short live record, a tiny balance, frequent deposits, hidden open trades or a change of broker mid-record all weaken the signal. Live trading carries the most weight, but only when it is long enough, transparent enough and traded under conditions similar to yours.

Evidence type	What it shows	Main weakness	How to use it
Vendor claim	Marketing summary and selected highlights	No verification and selective presentation	Turn each claim into a test question
Eurusd ea backtest	Simulated behavior on past data	Curve fitting, cost assumptions and fixed history	Use it to reject fragile logic, not to approve
Forward and EUR/USD EA demo test	Behavior on unseen prices with live timing	Demo fills differ from live fills	Require it before any live decision
Verified live result	Real fills, costs and psychology of real capital	Short samples, hidden exposure and account games	Give it weight only when transparent and long enough

A practical way to apply this ladder is to ask for evidence in order. Start with the full backtest report and settings file. Then ask for forward demo history with the same settings. Then ask for verified live tracking if it exists. If a developer jumps straight from claims to a sales page, you have your answer. Serious work leaves a trail. Marketing leaves a glow.

For deeper method on simulation controls, work through a step-by-step backtesting tutorial alongside the vendor report. For context on how accounts are compared in public, browse a ranking hub and notice how much extra information transparent records provide compared with screenshots.

How do you read the strategy instead of the equity curve?

Curves seduce. Strategy explains. A smooth historical line can come from sound trade management, but it can also come from holding losing positions until they recover, adding size into drawdown, or avoiding a stop so winners close and losers linger. Those designs look calm until the market stops returning to the entry. Your task is to look past the line and name the mechanism.

Start with position structure. Ask whether the robot holds one position at a time or can hold many. Ask whether new positions add in the direction of profit or in the direction of loss. Terms you will meet include grid, martingale and averaging. A grid opens orders at set price steps. Martingale-style sizing increases size after a loss. Averaging adds to an open losing position to improve the average entry. Each approach can produce long calm stretches followed by sharp exposure growth. None of them can be judged from a curve alone. You need the trade list, the lot progression and the floating drawdown through time.

Next, pin down stop-loss reality. A real stop is attached to each position, honored by the platform, and visible in the trade history as a bounded loss. A virtual or logic-based exit only works while price, connection and code cooperate. Ask what happens if the VPS drops, if the terminal restarts, or if price gaps through the exit level. Slippage means the fill can differ from the trigger. That is normal, but it should be discussed openly, not hidden. If there is no stop at all, say so plainly in your notes, because that single fact changes the entire risk picture.

Then examine profit handling. Some robots close quickly for small gains and let losses run. That pattern can show a high share of winning trades while overall balance depends on a few large losing events. Others wait for larger moves and accept more frequent small losses. Neither style is automatically good or bad. What matters is whether the math, after spread and slippage, survives losing streaks and difficult months. Ask to see the full distribution of wins and losses, not only totals.

Session and news handling belong in the same review. EUR/USD can drift in quiet hours and trend during active hours. A robot that scalps quiet periods needs different protection than a robot that rides London or New York momentum. Ask which hours it trades, whether it pauses for scheduled events, and how it behaves after a large candle. Vague answers suggest the developer has not studied regime change. Precise answers suggest real testing.

Design question	What good looks like	What should worry you
How many EUR/USD positions can be open together	A stated maximum with a reason tied to account size	No maximum, or a maximum that changes without explanation
How size changes after a loss	Fixed or clearly bounded sizing with documented logic	Size grows to recover prior losses
Where the stop sits	A real stop on every position plus a documented account-level halt	No stop, or an exit that only works when conditions are ideal
When the robot stands aside	Spread filter, news pause and time filter with values you can inspect	Trades through news, rollover and extreme spread without pause
What ends a bad day	A daily halt or a reduced-risk mode that is coded, not promised	No halt, or a halt you must apply by hand while drawdown grows

Write your findings in one paragraph you could defend. For example, you might note that the robot trades breakouts during active sessions, holds a single position with a real stop, avoids scheduled news, and halved activity when spread widens. Or you might note that it averages into losses, carries multiple correlated positions, and relies on price returning. Both notes are useful. The first describes bounded risk. The second describes exposure that needs far stronger proof before any demo time, let alone live capital.

