Trading Systems

I tested four trading ideas before releasing one EA. Here is everything

5 November 2026, 14:52
Fernando Alberto Rodriguez Lozano
Fernando Alberto Rodriguez Lozano
0
18

I wanted to build a gold EA I could sell without having to hide anything. So before releasing Glassbox Gold Breakout, I set myself three rules: test ideas on one period, confirm them on a period I had never looked at, and publish the results whatever they were.


Four ideas, three failures

  1. London session breakout on GBPUSD, USDJPY and EURJPY. 764 trades over 2023–2025, profit factor 0.99. Essentially random. Dropped.
  2. Daily trend following (50-day breakout) on four major FX pairs. Profit factor 0.49 on a 1,000 USD account and 0.75 on 10,000 USD over 2015–2022. It lost money. Dropped.
  3. Gold breakout with three sub-strategies (daily, 4-hour and 1-hour levels). The first run wiped out the test account, because of a position-sizing bug in my own code. After fixing it, the strategy was roughly break-even (profit factor 0.98). The trailing stop was cutting winners short: removing it moved the result to profit factor 1.06, still not good enough.
  4. The same gold breakout, daily levels only, no trailing stop. 2015–2021: 115 trades, profit factor 1.47, +33%, maximum drawdown 12%.


The honest caveat about idea 4

I chose the daily sub-strategy after seeing it perform best. That is exactly how curve-fitting happens. So I ran it once, unchanged, on data I had not used: 2022–2023.

Result: 33 trades, profit factor 1.26, +5.1%, maximum drawdown 8.9%. In 2022, when gold went sideways, it lost about 4%. In 2023, when gold rose, it made about 9%. Short trades lost money in both periods.

That is a pass, but a narrow one. And when I compared it with simply holding gold over the same years, the results were similar after adjusting for drawdown, not better.


So why release it?

Because what it does well is real: every trade has a fixed stop loss and take profit, there is no grid or martingale, drawdowns stayed small, and the rules are simple enough to understand and verify. Many gold EAs look better on paper and hide how they got there. This one shows you the whole path, including the parts that do not flatter it.


How to check it yourself

Glassbox Gold Breakout prints a results report at the end of every backtest: results by year, by direction (long/short) and by sub-strategy. Download the free demo, run XAUUSD on H1 with the default settings, and compare your numbers with mine.

Product page: https://www.mql5.com/en/market/product/206399

If you have questions about the testing method or the results, ask in the comments. I would rather lose a sale than have someone buy it expecting something it is not.


Trading gold on margin carries a high risk of loss. Past and backtested results do not guarantee future results.

#gold, xauusd, breakout, expert advisor, backtest, Out-of-Sample