Trading Strategies

London Breakout on Gold: 3.7 Years of Backtests, What Works and What Does Not

13 October 2026, 15:34
Shanawaz Mohammed Nabi Rasool Pinjori
Shanawaz Mohammed Nabi Rasool Pinjori
0
48

Gold (XAUUSD) is quiet during the Asian session and usually makes its first real move of the day when London opens. I tested a simple rule set built around that idea for 3.7 years of real-tick data. Here is what worked, what did not, and the numbers, including the bad parts.

The rules

  • Measure the Asian range: high and low from 03:00 to 08:00 server time.
  • At 08:00 place a Buy Stop above the high and a Sell Stop below the low, with a small buffer. When one fills, the other is cancelled (OCO).
  • Stop loss at the opposite side of the range, capped at 0.6% of price so wide days cannot blow up the risk.
  • Take profit at 2.5R. Move the stop to break-even at 1.5R.
  • Close everything at 21:00. Maximum one trade per day.
  • Skip the day if the range is smaller than 0.30% or larger than 1.60% of price.

Results (backtests, not live trading)

MT5 Strategy Tester, real ticks, January 2023 to September 2026, $5,000 start, 1% risk per trade:

  • Broker A data: net +$6,860, profit factor 1.38, maximum equity drawdown 17.3%, 715 trades, about 49% winners.
  • Broker B data (second check): net +$6,619, profit factor 1.37, maximum equity drawdown 18.8%, 716 trades.
  • At 0.5% risk: net +$2,091, profit factor 1.32, maximum drawdown 11.2%.

The settings were chosen on 2023-2025 data only. On the unseen 2026 data the profit factor was 1.61 over 157 trades. That is a good sign, but 157 trades is still a small sample.

What did not work

  • Tight stops inside the range. Stops at the middle of the range were hit by normal noise far too often.
  • No range filter. Very narrow ranges produce many false breakouts; very wide ranges mean the move already happened in Asia.
  • Holding overnight. Letting trades run past the New York session added swap and gap risk without improving the results.
  • Higher reward targets. 5R looked attractive but the profit factor dropped to 1.30 and drawdown went up.

A bug worth knowing about: lot size on gold

While testing on a second broker I found my first version opened trades about 10 times too large. The cause: SYMBOL_TRADE_TICK_VALUE returned a value that did not match the real contract on that symbol. The fix is to calculate the money lost per lot at the stop with OrderCalcProfit() and size from that, with the tick value only as a fallback. If you write gold EAs, check the loss of your very first test trade against your risk setting.

Honest limits

  • Drawdowns of 15-19% at 1% risk are normal for this system. For prop firm accounts with a 10% total limit, use 0.3-0.4% risk.
  • Long flat periods happen. The edge is small and needs hundreds of trades to show.
  • Session times depend on your broker's server time and should be checked once.

I am now forward testing it on a demo account and will post the results here. The EA is available in the Market as Gold London Range Breakout, and a free Session High Low Boxes indicator draws the same Asian range on your chart if you prefer to trade it manually.

Questions about the rules or the tests are welcome in the comments.

#gold, xauusd, breakout, london session, backtest