Skip to content
EA
Ranking

We backtested 19 expert advisors through the 2025 gold squeeze. 2 came through.

Between June 2025 and August 2025, gold moved far enough in one direction to end a lot of expert advisors that had never had a losing month. We ran 19 of the EAs listed on this site through that window on real tick data, at the vendor default settings, and ranked the survivors by return against maximum drawdown. 2 were wiped out. 10 ended down or flat. 1 finished up but got there through a drawdown past 40% or on an edge too thin to separate from chance. 1 did not trade enough for the result to mean anything. 3 made money with a strategy that has no upper bound on a losing sequence, so they are reported but not ranked. That leaves 2.

Test period
Jun 1, 2025 — Nov 1, 2025 (153 days)
2 products
Backtests we ran ourselves
Last updated

Demo account results. Demo fills are not identical to live fills — treat these as a filter, not as proof.

How this was tested

What was tested

Every expert advisor on this site that we hold a file for, and that we could run unattended. No selection by expected outcome — the list was fixed before any test was run, which is why most of the results below are bad.

The window

All results cover a single window that includes June 2025 to August 2025. This is deliberate. Gold moved far enough in one direction during that period to end a class of strategies that had looked flawless for two years, and a backtest that avoids it proves nothing.

How it was run

  • MetaTrader Strategy Tester, real tick data, every tick based on real ticks
  • Vendor default settings, with no parameter optimisation of any kind
  • Broker, spread, leverage and starting balance are shown per product in the table
  • Modelling quality is shown per product; anything below 99% is flagged

How they are ranked

By return divided by maximum drawdown, not by return — and only after four gates. A product is reported but not ranked if it ended down, if it was essentially flat, if maximum drawdown went past 40%, or if the profit factor came in under 1.20. Those gates decide more than the sort order does: the single highest return in this run is not in the ranking at all, because it was produced through a drawdown deep enough that most accounts would have been closed or margin-called long before the recovery. Ranking by return alone would have put it first. Sorting alone cannot fix that — a bad enough entry still lands in the top few.

What this is not

These are backtests, not live results. A backtest cannot reproduce slippage, requotes, or a broker widening spreads when it matters most. Treat a good result here as permission to test something on a demo account, never as evidence that it works. We use backtests only to exclude.

We backtested 19 expert advisors through the 2025 gold squeeze. 2 came through.

  1. 1

    Quantum OmniGold EA MT5

    MT4 Demo account

    201.5% return against a 13.1% maximum drawdown, over 231 trades.

    Days live
    273
    Return
    201.45%
    Max drawdown
    13.07%
    Trades
    231
    Full data and download →
  2. 2

    Gold Breaker EA MT5

    MT5 Demo account

    62.6% return against a 10.4% maximum drawdown, over 267 trades.

    Days live
    214
    Return
    62.62%
    Max drawdown
    10.36%
    Trades
    267
    Full data and download →

Also tested, not ranked

Publishing only the winners is how every other ranking on this subject is built. These are the rest of the same test run.

  • Dragon Multi EA MT4

    Return -97.1% · Max drawdown 97.15% · Trades 1899

    Account was wiped out during the test window.

  • Buster EA One Shoot XAUUSD MT5

    Return -102.01% · Max drawdown 100.55% · Trades 87

    Account was wiped out during the test window.

  • MFXT PHANTOM X MT5

    Return 262.72% · Max drawdown 94.26% · Trades 3500

    Finished up, but only after the account had been down far enough that most people would have been margin-called or would have switched it off. Recovering from a drawdown that deep is not the same as controlling it.

  • Lock30x Gold Scalper Robot MT5

    Return 1.72% · Max drawdown 1.03% · Trades 366

    Essentially flat — the result is indistinguishable from doing nothing.

  • GOLD FREQUENCY EA MT5

    Return 0.05% · Max drawdown 0.5% · Trades 985

    Essentially flat — the result is indistinguishable from doing nothing.

  • Smart Gold Hunter EA MT5

    Return 0.1% · Max drawdown 0.02% · Trades 161

    Essentially flat — the result is indistinguishable from doing nothing.

  • Capital Guard EA MT5

    Return 0.15% · Max drawdown 0% · Trades 134

    Essentially flat — the result is indistinguishable from doing nothing.

  • Super EA Access MT5

    Return -26.74% · Max drawdown 29.88% · Trades 808

    Ended the window down.

  • GoldTrap EA MT5

    Return -87.79% · Max drawdown 88.22% · Trades 1993

    Ended the window down.

  • Veloz Scalping Bot MT5

    Return -26.77% · Max drawdown 47.67% · Trades 1040

    Ended the window down.

  • EA GTX RnFXs MT5

    Return -76.3% · Max drawdown 88.41% · Trades 3421

    Ended the window down.

  • Gold Killer Pro EA MT5

    Return -30.77% · Max drawdown 30.94% · Trades 12020

    Ended the window down.

  • Gianluca Arrigo EA MT5

    Return -80.01% · Max drawdown 81.29% · Trades 5352

    Ended the window down.

  • Breakout Retest Pro EA MT5

    Return 10.09% · Max drawdown 7.73% · Trades 99

    Too few trades over the window to mean anything.

  • Full Throotle DMX MT5

    Return 149.53% · Max drawdown 12% · Trades 197

    Profitable here, but the strategy has no upper bound on a losing sequence — martingale, grid or averaging logic. Not ranked, on principle.

  • Mavrik Scalper MT5

    Return 19.12% · Max drawdown 8.21% · Trades 466

    Profitable here, but the strategy has no upper bound on a losing sequence — martingale, grid or averaging logic. Not ranked, on principle.

  • Wave Rider EA MT5

    Return 45.64% · Max drawdown 2.41% · Trades 430

    Profitable here, but the strategy has no upper bound on a losing sequence — martingale, grid or averaging logic. Not ranked, on principle.

What this means

The point of running the whole list rather than a shortlist is that the failures are the finding. 2 of 19 were wiped out and 10 ended down or flat — and every one of them is sold today with a marketing page showing an equity curve going up and to the right.

A further 1 finished the window in profit and are still not ranked. Ending up is not the same as being worth running: an account that recovers from a drawdown past 40% spent that window in a state most people would not have sat through, and a profit factor under 1.20 across a large number of trades is not an edge, it is a coin toss paying spread. Both are reported with their full numbers below.

Of the ones that made money, 3 did it with martingale, grid or averaging logic. Those can produce years of clean-looking returns and then give all of it back in a single sequence, which is exactly what the window above was chosen to expose. They are listed here with their numbers, and they are not ranked.

Next: the ones that came through go onto demo accounts with third-party verification, so that the results stop being ours to assert.