Stop Loss: Wide or Tight? Real Backtest Results on Gold

A tight Stop loss looks safe, but in practice it is the one that gets hit most often. Price swings back and forth all the time. Add a little spread, and that is enough to touch an SL that is too close, even when your direction is right. In our backtests on gold, an SL placed beyond a structure point (wider) beat a tight SL in almost every set. There is one important condition: a wide SL needs a smaller lot, so the money lost per trade stays the same.

This article follows on from Position size from your Stop loss. If you have not read it yet, start there.

Why a tight SL gets hit so often

There are two reasons.

1. Noise (normal swings). Price does not move in a straight line. Even in a clear trend, some candles flick the other way from time to time. An SL inside this swing range gets touched more or less at random, whether your idea was right or wrong.
2. Spread (the gap between the buy and sell price). A Buy closes at the Bid price, which is lower than the price you see on the chart by the spread. The tighter the SL, the bigger the share of the SL distance the spread takes.

Here is the effect of spread in numbers. Assume a round-trip spread on gold of about 0.27 dollars (close to the value we use in our tests).

• SL 2 dollars away: 0.27 ÷ 2 = 13.5% of the SL distance is lost to spread
• SL 10 dollars away: 0.27 ÷ 10 = 2.7%

So the tight SL gives five times as large a share to spread, before price has even moved. More detail in Small timeframes and the cost of spread.

Test 1: a 61.8 pullback with three SL distances

This system measures Fibonacci on a large swing. Level 0 is the end of the swing and 100 is the start. It enters when price pulls back to 61.8 and takes profit at 38.2. We changed only the SL position: 65 (just 3.2 units deeper than the entry), 78.6 or 88.6.

Before testing, we checked how much deeper price goes after touching 61.8. It turned out that 96% of the time, price went on down to 65, and 89% of the time to 70. So an SL at 65 is almost certain to be hit.

In the same 5-year dataset (2022–2026), looking at how far price moved in our favour after entry before hitting the SL, every table ranked the same way: SL 88.6 beat 78.6, and 78.6 beat 70.

Then we ran it as a real system. Conditions: gold (XAUUSD) M1, only swings at least 1% of price long, 0.10 lot per trade, starting balance 10,000 dollars, 2024–2026, with 2026 on real tick data. All results are backtests. Past results do not predict future results, and the numbers below are not returns you should expect.

SL	2024	2025	2026	3-year total	Max drawdown
65	−381	−815	−1,078	−2,274	21.8%
78.6	−177	−435	+3,827	+3,215	14.7%
88.6	+274	−864	+5,579	+4,989	17.3%

Drawdown is how far the account falls from its peak along the way.

SL 65 lost money every year, matching the statistic above. In 2026 it took 756 trades, compared with 197 for SL 88.6, because it kept getting stopped out and re-entering over and over.

To be honest, SL 88.6 did not win every year. In 2025 every version lost money, and most of the profit came from 2026. Also, this test used the same 0.10 lot for every version, so the wide SL risked more money per trade, and it still did better.

Test 2: a breakout system wins more often as the SL widens

This system places a Buy stop above the recent price range and a Sell stop below it, and enters whichever side breaks first. A breakout means entering in the direction price breaks out of a range. We measure the SL as a share of the range width, for example 0.2 is 20% of the range and 2.0 is twice the range. Gold M1, 0.01 lot per trade, 3,000 dollar account. The win rates below are for 2026.

The win rate rises at every step, no exceptions. But read this chart with two things in mind.

• Part of the higher win rate is just geometry. With the same target and a further SL, even random entries would win more often. What matters is the net money, not the win rate. More in Win rate and breakeven.
• The result is fragile. Run continuously from 2022 to August 2026, SL 2.0 times made 4,120 dollars, but SL 1.75 times lost 3,005 dollars and wiped out the account. A tiny change in the setting flipped the result, so it cannot be trusted yet (see Plateaus vs peaks when tuning settings).

The practical lesson from this set: a small range has a lot of noise, and the meaningful SL is at the far side of the range, not right next to the entry price.