If the developer cannot or will not explain entries, exits, maximum exposure and pause conditions, stop the review. Lack of disclosure is itself a finding. You do not need to argue about potential. Move on to a robot whose maker respects your need to understand EUR/USD EA risk before you accept it.

Which five numbers should you demand before you trust any EUR/USD expert advisor?

Vendors love totals. Totals hide behavior. Ask instead for the five numbers that reveal how a robot loses, how long it suffers, how much it carries, how robust its sample is and how sensitive it is to costs. If any number is missing or cannot be reproduced, treat that as a failed gate.

First, drawdown depth through time, including floating exposure, not only closed-trade balance. Balance can look smooth while equity dips deeply between closes. Ask for an equity-based drawdown series that includes open trades. You want to see when drawdown happened, how long it lasted, and what market behavior caused it. A single summary value without context tells you little. A dated series with notes tells you how the strategy suffers.

Second, exposure behavior at the worst point. Ask for the maximum number of simultaneous EUR/USD positions, the total open size at that moment, and how long that exposure lasted. This number connects strategy to survival. A robot that once held a large basket for many hours carries a different risk than a robot that never exceeds one bounded position. The trade list and the account statement should support the answer.

Third, sample size and period coverage. Ask how many closed trades the result contains, over which date range, and across which market regimes. A result with few trades or only calm months proves little. You want trades across trend periods, range periods, news weeks and different sessions. Small samples and short windows are easy to fit by accident. Demand breadth before you give weight.

Fourth, cost sensitivity. Ask how the result changes when spread and slippage assumptions become less kind. A serious developer can show the same backtest with higher costs and explain the break point where edge fades. Backtesting in MetaTrader lets you adjust these assumptions, and forward demo testing shows real spread behavior. If performance depends on perfect fills, it is not robust enough for live conditions where fills vary.

Fifth, stop-loss reality and tail behavior. Ask for the largest single losing trade, the average losing trade relative to the average winning trade, and the longest sequence of consecutive losses. Then ask which stop or account halt would have contained a worse run. These figures describe the tail you must be able to survive. Leverage cuts both ways, and a strategy that needs an unusually calm tail to survive is asking the market for favors.

Number to demand	Why it matters	What to ask for
Equity-based drawdown with dates	Shows real pain including open trades	Dated equity series with notes on cause and length
Maximum exposure held	Shows how much was at risk at once	Peak open positions, total size and holding time
Trade count and date range	Shows whether the sample means anything	Full trade list across varied regimes
Cost sensitivity	Shows whether edge survives real fills	Same test with higher spread and slippage settings
Loss structure and stops	Shows whether one bad run ends the account	Largest loss, loss sequence and halt mechanism

When you request these figures, ask for files, not screenshots. You want the Strategy Tester report, the settings file, the symbol and timeframe, the data source, and the exact inputs used. You also want read-only investor access or verified tracking for any live or demo claim, so you can check deposits, withdrawals and open trades yourself. A developer who trades transparently will understand the request. A developer who stalls, blames disclosure, or sends only cropped images has told you what you need to know.

Use a public ranking hub to see how transparent records present drawdown, exposure and history length. The format will train your eye for what a complete answer looks like, which makes incomplete answers easier to spot.

How do you test a backtest for dishonesty?

A backtest cannot prove a robot works, but a dishonest backtest can prove you should walk away. Your goal is not to confirm the curve. Your goal is to break it honestly and see what remains. Work through these checks in order, and write down what you find.

Start with data and modeling quality. In MetaTrader 4 and MetaTrader 5, the Strategy Tester reports modeling quality and lets you choose tick methods. Poor tick modeling, missing history, weekend gaps handled badly or a single broker feed with unusually smooth prices can flatter a scalper. Re-run the vendor settings file on your own data, preferably from your own broker, and compare. If the result changes sharply with data alone, the edge was data-dependent, not robust.

Next, inspect spread and slippage honesty. Many flattering reports assume a fixed low spread with no delay. Real EUR/USD spread moves, especially around news, rollover and thin liquidity. Re-run the same EUR/USD EA backtest with wider spread and with slippage applied, then observe whether the result degrades gracefully or collapses. A robust idea bends. A fitted idea breaks. Also check whether the robot trades during high-cost minutes. If most profit comes from moments when real spread would have been wide, the test is telling you more about assumptions than about strategy.

