EA Autopsy Report
Three fictional EAs that failed — and the red flags that were visible before purchase. Click each to expand.
Red Flags in the EA Itself:
What to Check Before You Buy
Published 17 June 2026 · 12 min read
The red flags that matter most are in the EA's strategy and backtest — not just the seller's marketing. The non-negotiables: a hard stop loss on every trade, no martingale lot scaling, a backtest covering 3+ years across multiple market regimes, and verified live results on a third-party platform. An EA that fails any of these tests should be disqualified regardless of how impressive the claims are.
The 4 Categories of Red Flags Inside an EA
Red flags inside an EA fall into four distinct categories. Understanding which category a flag belongs to helps you prioritise what to check first.
Category 1: Risk Management
Critical — disqualifying if present
- › No hard stop loss
- › Martingale or grid lot sizing
- › No maximum drawdown limit
- › Single trade can exceed 5% of account
Category 2: Backtest Integrity
High — requires explanation
- › Test period under 2 years
- › Only tests one market regime
- › Zero-spread assumption
- › Suspiciously smooth equity curve (no drawdowns)
Category 3: Strategy Transparency
Medium — evaluate case by case
- › No strategy description at all
- › Cannot explain when EA enters or exits
- › No logic for filter conditions
- › Refuses to run a different date range
Category 4: Live Proof
Medium — reduces confidence
- › Only demo results shown
- › Myfxbook unverified
- › Results from only one broker
- › No trades in drawdown periods shown
Why No Stop Loss Is the Ultimate Disqualifier
The single most important thing to check in any EA is whether it uses a hard stop loss on every individual trade. This is non-negotiable — not because losing trades are acceptable (they are), but because the alternative is categorically more dangerous.
EAs without hard stop losses typically use one of two approaches: they hold losing trades indefinitely hoping for a reversal (often called "no-SL grid" or "hedge grid"), or they add to losing positions at increasing lot sizes (martingale). Both approaches have the same fundamental problem: the losses compound mathematically with no defined ceiling.
In backtests, these approaches look extraordinary. The win rate is 95%+ because the EA rarely closes a losing trade — it just waits. The equity curve is perfectly smooth because drawdowns are hidden in open floating losses that the backtest doesn't show correctly. The reported return is often 500%+ per year.
In live trading, this works until there is a sustained directional move — a geopolitical event that sends gold 400 pips in one direction, an emergency rate cut, a flash crash. The open losses at increasing lot sizes create a margin call. The account can drop 60–100% in a single session. This scenario is not rare; it happens to multiple traders running martingale EAs every year.
The standard: any EA should have a hard stop loss on every trade.
Goldie Razor V2.8.4 uses a hard SL on every individual trade — this is the baseline expectation for any EA you should consider. There is no scenario where an EA without a hard SL is worth the risk, regardless of its backtest performance.
EA Red Flag Scan
Work through this 10-point checklist for any EA you are evaluating. Check each item only if you can confirm it is true — not just assumed. Items you cannot confirm count as red flags.
How to Verify Each Point Before Buying
Checking these items is not abstract — there are specific things to look for in the MT5 Strategy Tester report and in the EA parameters.
Checking for a stop loss
In the backtest report, look at the "Largest profit trade" and "Largest loss trade" values. If the largest loss is many times larger than the average loss, that is a sign of no hard SL. Also check the EA parameter list for any "StopLoss" or "SL" parameter — if it is set to 0 or does not exist, the EA has no stop loss.
Checking for martingale
Look at the equity curve shape. Martingale EAs produce characteristically serrated curves that go steadily up with occasional deep drops. Also look for parameters named "LotMultiplier", "Multiplier", "RecoveryFactor", or similar. Any parameter that scales lot size based on previous trade results is martingale logic.
Evaluating the backtest date range
Right at the top of the MT5 Strategy Tester report, there is a "Testing period" field. Check that it covers at least 2020–present at minimum, and ideally 2018–present to include both a strong trending period (2020) and ranging periods (2022). The quality setting should show "Every tick based on real ticks" or "99% modelling quality" — not "Open prices only".
Verifying the spread setting
In the Strategy Tester report, find the "Spread" field under the test parameters. If it shows 0 or a very low number (1–2 points), the test used an unrealistic spread. Real XAUUSD spreads during normal trading are 10–20 points (1–2 pips). During news events they can reach 50+ points. A 0-spread backtest is not a useful prediction of live performance.
Confirming live results are verified
The vendor should provide a direct link to a Myfxbook or FXBlue page (not a screenshot). Visit that page directly, check the "Verified" status, look at the growth chart (not just the summary statistics), and check whether there are periods of significant drawdown that match what is claimed.
Related Reading
Red Flags in the Seller's Behaviour
Separate guide: deceptive marketing, fake reviews, and misleading claims — distinct from flaws inside the EA itself.
Verifying the Backtest Quality
How to read an MT5 backtest report and identify signs of curve-fitting or unrealistic assumptions.
Questions to Ask About Each Red Flag
The exact questions to send a vendor that expose whether these red flags are present.
Verifying the Proof of Results
How to distinguish genuine verified results from fabricated or selectively shown performance.
Documentation to Request Before Buying
The minimum documentation package that any legitimate EA vendor should be able to provide.
Frequently Asked Questions
No hard stop loss combined with martingale lot sizing. This combination produces EAs that look statistically impressive in backtests — high win rates, smooth equity curves, minimal drawdown — because the EA never takes a realized loss; it just keeps adding to losing positions at larger lot sizes. The account survives until it doesn't. A single adverse news event (a 200-pip spike, a flash crash) hits all the stacked positions simultaneously. The resulting loss is catastrophic, sometimes wiping 80–100% of the account in hours. Any EA without a hard stop loss on every individual trade should be disqualified immediately.
Three checks: First, look at the backtest date range. If it only covers 2020-2021 (bull market) or another specific favorable period, it's suspicious. Second, look for suspiciously high parameter counts in the EA settings — many parameters mean many opportunities to over-optimize. Third, ask the vendor to run a test on a different date range (for example, 2016–2018) and see if the results hold up. A curve-fitted EA performs well only on the period it was optimized for.
Myfxbook or FXBlue tracking with a verified broker feed — not just screenshots from MT5. The key is third-party verification: the statistics come from an independent platform that connects directly to the broker, not numbers that the vendor calculated and typed into a spreadsheet. Look for: minimum 6 months of live data, consistent with the claimed strategy (number of trades, win rate, drawdown match the description), and ideally multiple accounts across different brokers. One account with 2 months of live data is insufficient evidence.
Not necessarily — more parameters can mean more flexibility for different market conditions. It becomes a red flag when: (1) the vendor provides default settings that were clearly optimized on the backtest period, (2) there is no guidance on how to set parameters for different conditions, or (3) the backtest only shows results with one specific parameter set without testing robustness across different settings. Robust EAs tend to work reasonably well across a range of parameter values — not just the exact optimized set.
You can ask, but do not expect it. Most commercial EA developers protect their source code — this is legitimate intellectual property protection, not a red flag. What you should have access to: the compiled .ex5 file (to run the EA), the strategy description, backtest reports (full, not screenshots), and ideally a demo license to test before buying. If the vendor refuses to share any backtest details, refuses to explain the strategy concept, and refuses to provide verified live results — that combination is a red flag regardless of whether they share source code.
Goldie Razor V2.8.4
M15 breakout + H4 EMA filter — built for XAUUSD on MT5