Test 3: forcing the SL to a fixed dollar amount breaks the system

One of our breakout systems on an 8-bar box (Donchian channel) uses an SL 0.6% of price away. We tried switching it to cut the loss at a fixed dollar amount per trade. Conditions: gold M1, 2026, real tick data, 0.05 lot per trade, 3,000 dollar account. Backtest results do not predict future results and are not returns you should expect.

SL	Net profit	Drawdown	Profit ÷ Drawdown	Win rate
0.6% of price (original)	4,575	392	11.68	95.8%
100 dollars	3,501	322	10.87	93.5%
60 dollars	1,925	590	3.26	87.6%
40 dollars	1,475	893	1.65	81.9%
20 dollars	289	1,408	0.21	66.6%

Profit ÷ Drawdown shows how many times the worst drop in the account you made back in profit. Higher is better. A 20 dollar SL at 0.05 lot (5 ounces) is a price distance of only 20 ÷ 5 = 4 dollars.

Check: 4,575 ÷ 392 = 11.67 (with the full figures, 4,575.25 ÷ 391.80 = 11.68), and 289 ÷ 1,408 = 0.21.

Profit fell at every step as the SL was squeezed. The number of trades rose from 960 to 1,118, because trades that used to dip and then recover were stopped out and had to re-enter. This system moves its SL along with price (trailing stop) once it is in a small profit. A tight SL cuts the trade before that mechanism gets a chance to work.

In 2026, an SL of 0.6% of price was roughly 130–147 dollars per trade, close to a fixed 140 dollar SL. Yet the fixed version made 3,875 dollars, about 15% less. A distance that adjusts with price and volatility has value of its own.

To risk less, cut the lot, not the SL. A smaller lot shrinks both profit and Drawdown in proportion, while a tighter SL usually cuts profit and raises Drawdown at the same time.

A wide SL must come with a small lot

This example uses the same setup as Test 1. A 10,000 dollar account risking 1% = 100 dollars. A gold swing 45 dollars long, entry at 61.8.

SL at 65

• SL distance = (65 − 61.8)% × 45 = 3.2% × 45 = 1.44 dollars = 14.4 pips (1 pip on gold = 0.10 dollars)
• lot = 100 ÷ (14.4 × 10) = 100 ÷ 144 = 0.694, rounded down to 0.69 lot
• spread 0.27 ÷ 1.44 = 18.8% of the SL distance

SL at 88.6

• SL distance = (88.6 − 61.8)% × 45 = 26.8% × 45 = 12.06 dollars = 120.6 pips
• lot = 100 ÷ (120.6 × 10) = 100 ÷ 1,206 = 0.083, rounded down to 0.08 lot
• spread 0.27 ÷ 12.06 = 2.2% of the SL distance

Check (1 lot of gold = 100 ounces)

• 0.69 lot = 69 ounces × 1.44 = 99.36 dollars
• 0.08 lot = 8 ounces × 12.06 = 96.48 dollars

Both risk about the same and no more than 100 dollars. But the SL at 88.6 does not get eaten by normal swings, and spread takes about eight times less of it.

Always fit the lot to the SL. Never fit the SL to the lot you want to open.

How wide is wide enough?

Wide does not mean the further the better. A good SL sits at the point where, if price gets there, the idea behind the trade is wrong, such as beyond the prior low or at the far side of a range. That distance comes from the structure of the chart, not from how much money you would like to lose.

Be careful with an SL placed right at the turning point (pivot). If you wait for ZigZag to confirm the pivot before entering, then in our gold M1 data, price had already moved back about 46–49% of the swing, in every year and with every ZigZag setting, before we even knew there was a pivot. This number comes from how ZigZag is calculated, not from market behaviour. The result: if you enter on confirmation and put the SL at the pivot, you are risking about half a swing to chase the remaining half.

Common mistakes

• Using a tight SL because it feels safe. A tight SL with a big lot risks the same money as a wide SL with a small lot, but gets hit far more often.
• Widening the SL without cutting the lot. You end up risking more per trade without meaning to. Recalculate the lot every time.
• Setting the SL as a fixed dollar amount, for example 20 dollars on every trade, without looking at where that distance falls on the chart.
• Only looking at the win rate. A wider SL naturally wins more often. Look at net profit and Drawdown too.
• Moving the SL away as price gets close. That is not a wide SL, it is no SL. Set the distance before you enter, and do not move it further away.