Asian range breakout on gold: how it works and fails

An Asian range breakout marks gold’s high and low in quiet hours and trades the first clean break of either edge; it fails on breaks that come back, a range drawn at the wrong hour and the entry spread.

• An Asian range breakout marks gold’s high and low during quiet hours, then trades the first clean break of either side, usually around the London open.
• What breaks it is mostly mundane: breaks that fail and come back, a range drawn at the wrong hour because the broker’s clock moved, and the spread at the moment of entry.
• Test it in hours, not in dollars. Most of what goes wrong is a timing error you can see in the tester’s deal list.

Search for a london breakout EA gold and page one is vendors and code samples, most of them describing the setup in a sentence and the results in a table. This guide is about the part in between: what the Expert Advisor is actually measuring, and the three places it goes wrong on XAUUSD.

How the Asian range breakout works

The idea has two halves. During the hours when most of the gold market is asleep, price tends to drift inside a band. The EA records the high and the low of that band over a fixed window of hours. Then, when a busier session begins and volume arrives, it trades the first move that clears one edge: a buy above the high, a sell below the low. The London open is the classic trigger, which is why the two names are used for one strategy.

Most versions place a pair of pending orders — a buy stop above the range and a sell stop below it — and cancel whichever is left once the other fills, or once a deadline passes. The stop usually sits on or inside the far side of the range, so the width of the range sets the risk of the trade.

Every part of that description depends on the clock. The range is “the high and low between two times”, the trigger is “a time”, the deadline is “a time”. That is where the trouble starts.

What breaks it, part one: the break that comes back

A breakout strategy is a bet that leaving the range means something. Often it does not: price pokes through the edge, fills the stop order, and falls back inside. The families guide puts it plainly — a market that keeps poking out of the range and coming straight back costs a small loss each time.

So expect a low win rate from any honest breakout EA, and judge it on whether the winners are larger than the losers. Our own session strategy is the example. In its Strategy Tester run, 1 January to 7 September 2026 on a $25,000 deposit, Dawnbreak won 39.39 % of 556 trades, with an average win of 499.86 against an average loss of −287.69, and a longest losing run of 15 trades. Most of its breaks failed. A buyer who expected otherwise would have switched it off in the first losing fortnight.

What breaks it, part two: the broker’s clock

A MetaTrader EA reads the trade server’s clock, not yours and not London’s. TimeCurrent() “returns the last known server time” (MQL5 documentation), and a range defined as “00:00 to 06:00” means server hours.

Some brokers move that clock with US daylight saving. IC Markets, whose data our tests use, wrote that from 8 March 2026 “the server time will be adjusted from GMT+2 to GMT+3”, and that “the change to GMT+3 may affect your EAs” (IC Markets).

The US and the UK do not change on the same day. US daylight saving begins “on the second Sunday of March” and ends “on the first Sunday of November” (NIST); UK clocks go forward “on the last Sunday in March” and back “on the last Sunday in October” (GOV.UK). On a server that runs at GMT+2 and moves to GMT+3 on the US date, 08:00 in London is 10:00 server time in winter and in summer, and 11:00 for the weeks in between: in 2026, 8 to 29 March and 25 October to 1 November. An EA hard-wired to 10:00 trades an hour early for those weeks.

Why a backtest will not warn you about the clock

The obvious fix is to compute London time from GMT. In a backtest that fix does nothing: “During testing in the strategy tester, TimeGMT() is always equal to TimeTradeServer() simulated server time” (MQL5 documentation). An EA that converts from GMT behaves one way in the tester and another on a live account.

So test the clock by eye. Run the visual tester over the second and third weeks of March and the last week of October, and read the entry times in the deal list against the hour the strategy claims to trade.

What breaks it, part three: the spread at the entry

A stop order above the range fills as the move starts, which is when the spread is least likely to be at its tightest. The families guide notes that a breakout entry is taken at the worst moment for the spread, by definition, and the tighter the range, the larger the share of the stop that spread takes.

The tester does not invent a spread for you: “During testing, the spread is not modeled but is taken from historical data” (MQL5 documentation). That makes your own broker’s ticks the only honest test. A breakout backtested on one broker’s spreads says little about another’s.

Where Dawnbreak sits

Dawnbreak, our session breakout EA, is a relative of this setup, not a copy of it. It sits out the opening of each trading day, lets gold show where it is willing to trade, and opens a single position when price breaks clear of that early zone and the wider daily direction agrees. Its exits are fixed when the trade opens, and it is flat before the trading day ends. Because it is anchored to the start of the broker’s trading day rather than to a London clock time, the hour that matters for it is where your broker’s day begins, so check that on your own server before anything else.

Its free Backtest Edition runs in the Strategy Tester on your own broker’s ticks, from the strategies page.

Before you act on any of this

Every figure above comes from the MetaTrader 5 Strategy Tester over 1 January 2026 to 7 September 2026 on IC Markets demo data, never from live trading, and no result is promised anywhere on this site. Two departures apply to all of them: the tester charged no commission on XAUUSD, while real trading on that account cost roughly $7 per lot round-turn that the net does not deduct, and fills are modelled rather than executed. History Quality was 63 % because real ticks begin on 21 April 2026. Before you run any breakout EA, read the risk disclosure.