Blog ·

How to choose a gold EA (XAUUSD) for MetaTrader 5: 8 checks

Gold is the favourite market for EAs: it moves a lot, spreads are reasonable and it trades almost 24 hours a day. That is also why it is where the most robots with "too good to be true" results are sold. These are the checks we run before releasing an EA, and the ones we recommend before you buy one.

1. Every trade must have a stop loss

This is rule number one. Many EAs with perfect curves use martingale (doubling the lot after a loss), grid (opening orders every X points with no stop) or averaging. They work for months and, when gold makes a big move against them, they can wipe the account in a day. Look for "no martingale, no grid" in the description and check in the backtest that the size of the losses is limited.

2. Real-tick backtest, not candles

The MT5 Strategy Tester has several modes. "Open prices only" or "1 minute OHLC" are fast but guess what happens inside the bar. For an M1 or M5 gold EA, the only reliable mode is "Every tick based on real ticks", with history quality close to 100%. If the seller doesn't say which mode they used, ask. We show you how to do it yourself in this backtest guide.

3. A period the EA has never seen

Optimising an EA on a period and showing the result of that same period is easy: the inputs are fitted to the past. What matters is how it does on a later stretch that was not used for optimisation (forward or out-of-sample). If the result collapses outside the optimised period, it is overfitted.

4. Drawdown matters more than profit

+80% with a 40% drawdown is worse than +10% with 3%: with the second you can raise the risk under control, with the first you never know when the next fall comes. Look at the maximum equity drawdown (not only balance) and the recovery factor (net profit divided by maximum drawdown). Above 2–3 over a few months of testing is a good sign.

5. A realistic profit factor, and be careful with win rate

A profit factor (gross profit / gross loss) between 1.2 and 2 over hundreds of trades is believable. Values of 5 or 10 with few trades are usually luck or overfitting. Win rate is misleading: an EA can win 95% of the time with small gains and lose everything in one trade. Always compare the average win with the average loss.

6. Sensitivity to spread and broker

On M1 and M5, a few points of spread change the result a lot. A good EA has a spread filter and tells you the maximum spread it is designed for. Before going live, run it in the tester on your broker's data.

7. Required capital and minimum lot

On gold, a 0.01-lot trade with a 2–3 × ATR stop can risk several dollars. On a small account the minimum lot already exceeds the risk you set. Ask the seller what capital they recommend for the default risk. For our EAs it is about 3,000 USD for 0.3% per trade.

8. Live results

The backtest is the starting point. A live signal (MQL5 Signals, Myfxbook) or a demo account you can follow confirms that the EA behaves live as it did in the tester.

How we apply this at JAIAK

The four JAIAK V1 EAs follow these rules: a stop on every trade, no martingale or grid, real-tick tests and a separate forward period. Each one attacks gold differently: Trend Jury (vote-based trend), Angle Shift (momentum), Breakout Rider (trend flips) and Ice Cross (EMA crossover). Each page has the full backtest and its conditions.

The 4 JAIAK V1 EAs

You may also like

Trading leveraged CFDs and forex carries a high risk of loss. Past results, live or simulated, do not guarantee future results. This content is not financial advice.