Then look for curve fitting through inputs. Open the inputs and ask what each parameter does. If the robot has many finely tuned values that only work in a narrow band, small changes should not destroy a sound idea completely, but they will move results. That is normal. What is not normal is a report where one magic combination shines while nearby values fail. Use the optimizer sparingly and honestly. Test nearby values, adjacent date ranges and out-of-sample periods the developer did not submit. A strategy that only works on one exact setting and one exact window is a description of the past, not a plan for the future.

Session and symbol tricks deserve their own pass. Some reports mix timeframes, switch symbols mid-test, or include only favorite months. Confirm the symbol is EUR/USD, the timeframe matches the vendor file, and the date range covers varied conditions. Then split the test by session and by year. If all strength clusters in one quiet regime while active periods struggle, you have learned where the risk lives. That knowledge is more valuable than any total.

Finally, read the trade list line by line. Look for long holding times on losers versus quick exits on winners, growing size after losses, gaps in trading that suggest manual intervention, and profit concentrated in a few unusual trades. Check whether stops appear on every position or only in description text. Check whether the worst floating exposure ever approached an account halt. A trade list tells the truth that a headline hides.

Pay separate attention to out-of-sample behavior and tick-data realism. A developer can tune inputs until one historical window looks attractive, so always reserve a later period the tuning never saw and run the same file there without changes. Then repeat the run with high-quality tick data and realistic delay settings, because scalping and news logic are sensitive to how prices are modeled between candles. When results hold broadly across unseen periods and stricter modeling, confidence grows modestly. When they fall apart outside the marketed window, you have found fitting rather than edge, and rejection is the correct outcome.

Keep notes as you go, because memory fades and marketing sticks. Record the data source, spread setting, slippage setting, date range, input file name and result for each run. When you later compare demo behavior to backtest behavior, those notes let you distinguish normal variance from a genuine mismatch. That comparison is the heart of honest evaluation.

How should you size positions when an EA trades for you?

Position sizing is where automation risk becomes personal. A robot can follow perfect entry logic and still end the account if size is wrong for the balance, the stop distance or the loss limit. Think of sizing as the bridge between strategy behavior and account survival. You control that bridge. Never delegate it fully to default EUR/USD EA settings.

Start with the account constraint, not the robot. Funded challenges and many live accounts enforce daily and total loss limits. Those limits decide how much heat you can take before the account closes. Your sizing must keep a normal losing sequence well inside those lines, with room for spread widening and slippage. If you do not know the exact loss rules for the account you plan to use, stop and read the terms first. An EA cannot respect a limit you never entered into its controls.

Next, translate the strategy into exposure. A single-position robot with a real stop risks a bounded amount per trade. A basket robot can risk far more at once because several positions share the same direction. Ask for the maximum simultaneous exposure in plain terms, then map that to your balance. Walk through a difficult week on paper. Suppose several trades lose in a row while one basket stays open and floating loss grows. Could you still trade the next day without breaching a limit. If the answer is unclear, size is too large or the design is too opaque for that account.

Use a worksheet instead of memory. The free checklist with this guide includes a position-size worksheet that forces you to write down balance, loss limits, maximum open trades, stop handling and the account halt level. Writing forces honesty. It also creates a record you can compare against demo behavior. Many traders discover at this stage that default inputs assume a larger balance or a looser limit than they actually have. Defaults serve demonstration, not your account.

Controls matter as much as math. Look for a maximum spread filter, a maximum open-trade cap, a daily loss halt, reduced trading after consecutive losses, and a news pause you can configure. Test each control on demo to confirm it triggers. A control you never tested is a hope, not a control. Also confirm how the robot behaves after a halt. Does it close baskets cleanly, or does it leave exposure running while you sleep. The answer decides whether you can trust it on a VPS without watching every tick.

Review costs in the same pass. Spread and slippage act like a constant headwind, especially for frequent trading. Commission and swap terms vary by broker and account type, so check the contract specification on your own platform before you judge profitability. Comparing reputable broker specification pages can help you understand account types and cost structures, but the final check always happens on your own feed with your own settings. If costs consume the edge on demo, live trading will not rescue it.

