This EA Super Trend XAUUSD review is about a model that only knows the years it was trained on. From May 2024 to now
the backtest climbs almost every month. Run 2022 and 2023 with the same EA and the same settings, and it breaks even with a
35 % drawdown.

This one is in the second group. There are no stored answers like in
Quantum iGold. The gap is between what the page says the EA does and
what it actually does.

Tested: EA Super Trend XAUUSD v3.0, the Market demo, 31 October 2026, real ticks from Pepperstone and Exness.

What does the page say?

$299, 45 reviews at 3.29 stars. Gold on H1, "2 ONNX models based on SuperTrend". No grid, no martingale, and that part
checks out: fixed 50 USD stop and target, a trailing stop, at most 3 trades open.

Is SuperTrend doing anything?

No. I changed the SuperTrend ATR period and multiplier from 10 × 3.0 to 20 × 5.0, then to 5 × 1.0. All 121 trades came
out identical. Same prices, same exits. The line is only drawn on the chart.

The trade direction matches SuperTrend on H1 just 33 % of the time. Other settings and timeframes didn't line up either.

EA Super Trend XAUUSD review: when does it make money?

Same EA, defaults, 0.01 lot per position, no swap:

Year Profit (0.01 lot) Win rate
2022 −132 USD 53 %
2023 +66 USD 54 %
2025 +7,771 USD 86 %
2026 (to 23 Oct) +11,028 USD 89 %

2024 is the one to look at. January to April loses, then from May it goes up nearly in a straight line:

EA Super Trend XAUUSD backtest 2024, Exness real ticks on a 2-digit custom gold symbol, 1,000 USD start, auto lot 0.01 per 1,500 USD: flat January to April, steady climb from May

The author said in the comments he optimised on 2024 to now. So thats exactly where the profit is. Bigger gold moves
explain part of the dollar gap. They don't explain a win rate jumping from 53 % to 89 % with the same exits.

One more hint. Version 2.0 was retrained on 2020 to now. Six days later the author pulled it because it didn't meet
expectations, and v3.0 brought back v1.0. More training data, worse results. That's how a model fitted to one period usually behaves.

What about the live signal?

The author's signal, started 24 September 2026, shows 70 % algo trading. So 30 % of the trades were manual. On 31 October the balance
showed 0.00. A 100 USD account running the EA was reported blown on 30 October.

Verdict

I wouldn't buy it. All the edge comes from the direction the model picks, and outside its window that direction had none.

Want to check yourself? Run the demo on H1, defaults, 2022 and 2023, on gold with 2 decimals. If you dont see the 2025
kind of curve, you've seen the problem.

The two kinds of problem in this series: MQL5 EAs I wouldn't buy.