Gold vs EURUSD: Same Rules, Same Code — the Spread Was 7% of a Gold Bar and 75% of a EURUSD Bar

In my gold session-by-session test four scalping rules produced 15 positive session results out of 32 after costs. Most of the same rules had already been run on EURUSD in earlier tests on this site, and they mostly lost. The obvious question is whether gold is simply a better market for indicators, or whether something more boring is going on.

The fair way to answer it is one piece of code, two data files, the same months, and nothing changed except the cost of a trade.

How were the two markets compared?

Gold is 13,736 bars of COMEX gold futures (GC=F), June 16 to August 26, 2026. EURUSD is 16,926 bars, June 3 to August 26, 2026. Both are 5-minute bars, all times in UTC.

The four scalping rules are the ones from the gold session test — EMA 9/21 cross, RSI 7 fade at 20/80, 1-hour breakout of the last 12 bars, Bollinger 20/2 fade — each entered at the close of the signal bar and exited 30 minutes later. The two trend rules are always in the market: the EMA 50/200 cross and Supertrend 10/3, each reversing on a flip.

Gold pays $0.30 per trade, an assumed retail cost. EURUSD pays 1 pip, the figure from my spread and swap cost test. Because dollars and pips cannot be compared directly, every result is also shown in ATR units: the trade’s result divided by the 14-bar Average True Range at entry. An ATR unit means “one typical bar of movement” in either market.

How big is the cost compared with the movement?

Relative to its own price, a gold 5-minute bar is about eight times bigger than a EURUSD bar. The spread does not grow with it. That single fact decides most of what follows.

It also changes by session. EURUSD bars during the London–New York overlap had a median of 2.61 pips, so the pip cost fell to 38.3% of a bar there, against 74% to 76% in the other three sessions. Gold’s cost ran from 4.7% of a bar in the overlap to 8.8% in the New York afternoon. Even gold’s most expensive session is more than four times cheaper, relative to movement, than EURUSD’s cheapest.

Same scalping rule, which market paid?

Before costs, the two fade rules were positive in both markets. After costs, gold kept one of them and EURUSD kept neither. On EURUSD every scalping rule finished negative, and the two momentum rules — the EMA cross and the breakout — lost more than a pip and a half per trade.

Is EURUSD the worse market, or just the more expensive one?

This is where ATR units help. They strip out the difference in price and show how much of a typical bar each rule captured.

Measured in volatility, the picture flips for the fades. EURUSD rewarded fading far more than gold did: RSI 7 captured +0.442 ATR per trade before costs on EURUSD against 0.000 on gold, and Bollinger captured +0.353 against +0.016. EURUSD mean-reverts on the 5-minute chart. Gold barely does.

And EURUSD still lost, because the cost of a trade there is about half an ATR, while on gold it is less than a tenth of one. The edge was real. It was simply smaller than the spread.

The momentum rules tell the opposite story. On EURUSD the EMA cross and the breakout were negative even before costs, at -0.349 and -0.305 ATR — the market reversed against them. On gold they were close to zero. Momentum does not work on EURUSD M5; on gold it merely does not lose much.

So neither market is “better for indicators”. EURUSD suits fading and cannot afford it. Gold can afford almost anything and offers little edge to any fast rule.

What happens with a slow trend rule?

The 50/200 cross is the clearest split in the whole comparison: positive on gold, negative on EURUSD, with 80 and 100 trades. In ATR units it made +1.429 per trade on gold and lost 0.853 on EURUSD.

Before calling gold a trend market, look at what gold did. It rose from $4,350.80 to $4,674.90. On gold the 50/200 cross’s 40 long trades averaged +$10.58, and its 40 short trades averaged +$1.96. Both sides were positive, which is encouraging, but the long side carried most of the total. A trend-following rule tested during a trend will look like a trend-following rule that works.

Supertrend 10/3 lost in both markets, but the size of the loss is the cost story again: -0.028 ATR per trade before costs on gold and -0.693 on EURUSD. With 738 flips in under three months, Supertrend on EURUSD M5 pays the spread far too often. What an ADX filter does to it is in my ADX vs Supertrend filter comparison and the indicator itself in my Supertrend guide for MT4.

Do gold and EURUSD move together?

Somewhat. On 13,574 matched 5-minute bars, the correlation of gold’s returns with EURUSD’s returns was +0.423. Both are priced against the US dollar, so a dollar move tends to push both in the same direction within the same bar.

That matters if you trade both. A buy on gold and a buy on EURUSD at the same time are partly the same trade — a short-dollar position — and your risk is larger than two separate positions suggest. It is also why the two markets can share good and bad days even when the rules behave so differently.

What I would take from this

• Divide your spread by your median bar before choosing a scalping rule. On EURUSD M5 it was 75%. Almost no fast rule can clear that, and none of the four here did.
• If you fade on EURUSD, fade in the overlap or on a slower chart. The edge existed at +0.44 ATR before costs. The overlap is the one session where the pip cost fell to 38% of a bar.
• Do not port a gold setting to EURUSD, or back. The EMA 50/200 cross made +$501 on gold and lost 137.5 pips on EURUSD with the same code.
• On gold, costs are not the problem; edge is. Gold’s gross results for fast rules sat between -0.045 and +0.033 ATR. Getting more out of it means better exits, which is what my gold indicator settings page and my fixed vs ATR stop test are about.
• Count gold and EURUSD longs as partly one position. A +0.423 return correlation is not a hedge.

For the EURUSD side of the same rules in more depth, the RSI period question is in my RSI 7 vs RSI 14 scalping test, the band rules are in the Bollinger squeeze test, and the moving average side is in the moving average crossover test.

Frequently asked questions

Is gold easier to scalp than EURUSD?

Cheaper, not easier. A $0.30 gold spread is 7.3% of a median 5-minute bar, while a 1-pip EURUSD spread is 75%. That let gold’s rules finish near break-even while the same rules lost on EURUSD. But the fast rules found almost no edge on gold before costs, between -0.045 and +0.033 ATR per trade.

Why do indicators that work on EURUSD fail on gold?

In this test the fade rules captured far more on EURUSD than on gold — RSI 7 fade made +0.442 ATR per trade before costs on EURUSD and 0.000 on gold. EURUSD tends to snap back on the 5-minute chart; gold tends to keep going. A mean-reversion setting tuned on EURUSD has less to work with on gold.

Does the same moving average strategy work on gold and forex?

Not the same way. The always-in EMA 50/200 cross made +$6.27 per trade on gold over 80 trades and lost 1.38 pips per trade on EURUSD over 100. Gold rose $324 during the test, which helped the long side, so treat the gold number as partly trend.

Are gold and EURUSD correlated?

Positively, at +0.423 on 13,574 matched 5-minute returns in this data. Both are quoted against the US dollar, so dollar strength or weakness tends to move them together. Trading the same direction on both adds up to a larger dollar position.

How should I compare results between gold and forex pairs?

Divide each trade’s result by the ATR at entry. Dollars on gold and pips on EURUSD are different units, but ATR is the typical bar in either market, so a result of +0.2 ATR means the same thing on both. Do the same with your spread: cost divided by ATR tells you how far behind each trade starts.