End this step with a written sizing rule you can follow without judgment calls. State the balance you tested, the exact inputs you used, the maximum exposure you accept, and the condition that stops trading for the day. If you cannot write that rule clearly, you are not ready to attach the robot to anything beyond demo.

How do you run a 14-day demo protocol that actually tells you something?

A casual demo proves little. You glance at profit, ignore exposure, change settings midweek and declare the robot ready. A structured protocol does the opposite. It fixes one configuration, records behavior daily, and forces a decision based on evidence. Fourteen days is a minimum window to see varied sessions and at least some adversity, not a magic number that guarantees insight. Longer is better when trade frequency is low.

Preparation comes first. Create a demo account that mirrors your intended live conditions as closely as possible, including the same MetaTrader version, the same broker type, similar balance and leverage terms, and the same EUR/USD symbol specification. Install the correct build for your platform, because an EUR/USD EA mt4 file and an EUR/USD EA mt5 file are not interchangeable. Load the exact settings file you plan to judge, enable the spread filter and news handling you intend to use, and host the terminal on a stable VPS if you plan to trade that way live. Document everything before the first trade.

During the test, change nothing unless safety demands it. One configuration, one record. Log date, session traded, number of trades, open exposure peak, spread behavior, any news events, any disconnects, and how the robot behaved around them. Save the daily statement and keep screenshots of floating drawdown when baskets stay open. If you feel tempted to adjust inputs after a losing day, write the urge down instead of acting on it. Mid-test tuning restarts the clock because the result no longer belongs to one configuration.

Judge behavior, not balance alone. Compare demo fills to the backtest assumptions. Did spread widen during the robot’s favorite minutes. Did slippage appear on exits. Did the robot respect its time filter and pause logic. Did maximum exposure stay within the worksheet limit. Did losing trades match the loss structure the developer disclosed. Small differences are normal between simulation and demo. Large or repeated mismatches suggest the backtest was optimistic or the settings file differs from the marketed one.

Day block	Focus	What to record
Days one to three	Installation and baseline behavior	Fills, spread filter triggers, exposure peak and any setup errors
Days four to seven	Session coverage	Behavior across quiet and active sessions, overnight holds and news handling
Days eight to eleven	Adversity and controls	Losing sequence handling, daily halt test, reconnect behavior on VPS
Days twelve to fourteen	Review and decision	Full statement vs backtest, cost drag, setting stability and written gate

At the end, write a short verdict tied to your gate. State whether the robot behaved as documented, whether costs and exposure stayed inside limits, and what would need to change for a longer test. If trade frequency was too low to judge, extend the demo rather than guessing. If the robot breached exposure limits, ignored filters or required manual rescues, reject it even if the balance line looks acceptable. Behavior is the test. Balance is only one output.

Keep the statements, settings file and daily notes with your checklist. That packet is your memory when marketing tries to rewrite history. It also makes later comparison easier if you test an updated build. Version control matters, because even a small logic change can alter risk completely.

What red flags should stop you immediately?

Red flags save time. They let you reject a robot in minutes instead of weeks. Treat the list below as hard stops, not debate points. One clear flag is enough to walk away, because serious developers rarely trigger any of them.

Opacity is the first flag. If there is no strategy explanation, no settings documentation, no trade history and no willingness to discuss losing periods, you are being asked to trust a black box with your capital. Refuse. A serious maker can explain entries, exits, maximum exposure and pause logic without revealing proprietary code. Secrecy about risk is not intellectual property protection. It is a warning.

Pressure is the second flag. Countdown timers, private deals that expire tonight, demands to fund a specific account before you see evidence, and refusal to offer a demo-readable version all point the same way. Sound evaluation takes days. Anyone who rushes you past demo testing benefits from your haste. Take your time, and let urgency disqualify the offer.

Curve games are the third flag. Be wary of equity curves without trade lists, balance lines that hide floating drawdown, backtests with no settings file, verified badges that link nowhere, and Myfxbook records with hidden open trades or frequent deposits that reset the curve. Ask for investor-password access or a full export. If access is refused, assume the missing part contains the risk.

