Subscribe to our channel, here you will find the best 👇
The retail market is full of gold robots that promise precision, stability, and “smart” automation, yet most reviews still make the same mistake: they compare headline profit without examining how that profit was produced. You can learn about our approach to testing expert advisors on tick history with a real spread on the relevant page – Our principles.
That is the wrong way to evaluate a gold trading system.
In this article, I compare three MT4 expert advisors for XAUUSD — Gold or MT4, Secret Impulse, and CoreX G — using independent Tick Data Suite reports with 99.90% modeling quality and real variable spread. The goal here is not to praise any of these systems. The goal is to answer a harder and more useful question:
Which backtest looks profitable, which one looks believable, and which one looks dangerous despite attractive statistics?
That distinction matters because gold is one of the most execution-sensitive markets in retail trading. XAUUSD can produce strong directional moves, violent reversals, and fast spread changes during news. A robot that looks excellent in a clean backtest can still fail in live trading if its edge is too thin, too optimized, or too dependent on perfect fills.
Why raw profit is a misleading comparison
Before looking at the three systems one by one, we need to remove the first analytical trap.
At first glance, Gold or MT4 appears to crush the comparison with $33,549.96 net profit, versus $7,105.53 for CoreX G and $2,557.77 for Secret Impulse. But that raw ranking is distorted by position size.
Gold or MT4 was tested at a fixed 0.10 lot, while Secret Impulse and CoreX G were tested at fixed 0.01 lot. That means Gold or MT4 was effectively running at ten times the nominal trade size of the other two systems. So a serious comparison cannot stop at raw dollars.
Once you normalize profit to a 0.01-lot equivalent, the picture changes:
- Gold or MT4 falls to roughly $3,355
- Secret Impulse stays at $2,557.77
- CoreX G stays at $7,105.53
That does not automatically make CoreX G the best robot. It simply means the raw profit leaderboard is not trustworthy on its own.
This is exactly why professional analysis starts with structure, not marketing-style outcome.
Gold or MT4: the most believable profile, but not the most efficient one
Gold or MT4 produced the strongest headline result in the test:
- Net profit: $33,549.96
- Profit factor: 1.54
- Relative drawdown: 26.38%
- Total trades: 785
- Win rate: 61.02%
Its first advantage is that the report looks structurally coherent. Average profit trade was about $199.64, average loss trade about -$202.87, so this is not a system that wins through favorable risk/reward asymmetry. It wins because it is correct often enough, not because its winners are dramatically larger than its losers.
That matters.
A system with nearly symmetric payoff distribution is easier to read. There is no illusion here. If the win rate drops, the edge deteriorates quickly. In that sense, the backtest looks more like a real trading strategy and less like a marketing artifact.
Another important detail: despite being presented as a “scalper,” the report behavior does not look like classical scalping. The robot traded on H1, held positions for roughly 30 hours on average, and never stacked multiple positions at once in the test log. That is much closer to a short-term directional swing model than to a true scalp engine.
This is actually a positive sign from an analytical standpoint. Scalping claims are often used loosely in retail EA marketing. Here, the report itself shows a slower and more understandable trading structure.
But Gold or MT4 also has clear weaknesses.
First, the profit factor of 1.54 is respectable, not exceptional. It suggests a moderate edge, not a dominant one. Second, the 26.38% relative drawdown is meaningful. For gold, that level of drawdown is not shocking, but it is large enough to matter both financially and psychologically. Third, because average win and average loss are almost identical, this system depends heavily on maintaining its hit rate. If changing market regimes push the win rate even modestly lower, the strategy quality can deteriorate fast.
So what is the verdict?
Gold or MT4 is not the cleanest robot in the comparison, but it may be the most believable. Its statistics look like those of a normal, imperfect, real trading system. That is often worth more than a backtest that looks almost too perfect to question.
Secret Impulse: profitable on paper, but structurally fragile
Secret Impulse is the most active system in the group:
- Net profit: $2,557.77
- Profit factor: 1.77
- Relative drawdown: 25.00%
- Total trades: 2,594
- Win rate: 67.46%
- Expected payoff: $0.99 per trade
At first glance, the profit factor of 1.77 looks better than Gold or MT4. The win rate is also higher. A superficial reviewer might stop there and call it stronger.
That would be a mistake.
The real problem with Secret Impulse is not whether it made money in the backtest. The problem is how thin its edge is on a per-trade basis.
An expected payoff of $0.99 per trade on XAUUSD is not much. Average profit trade was just $3.35, while average loss trade was -$3.91. In other words, the strategy does not have positive asymmetry. It needs a reasonably high win rate just to function, and even then the monetary edge per trade remains extremely small.
That profile is dangerous on M1.
The system traded on the 1-minute timeframe, executed 2,594 trades, and the median holding time in the report was around 35 minutes. On top of that, the log shows as many as six simultaneous open positions, which means this is not just a simple one-entry scalper. It can stack exposure.
The exit structure is also revealing. Only a minority of trades were closed by take profit, and almost none by stop loss. Most positions were closed through generic “close” actions rather than clean TP/SL resolution. That suggests a more adaptive or internal exit logic, but it also makes the strategy harder to evaluate and easier to degrade in live conditions.
This is where the testing limitation becomes critical: real spread was used, but slippage was not modeled in the published setup. For an H1 robot, that is a limitation. For an M1 gold robot with a thin per-trade edge, it is a major limitation.
