EA Due Diligence
To a casual observer, a quality XAUUSD EA and a scam EA look identical on MQL5. Same page format, same type of screenshots, sometimes even similar pricing. This 12-point audit provides the framework to distinguish them in under 15 minutes — before you risk any real money.
Backtest fraud specifically is covered in our backtest fraud guide. The documentation you should demand is listed in our documentation checklist.
Prosecution vs Defense — how the same evidence looks on a bad EA versus a good one.
Verdict
The accused EA failed 8/12 quality criteria. You would have known in under 5 minutes of due diligence.
Check each criterion the EA passes. 1 point per criterion — 12 maximum.
This EA fails the basic quality standard. Move on — there are too many verified alternatives to justify the risk of an EA failing 8+ criteria.
The MQL5 marketplace template is both a strength and a vulnerability for buyers. As a template, it ensures every listing has the same categories: description, screenshots, reviews, ratings, and a backtest chart. As a vulnerability, it means a developer with fabricated results uses exactly the same visual format as a developer with a genuine 18-month live track record.
The screenshots that appear on EA listings are Strategy Tester results — not live trading screenshots. Anyone can generate an impressive-looking backtest on MT5 Strategy Tester in minutes. The difference between a legitimate backtest and a fabricated one is in the parameters: what date range was used, what modelling quality was selected, what spread was applied, and whether the EA was optimised on the same data it was tested on. None of these parameters are visible in a screenshot — they require reading the raw backtest report or asking the developer specific questions.
Red flags in the EA itself (beyond these 12 points) are in our EA red flags guide.
Cherry-picked date range
Running the backtest only on a period where the strategy happened to perform well — often just a bull market period for gold, or the most recent 3–6 months. Ask: "Does the backtest cover 2020, 2021, 2022, and 2023?" Four calendar years includes bull runs, corrections, rate hike cycles, and choppy ranging periods. Any strategy that only works in one of these regimes is not a robust strategy.
Zero or unrealistic spread
Setting spread to 0 pips in the backtest parameters. Real XAUUSD ECN spread during the London session is 8–15 pips; during off-hours, 20–40 pips. A backtest at 0 spread gives every trade approximately $8–$15 of free advantage per 0.1 lot that does not exist in live trading. For a high-frequency EA taking 10 trades/day, this can represent $80–$150/day of phantom profit — thousands of dollars overstated annually.
Over-optimisation on in-sample data
The developer runs the MT5 optimisation tool with thousands of parameter combinations on historical data and selects the parameters that performed best on that specific history. These parameters are maximally tuned to past data and will almost certainly underperform on future data. This is called curve-fitting. The test is: "Did you run an out-of-sample forward test on data you excluded from optimisation?" A developer who cannot answer this has almost certainly not done it.
90% modelling quality on MT4
Approximately 10% of the trades in a 90% quality MT4 backtest use interpolated tick data rather than real historical ticks. For gold strategies, this means session-open entries and news-period behaviour are less accurately modelled. An EA that shows excellent results on a 90% MT4 backtest may show significantly different results when tested at 99% quality on MT5 with real tick data.
No slippage or commission
Backtest parameters allow setting slippage to zero and commission to zero. Real ECN gold trading has slippage of 0–3 pips on fast-moving markets and commission of $3–$7 per round trip on 0.1 lot. A high-frequency gold EA running 10 trades/day with 0 commission in backtest is overstating annual returns by $600–$1,400 at 0.1 lot — potentially more at larger lot sizes.
Rather than describing an abstract "good EA," this section uses Goldie Razor V2.8.4 as a concrete example of what meeting the 12-point audit looks like in practice:
| Criterion | Goldie Razor V2.8.4 |
|---|---|
| 3+ year backtest | ✓MT5 99% tick-data backtest covering multiple market regimes including 2022 rate hike period |
| Realistic spread | ✓Backtest run with ECN-realistic spread applied — not zero |
| Live track record | ✓Myfxbook-verified live account, real money, 6+ months visible history |
| Strategy explanation | ✓M15 range breakout with H4 200 EMA trend filter; signal, filter, and exit logic all documented |
| Parameter documentation | ✓Each parameter described in user guide with recommended ranges by account size |
| Active development | ✓Version 2.8.4 is the current active release — changelog available showing update history |
| Support channel | ✓Email and dedicated channel; response time typically under 48 hours |
| Review variance | ✓Mix of positive and constructive feedback in public reviews — not uniformly edited |
The most common reason traders skip this process is assuming it takes too long. It does not. Here is the 15-minute sequence:
Open the MQL5 listing and find the backtest report
Look for a downloadable backtest PDF or screenshots showing Strategy Tester parameters. Note: date range, modelling quality percentage, and spread setting. If none are visible, that is itself a red flag.
Find and verify the Myfxbook or track record link
Click any track record link. Verify: real money account (stated in account info), recognisable broker, account age 6+ months, no visible resets, drawdown under 30%. If no link exists in the listing, search for the EA name + Myfxbook — if not found, the track record does not exist.
Read the strategy description for plain-language explanation
You should be able to answer in your own words: what generates the entry signal? What prevents bad entries (filter)? How does the EA exit? If you cannot answer these after reading, the description is insufficiently transparent.
Check developer activity
Find the last update date on the product page. Look for a changelog or version notes. Send a pre-sales question via the MQL5 message system ("What is the average modelling quality used in your backtests?"). Note if and when you receive a response.
Read 10 reviews, looking for variance
Scroll to the most recent reviews. Are there any critical or 3-star reviews? What do reviewers say the EA is weak at? Uniform 5-star reviews from new accounts are not social proof — they are a red flag.
Related Reading
Goldie Sniper, Goldie Razor V2 and V2.8.4, Blind Sniper, and Hybrid Scalper. Full suite for every market condition.
Not sure which EA fits your account size or trading style? Email us and we will help you choose.
Pro-Scalper EA Range
Session breakout · M1 · London & NY
High-frequency session breakout capturing London open and NY session momentum on the M1 timeframe.
Range breakout · M15 · H4 EMA filter
M15 range breakout with H4 200 EMA trend filter, 6-level trailing stop, and failed-breakout recovery.
H1 range breakout · proven track record
The original Goldie Razor — H1 breakout strategy with a long live track record on XAUUSD.
Triple-confirmation · low frequency
Low-frequency sniper requiring triple signal confirmation — fewer trades, higher per-trade selectivity.
Manual entry · automated exits
You control entries; the EA manages exits, trailing stops, and position management automatically.
Full Pro-Scalper suite
All five Expert Advisors at a bundle price — cover every market condition and trading style.
Goldie Razor V2.8.4
M15 breakout + H4 EMA filter — built for XAUUSD on MT5