The 9/21 EMA crossover with RSI confirmation and ATR sizing is the most-recommended beginner gold strategy on the internet. We included it deliberately — to show, with data, why 'obvious' isn't the same as 'edge'.
A textbook long/short crossover of the 9- and 21-period EMAs on XAUUSD M15, filtered by RSI and sized by ATR. Entry on bar close, stop beyond recent structure. No curve-fitting, no hidden parameters — the version a beginner would actually run.
Across 8 anchored out-of-sample folds the strategy held up in only 3. Profit factor came in at 0.81 (below the 1.0 break-even line) and the win rate of 44% wasn't high enough to overcome spread and the mediocre reward-to-risk. In plain terms: it loses money once you account for real-world costs and unseen data.
Crossovers are lagging by construction — they confirm a move after the easy part is over, which on gold's fast M15 tape means you buy the top of the impulse and sell the bottom. Without a liquidity or session-context filter, the signal fires constantly in chop, and chop is where XAUUSD spends most of its time.
This is the baseline every other experiment is measured against. A raw indicator crossover is a starting point for learning to code an EA — not a deployable edge. The strategies that passed our walk-forward all added structure (liquidity, session timing, higher-timeframe alignment) on top of a signal, not a signal alone.