A strategy making roughly one dollar of expected value per trade in backtest conditions can lose most of that edge through live execution friction, especially on XAUUSD during volatile sessions.
So how should Secret Impulse be interpreted?
It is not a fake-looking backtest. The growth curve is gradual and the drawdown is not absurd. But this is the system I would call the most fragile in the comparison. It depends on high-frequency execution, small edge capture, and a market where costs can expand suddenly. That is not a comfortable combination.
CoreX G: the most impressive numbers and the highest verification burden
CoreX G delivered the most extraordinary statistics in the entire comparison:
- Net profit: $7,105.53
- Profit factor: 20.31
- Relative drawdown: 6.54%
- Total trades: 864
- Win rate: 98.38%
- Only 14 losing trades across nearly five years of testing
If you look only at the report summary, CoreX G is the clear winner. On a normalized 0.01-lot basis, it outperformed both Gold or MT4 and Secret Impulse. Its equity curve is extremely smooth. Every month in the tested period finished positive. The drawdown remained low. The win rate was almost absurdly high.
And that is exactly why a serious trader should slow down here.
The question is not whether these numbers are attractive. Of course they are. The question is whether such a profile is robust enough to survive outside the specific historical sample, broker feed, and test configuration used in the report.
CoreX G uses a classic high-win-rate structure: the average winning trade was about $8.79, while the average losing trade was -$26.28. That means the strategy accepts losses roughly three times larger than its average win and compensates through a very high hit rate. In principle, that can work. Many mean-reversion and precision-entry systems are built that way.
But a 98.38% win rate on gold over 864 trades is not just strong. It is statistically extraordinary for a public retail EA. Add the extremely smooth equity curve and the absence of losing months, and the required level of skepticism rises sharply.
The trade log shows up to two simultaneous positions, not a deep grid, but still some layering. The parameters also show fixed lot trading rather than auto lot escalation, which means the result was not artificially boosted by aggressive compounding. That makes the performance look even more striking.
There are two ways to read this backtest.
The optimistic interpretation is that CoreX G is a highly selective, unusually stable XAUUSD system.
The critical interpretation is that the report shows a profile that is consistent with a heavily optimized historical fit and therefore requires much more validation than the other two robots.
I lean toward the second interpretation, not because the result must be false, but because the burden of proof is much higher when a backtest becomes this clean.
This issue becomes even more important because the product is described with AI, neural-network, and machine-learning language. That kind of branding may sound sophisticated, but it does not reduce model risk. In trading, “AI-powered” is not evidence. Robustness is evidence.
Which robot has the most credible edge?
If the goal is to identify the most believable backtest rather than the prettiest one, the ranking changes.
Gold or MT4 has the most ordinary and therefore arguably the most credible profile. Moderate profit factor, meaningful drawdown, realistic win rate, and no perfect-looking smoothness. It behaves like a normal H1 gold system with an imperfect but visible edge.
Secret Impulse has a thinner edge than its profit factor suggests. The M1 timeframe, very small expected payoff, multi-position stacking, and dependence on unmodeled slippage make it the most execution-sensitive and fragile of the group.
CoreX G has the best report on paper, but also the highest probability of disappointing in live or out-of-sample conditions if the historical fit is too specific. The stronger the backtest, the stronger the required skepticism.
That is the paradox many retail traders miss.
The system with the most believable backtest is not always the one with the highest historical return. And the robot with the smoothest curve may deserve the most doubt, not the most trust.
Final verdict
From a professional trading perspective, these three XAUUSD EAs fall into three very different categories.
Gold or MT4 is the most believable. It does not look magical, but it does look understandable. Its main weakness is efficiency: moderate profit factor, meaningful drawdown, and dependence on win rate.
Secret Impulse is the most fragile. It made money in the report, but the edge per trade is too thin to inspire confidence on M1 gold without aggressive forward validation under real execution.
CoreX G is the most impressive and the most suspicious at the same time. If the report reflects genuine robustness, it is the strongest system here by a wide margin. But until that profile is confirmed with out-of-sample data, broker variation, and live-forward stability, it should be treated as the system with the highest verification burden, not the safest choice.
So if the question is, “Which gold EA looks best in backtest?” the answer is probably CoreX G.
But if the real question is, “Which backtest would I trust first as a trader who understands market noise, execution risk, and overfitting?” the answer is Gold or MT4.
And that is the more important question.
Conclusion
A profitable backtest alone is not enough to justify trust in a gold trading EA. In the XAUUSD segment, where volatility, spread expansion, and execution quality play a major role, the real value of a system is defined not by headline profit, but by the durability of its edge.
In this comparison, Gold or MT4 appears to offer the most realistic and structurally believable profile, even if its performance metrics are far from exceptional. Secret Impulse remains profitable on paper, but its low expected payoff per trade makes it highly vulnerable to live-market friction. CoreX G delivered by far the strongest historical result, yet its unusually smooth statistics and extremely high win rate demand the highest level of caution and additional validation.
For traders evaluating gold robots seriously, the main takeaway is simple: do not rank EAs by profit alone. A robust system should be judged by the relationship between return, drawdown, trade structure, execution sensitivity, and the likelihood that its historical edge can survive in live conditions. In that context, this comparison shows that the most attractive backtest is not always the most trustworthy one.
Subscribe to our channel, here you will find the best 👇