How Gold EA Spread Cost Really Works — and Why Backtest ≠ Live

Gold (XAUUSD) is the most popular symbol for EAs — and the one whose execution cost is most easily underestimated. Many people trade on a beautiful backtest, then find live profit shrinks sharply. The culprit is usually spread cost. Here's the true cost of a gold trade, broken down.

The true cost is more than the spread

Every gold trade's real cost has four parts:

• Spread: the bid-ask gap, paid the instant you open. Gold is usually ~1.5–3.0 points (varies by broker and session) and can blow to dozens around news.
• Commission: ECN/raw-spread accounts have a tight spread but charge commission separately, typically a few dollars per lot per side.
• Slippage: the gap between your requested and filled price, worst in fast markets and on stop orders.
• Swap: overnight interest by direction; gold longs are usually negative and add up the longer you hold.

How to actually calculate it: an example

One standard XAUUSD lot = 100 ounces; a 0.01 (one-point) move is about $1 P/L on 1 lot. Say the spread is 2.5 points:

• Spread cost on 1 lot ≈ 2.5 × 1 = $2.5, carried the moment you open;
• On an ECN account at a 0.5-point spread + $3/lot/side commission, cost ≈ 0.5 + 6 = $6.5 round-trip — but the spread is stable;
• So the trade must first climb out of a $2.5–6.5 hole before it's truly profitable — that's your break-even spread.

For a trend strategy taking a few trades a day, this is trivial; but for a scalper taking dozens a day, cost is lethal: 0.2 extra points per trade × thousands of trades is a large slice of profit quietly eaten. That's why gold scalpers are so picky about broker spreads.

Why the backtest is pretty and live shrinks

Backtests default to a fixed spread and ideal fills, but live spread floats: it widens in thin Asian hours, around news, and at month-end — and slippage is inevitable. If the backtest used a 1.0-point fixed spread and live averages 2.5 points + slippage, a high-frequency curve will sit clearly below the backtest. A correct backtest must use real tick data plus spread/commission close to live (how: backtest guide) — otherwise you're validating a world that doesn't exist.

How to cut the cost

• Pick a low-spread broker: for high-frequency gold, prefer a raw-spread + commission ECN account — total cost is usually lower and steadier over time. We recommend Tickmill (low spreads + good history data, the community's common backtest/live environment).
• Avoid high-cost windows: spreads are worst around news and in thin Asian hours; let the EA's news filter dodge them (whether to pause: here).
• Compute the break-even spread first: before picking an EA, ask its average profit target (how many points). If a trade targets only 3–4 points while your spread is 2.5, the strategy barely works on your account.

When choosing a gold EA, make "can it still profit at my account's real spread" a hard criterion (how to choose: picking a gold EA). Pick one with a published live signal from the store (use WELCOME10 for 10% off) — its signal numbers were earned at a real spread.

Risk note: this is an execution-cost explainer, not investment advice. Spread, commission and slippage vary by broker and market; your actual cost is whatever your account shows. EA trading is high risk; past performance doesn't represent future returns. Only trade with money you can afford to lose.