Gold 1-Minute Scalping Strategy: An EMA Pullback Scalp Tested, and What Costs Do to It

The short answer

We tested a popular 1-minute gold scalping strategy, a pullback to the 21 EMA inside a stacked 9/21/50 EMA trend, on 27,510 one-minute bars of COMEX gold (2 to 30 October 2026, 21 trading days). It made 343 trades, won 41.4% of them, and lost $143.80 per ounce after a $0.30 per ounce cost on every trade.

The more useful finding is about costs. Before costs the same trades lost only $0.12 an ounce each. The $0.30 cost added another $103 of losses, and at $0.60 the loss reached $247 an ounce. On the 1-minute chart, the spread is not a detail; it is most of the result.

The strategy: a 9/21/50 EMA pullback scalp

This is one of the most common M1 setups taught for XAUUSD: trade only in the direction of a stacked EMA trend, wait for price to dip into the 21 EMA, and enter when it snaps back above the fast 9 EMA. We wrote it as fixed rules so that a computer could trade it without judgement.

Exact rules

The full lesson on building an M1/M5 framework, including round-number sweeps and session rules, is in our gold scalping strategy lesson. This page is the stress test of one specific setup.

How we tested it

The results

Split by week (main rules)

Every week lost. The worst run was 15 losing trades in a row.

Costs: the same trades at $0, $0.30 and $0.60

Because every trade is recorded before costs, we can re-price the identical trade list at different costs. This is the table that matters for any 1-minute trader.

Two things stand out:

• Costs, not the signal, did most of the damage. For the main rules, costs were $103 of the $144 loss. The 1R version was almost flat before costs (−$0.04 a trade) and lost $130 after them.
• The wider stop was the only version positive before costs, by a tiny $0.05 a trade. A $0.30 cost is six times that edge. Doubling the cost to $0.60 doubled the loss.

Why M1 is so sensitive: cost as a share of the bar

We measured the average ATR(14) during London and New York hours on each timeframe, then expressed a round-trip cost as a share of it.

With a 1.5 ATR stop of about $3, a $0.30 cost means every trade starts 0.1R behind. At $0.60 it starts 0.2R behind. A strategy that wins 50% of the time at 1:1 would need to win about 55% just to break even at $0.30, and about 60% at $0.60. Very few simple signals have that kind of edge on a 1-minute chart.

What the results say, in plain English

• The signal itself had no edge in this sample. Before costs, the main rules lost a few cents per trade. A pullback-to-the-21-EMA entry on M1 is close to random over four weeks of data.
• Higher win rate did not save it. The 1R version won 51% of its trades and still lost, because the average loss plus cost was larger than the average win.
• Wider stops reduce the cost burden. Doubling the stop cut the number of trades from 343 to 212 and made the cost a smaller share of each trade. It still lost after costs.
• The trend of the month mattered. Gold fell about $199 an ounce over the window. The main rules lost $115 on longs and $29 on shorts. In a rising month the split might reverse.

When this kind of scalp fails

• Chop around the EMAs. When the 9, 21 and 50 EMAs bunch together, the “trend” flips every few minutes and pullback entries are coin flips.
• News minutes. Around 08:30 New York data releases, spreads widen and a $3 stop can slip to $6 or more. Our test used a flat cost and does not capture that.
• The Asian session. We excluded it. Quiet M1 bars make the cost share even larger.
• Widening spreads at session turns. The rollover hour and the Sunday open have spreads many times larger than normal.

Risk sizing for an M1 scalp

Size from the stop, never from the account. In this test the average stop was about $2.98 per ounce. If you risk $50 per trade:

• $50 / ($2.98 + $0.30 cost) is about 15 ounces, or 0.15 lots.
• At 0.15 lots, each $1 move is $15, and the 15-trade losing streak in our test would have cost about $740 before any recovery.

Run your own numbers with the position size calculator and check what each dollar move is worth with the gold pip calculator. Our guide to XAUUSD spreads, commissions and swaps explains how to find your true round-trip cost, which is the single most important input for an M1 test.

If you still want to scalp the 1-minute chart

1. Measure your real cost first. Spread plus commission plus typical slippage, per ounce, at the hours you trade.
2. Make the target large relative to the cost. A target under ten times your round-trip cost is very hard to beat.
3. Trade fewer, better minutes. The London open and the New York morning have the largest M1 bars; the cost share is lowest there.
4. Consider the 5-minute chart. Our 5-minute scalping test lost too, but each bar is about 2.7 times larger, so costs bite less.

Test it yourself

Write the rules above into your platform’s strategy tester, use your broker’s XAUUSD 1-minute history, and run it three times: at zero cost, at your real cost, and at double your real cost. If the result only works at zero cost, you have learned something valuable for free. Our lessons on backtesting a gold strategy and backtesting with free data show how to set this up, and the gold strategies tested hub compares this result with 13 other named strategies tested the same way.

Practise it with other traders

Scalping is easier to judge honestly when other people see your numbers. In the XAUUSD Global community, traders post their M1 and M5 journals, real spreads and cost-adjusted results, and pick each other’s setups apart before money is at stake. The first month is free, then $19.99 a month.

Educational content only, not financial advice. Trading gold with leverage carries a high risk of loss.

Frequently asked questions