How much history do I need to backtest gold?

At least three years, which on gold covers a trending phase, a compressive range and at least one shock event. Anything shorter usually samples a single regime.

Do I need tick data for XAUUSD?

For anything with a stop under about three dollars per ounce, yes — bar data hides the wick sequence that decides whether your stop or your target was hit first.

Why does my backtest beat my demo results?

Almost always spread, slippage and rollover. Add realistic costs to the backtest and the two converge; if they still diverge, your execution timing is the difference.

BACKTESTING GOLD

Backtesting gold requires explicit spread, slippage, commission and swap assumptions because XAUUSD trading costs vary dramatically between the London session and the 21:00 UTC rollover.

Results should be split by market regime — trend, range and shock — rather than by calendar date, and validated on out-of-sample data before any live capital is committed.