Why do currency-pair settings fail on gold?
Because they are usually expressed in fixed pip or point distances, and gold's typical range is a large multiple of a major pair's — so the same number means a completely different thing.
A 25-pip stop on EUR/USD is a normal intraday distance. The equivalent number of points on XAUUSD sits well inside ordinary noise, and a strategy carrying it will be stopped out constantly by movement that means nothing.
The fix is the same one that makes any strategy portable: express every distance as a multiple of ATR on the operating timeframe. The rule then adapts to whatever the instrument is doing rather than to what EUR/USD was doing when the parameter was chosen.
This also removes a subtler problem within gold itself. Its volatility regime shifts substantially over the years, so even a gold-specific fixed distance tuned on one period will be wrong on another.
What is different about gold's spread?
Gold's spread is wider relative to its typical move than a major pair's, and it widens more dramatically around the events gold actually responds to — which are the moments many strategies want to trade.
Scheduled macro releases move gold sharply, and spreads there can widen by a large multiple for a period long enough to matter. A backtest carrying a single average spread has modelled a market that is calm at exactly the moments it is not.
The daily rollover deserves the same attention it does in forex, and gold's rollover widening is typically more pronounced. A strategy with trades in that window and a flat spread assumption is being credited with fills nobody was quoting.
Measuring your own broker's gold spread by hour over several weeks and feeding that profile to the tester is the single highest-value change available, and it is an afternoon's work.
Does gold have a session structure?
It does, and it is not the same as the currency session structure most strategies encode — gold responds to the London fix, to US data releases and to physical market hours in ways that do not line up with an FX session template.
A strategy carrying a "London session" filter written for EUR/USD is applying a window defined by currency market participation to an instrument with a partly different set of participants. The filter may still help; the reason it helps is not the reason encoded in the comment.
The empirical approach beats the inherited one here. Bucket gold's historical moves by hour and look at where the volatility and the directional persistence actually sit, rather than importing a session definition from another market.
As with any such split, each candidate window is a trial. Testing eight session filters and reporting the best is a search, and the winner needs an out-of-sample check before it is a finding.
What about the contract and the data?
Spot gold at a retail broker, gold futures and a gold ETF are three different instruments with different prices, different costs and different trading hours, and a backtest on one is an estimate for another.
Spot XAUUSD at a CFD broker is a contract with that broker, priced from their aggregation, carrying their spread and their overnight financing. Two brokers' XAUUSD series will differ more than two brokers' EUR/USD series.
Futures have real exchange volume and a roll schedule, and a continuous futures series is a construction: the method used to join contracts at each roll changes the historical prices, and a strategy sensitive to exact levels is sensitive to that choice.
Overnight financing is the cost most often forgotten on spot gold. It applies to every position held across the rollover, it compounds for multi-day holds, and a swing strategy tested without it is missing a recurring charge on its longest and most profitable trades.
Check the contract size your tester assumes as well. Gold is quoted per ounce and brokers differ in what one lot represents, so a position size that looks conservative in lots can be a much larger notional than intended — which affects both the financing charge and the drawdown the account actually experiences, without changing a single number in the strategy.