Logic flags come next. No stop loss, growing size after losses, unlimited baskets, trading through major news without a documented reason, and extreme sensitivity to spread all deserve instant rejection unless the developer provides unusually strong and transparent proof. Even then, remember that leverage magnifies adverse moves and that past calm does not predict future calm. Your default answer to opaque exposure should be refusal.

Flag group	Example	Your response
Opacity	No strategy note and no settings guide	Stop and request disclosure, then reject if refused
Pressure	Deposit demand before demo evidence	Walk away and document the tactic
Curve games	Screenshots without trade lists or settings	Demand files and verified history, then reject if missing
Risky logic	Averaging without a firm halt	Require written exposure proof, default to rejection
Support failure	No update path and no disconnect guidance	Reject for live use, since unattended trading needs support

A related flag is testimonial theater. Anonymous messages, unverifiable account images and borrowed lifestyle imagery prove nothing about execution, costs or drawdown. Ignore them. Focus on files, statements and your own demo record. Education before conversion is the rule that keeps you safe. If removing the sales material leaves no method to judge, the product was never evaluable.

For perspective on how marketing language differs from evidence, read an independent review example and notice which claims link to checkable records and which do not. Use the same lens on every EUR/USD offer you meet.

How do you turn all of this into a go or no-go decision?

Evaluation without a decision becomes open-ended demo trading. You need a written gate that says yes only when every check passes, and says no the moment one critical check fails. The free checklist with this guide exists for exactly that purpose. It turns the method above into pages you can print, fill in and keep.

The checklist starts with provenance. You record where the file came from, which version you tested, whether it targets MetaTrader 4 or MetaTrader 5, and who published it. A resource such as an EUR/USD EA free download deserves the same provenance care as paid software, because malicious or mismatched builds can harm an account before strategy even matters. Scan files, confirm the correct platform build, and keep the original archive with your notes.

Next comes strategy teardown. You write the entry trigger, the exit logic, the maximum simultaneous positions, the stop-loss arrangement and the pause conditions in your own words. If you cannot complete that page, the gate fails early. That early failure is valuable. It prevents weeks of testing on logic you never understood.

The five numbers have their own section. You log drawdown behavior with dates, peak exposure, trade count and range, cost sensitivity and loss structure. Each number needs a source file or statement attached. A number without a source is a rumor. Your future self will thank you for stapling proof to every claim.

Backtest honesty tests follow. You record data source, spread and slippage settings, date ranges, input file names and the outcome of each stress run. You note where the backtest bent gracefully and where it broke. You compare vendor results to your own runs on your own feed. Differences go in writing, not in memory.

The 14-day demo protocol has daily rows. You log trades, exposure peaks, filter triggers, news behavior and any intervention. You keep the statements. At the end, you compare demo behavior to the documented strategy and to the backtest assumptions. Consistency earns further testing. Repeated mismatch earns rejection.

The position-size worksheet ties the decision to your account. You write the balance tested, the loss limits that apply, the maximum exposure you accept and the halt rule you will enforce. You confirm the controls triggered on demo. If the math only works on a larger balance or under looser limits than you have, the answer is no for your account, even if the robot might suit someone else.

Finally, the red-flag page and the go or no-go gate close the loop. You check opacity, pressure, curve games, risky logic and support. Then you sign one of two statements. Either the robot passed every gate and earns a longer demo or a small live probation under the same controls, or it failed a named gate for a named reason. Both outcomes are wins, because both protect you from drifting into live risk without a record.

Download the checklist, print it, and use it for every candidate, including updates. Developers release new builds, and a new build is a new candidate. Keep each packet separate. Over time, your shelf of completed checklists becomes a personal database of what survives contact with real spread, slippage and sessions. That database is worth more than any single download.

If you want broader context while you study, compare cost structures through a broker shortlist, review loss-control thinking in a live drawdown control method, and browse the shop only after your gate says a longer test is justified. Tools support judgment. They do not replace it.

Trading foreign exchange on margin carries a high level of risk and may not be suitable for every reader. Prices can move sharply against open positions, leverage magnifies both favorable and adverse moves, automation can malfunction or disconnect, and historical simulations do not predict future performance. Test every system on demo, size positions so a normal losing sequence cannot end the account, and never commit funds you cannot afford to lose.