What is spread? The small cost that eats more profit per trade than you think

Spread is the gap between the price you can buy at and the price you can sell at. You pay it every time you open an order, whether that trade wins or loses. Per trade it looks small, but if you trade often or use a tight Stop loss, it can eat almost all of your profit.

Where spread comes from

At any moment, your platform shows two prices.

• Bid is the price you can sell at
• Ask is the price you can buy at. It is always higher than Bid

The difference between the two is the spread. When you open a Buy, you buy at the Ask, but when you close you sell at the Bid. Open and close straight away and you lose exactly the spread. So every order starts slightly in the red. Price has to move your way by more than the spread before you start to make a profit.

How to turn it into money

The formula: cost = spread (in price) × contract size per lot × number of lots

For gold (XAUUSD), most brokers set 1 lot = 100 ounces (check your own broker’s contract specs). So 0.01 lot equals 1 ounce.

The average gold spread we measured from real ticks, January to August 2026, on the account we backtest with, was about 0.272 dollars per ounce. So 0.01 lot pays about 0.272 dollars per trade (once per open-and-close round trip). We’ll use that number as the example.

Lot size	Size in ounces	Spread cost per trade
0.01	1	0.272 × 1 = 0.272 USD
0.05	5	0.272 × 5 = 1.36 USD
0.10	10	0.272 × 10 = 2.72 USD
1.00	100	0.272 × 100 = 27.20 USD

If you count 1 pip of gold as 0.10 dollars, a 0.272-dollar spread is about 2.7 pips.

For currency pairs, say EURUSD has a 1 pip spread and you open 0.10 lot, which is worth about 1 USD per pip. The cost per trade is then about 1 USD.

Compare it with profit per trade, not total profit

A more useful question than “is the spread expensive?” is “how many times bigger is my system’s average profit per trade, before costs, than the cost per trade?”

Here is an example from our backtests: a dip-buying rule on gold that trades very often.

1. Over 5 years it opened 17,772 trades, with a net result of −2,626 USD
2. Average per trade = −2,626 ÷ 17,772 = −0.148 USD
3. Add the spread back = −0.148 + 0.272 = +0.124 USD per trade, before spread
4. Total spread paid ≈ 17,772 × 0.272 ≈ 4,834 USD

So this rule had a small edge before costs, but the spread per trade was about 2.2 times bigger than that edge (0.272 ÷ 0.124). That is why the net result was negative.

Another system we tested on real 2026 ticks (198 trades, 0.01 lot) averaged +1.66 USD per trade after costs. That is 1.66 ÷ 0.272 ≈ 6.1 times the spread. For a system like that, the spread barely changes the overall picture.

Both examples are backtest results. They do not predict future results. They are here only to show the way of thinking.

Systems that trade often but make little per trade are the most sensitive to costs. A small change in spread can flip the total from positive to negative.

The tighter the Stop loss, the heavier the spread

The spread is a fixed distance, but everyone’s Stop loss (the point where you cut a loss) is different. Look at the spread as a share of your Stop loss distance and you’ll see: the tighter the SL, the more the spread takes.

Stop loss distance (gold)	Spread of 0.272 equals
1 dollar	0.272 ÷ 1 = 27.2%
2 dollars	0.272 ÷ 2 = 13.6%
5 dollars	0.272 ÷ 5 ≈ 5.4%
10 dollars	0.272 ÷ 10 ≈ 2.7%

This can happen within one system too, if it sets the SL from the size of the price range. In our 2026 gold test, a medium range gave an SL of about 3.68 dollars. The spread was 0.272 ÷ 3.68 ≈ 7.4% of the SL.

Now imagine a year where price moves in a narrower range, so the same rule gives an SL of about 1.44 dollars (a rough figure, not measured). The spread becomes 0.272 ÷ 1.44 ≈ 18.9% of the SL. Same rule, but the cost is about 2.5 times heavier. The system might look broken in a year like that, when what’s really eating the profit is the cost.

Spread is not fixed

The spread you see when the market is quiet is not the only spread you pay. It often widens:

• When the market opens after a weekend or holiday
• At the broker server’s daily rollover, when liquidity is thin
• When major news comes out

In our backtest on real 2026 tick data, we entered one gold trade a day with a 5-dollar SL. Entering at 07:00 Thai time (UTC+7), which matched the daily rollover of the test server, gave worse results than 06:30, both for buying every day and for selling every day. Our guess is that the spread widened right after the rollover and hit the tight Stop loss. Rollover time differs by broker, so check your own server time.

Common mistakes

• Looking only at spread and forgetting other costs. Real costs also include commission and swap (overnight interest when you hold past the daily rollover).
• Checking the spread when the market is quiet and assuming you pay that on every trade.
• Setting a very tight Stop loss without allowing for spread. Trades where price barely moved against you can still get stopped out.
• Looking at total profit, not profit per trade. A large number of trades can make the total look fine and hide how much the costs are taking.

Before using any system, divide the net profit by the number of trades. Then ask: if costs rise a little, is this number still positive?