Spread, slippage and commissions: the costs a trading robot pays without you noticing

Spread, slippage, commission and swap are the four costs that any trade a robot opens can run up, and none of them shows up in the profit figure: they’re taken out before or after, in columns almost nobody looks at. On gold (XAU/USD) they weigh more than you’d think, especially if the system enters with stop orders and holds positions overnight. Here we define each one with examples, add them up on a 0.05-lot trade and show you how to see them in MetaTrader 5.

What is spread in trading? The toll you pay the moment you enter

The spread is the difference between the price at which the broker sells to you (the ask) and the price at which it buys from you (the bid). You buy at the ask and sell at the bid, so the moment you open a trade you’re already down by that difference. On a standard account it’s the main cost: there’s no separate commission because it’s built into the spread.

Most brokers quote XAU/USD with two decimals ($3,300.25, for example), so 1 point = $0.01 per ounce. One lot is 100 ounces: a point is worth $1 per lot and $0.05 on 0.05 lots. A 25-point spread is $0.25 per ounce, within the $0.10–0.35 range we give as a reference in our guide to choosing a broker. On a raw or ECN account it drops to 5–15 points, in exchange for a fixed commission we’ll cover below.

When the spread on gold widens

The advertised spread is the best one of the day; the one that matters is the spread during the hours your robot trades. Four moments when 25 points turn into 80 or 120:

• The Asian open, shortly after midnight Spanish time (Central European Time): liquidity is at its lowest.
• Top-tier news (US nonfarm payrolls, inflation, the Federal Reserve), before and after the release.
• Friday afternoon, in the last hours before the weekly close.
• The server’s daily rollover, around 11 p.m. to midnight Madrid time, when swaps are applied.

A robot doesn’t know it’s “a bad time”: if its rule is met, it enters and pays the wide spread.

What is slippage? When the price you ask for isn’t the price you get

Slippage is the difference between the price at which an order is sent and the price at which it’s filled. It’s negative when you get a worse price and positive when you get a better one. In normal conditions it’s small and goes both ways; if your history only ever shows it against you, the problem is the broker.

It comes from the latency to the server and from the liquidity available when the order arrives. On gold, which moves much faster than a currency pair, it can run to tens of points on an inflation release or at the Monday open.

Why stop orders suffer more than limit orders

A stop order (a buy stop above the price, a sell stop below it) is a pending order that, when the market touches its level, turns into a market order: it’s filled at the first available price, whatever that is. In a gap, that first price can be 50 or 100 points further away. A limit order does the opposite: it only fills at its price or better, with no negative slippage, but it may never fill at all.

Many trend-following systems, AUREUM among them, enter with buy stop and sell stop orders to ride the move once it starts, not to anticipate it. That fits this kind of strategy, but it comes at a price: in gaps and on news, stop orders are the ones that suffer the most slippage. Check your robot’s history to see which order type it enters with; a fast-execution broker with no requotes reduces slippage, but doesn’t eliminate it.

Broker commissions and swap: what you pay on top of the spread

The commission is a fixed, volume-based fee on raw or ECN accounts, in exchange for a spread with almost no markup. On gold it’s usually $6–7 per lot “round turn,” meaning for opening and closing one full lot. It’s charged when you open (sometimes half on opening and half on closing) and has its own column in your history; with 0.01 lots it’s a matter of cents, but those cents are multiplied by hundreds of trades.

Swap is the financing cost of holding a position open overnight. It’s applied every night at the server’s rollover and, one day a week (usually Wednesday), it’s charged triple to cover the weekend. On gold, long positions almost always pay negative swap, because gold earns no interest and financing the position costs money; shorts sometimes earn a little and often pay too. As a rough order of magnitude, and only as an example, it’s not unusual to see $30 to $60 per night on a one-lot long; your broker’s exact figure is in the symbol’s Specification in MT5 (right-click XAUUSD in Market Watch). Remember the rule: spread and commission are paid per trade; swap, per night. A system that holds positions for several days can pay more in swap than in spread.

How trading costs add up: a 0.05-lot gold trade, by the numbers

The robot opens a 0.05-lot buy on XAU/USD on a raw account, with illustrative figures: a 25-point spread on entry, a commission of $7 per lot round turn and a swap of −$40 per lot per night on longs. The trade is held for one night, with no triple charge.

At 0.05 lots each point is worth $0.05, so $3.60 is 72 points ($0.72 per ounce) that gold has to move in your favor just to bring the trade back to zero. If it stays open through the triple-swap night, the total rises to $7.60: 152 points.

Two account types, same trade, no swap:

• Standard account, 35-point spread and no commission: 35 × 0.05 = $1.75.
• Raw account, 10-point spread plus $7 per lot: 0.50 + 0.35 = $0.85.

Ninety cents looks like nothing until you multiply it: AUREUM’s published backtest contains 439 simulated trades between 2020 and 2026; at 0.05 lots (the size AUREUM calculates by default with a 5,000 balance), that difference adds up to almost $400 over the period, not counting swap. The account label doesn’t matter: what decides it is the total cost per trade at the hours your robot trades.

Why a fixed-spread backtest can lie to you

The MetaTrader 5 Strategy Tester lets you simulate with the symbol’s current spread or with a fixed value. Setting a fixed 10 points for the whole history assumes the market never widened: not in Asia, not on payrolls, not on Fridays. A system that enters right at those moments can look excellent on paper and only break even live, not because the logic fails, but because the cost the backtest assumed never existed. A simple test: run the same backtest with a low fixed spread and with the real one; if the curve changes shape dramatically, the system depends on conditions you don’t control. More in our guide to backtesting and overfitting.

That’s why the AUREUM backtest we publish on the homepage was run on XAU/USD history from 2020 to 2026 with the real historical spreads, not a fixed value. The published figures from that simulation are +410%, a maximum drawdown of 21.84%, 439 trades and a profit factor of 1.49. Read them for what they are: a simulation on past data, not a live account, under the conditions of the broker used for the test; your broker will have its own commission and swap, and comparing them is your job. If you don’t know how to read profit factor, start with our guide to the Strategy Tester report.

Choose a broker by total trading cost, not the advertised spread

The temptation is to hunt for the lowest spread on the internet. The problem is that the most aggressive “from 0.0” offers usually come from brokers without European regulation: what you save on the toll, you put at risk by not knowing where your money is. We cover regulation, execution and platform in depth in which broker to choose for an Expert Advisor; here we only add what changes when you look at costs up close:

• Compare the total cost per trade (spread + commission + expected swap, based on your system’s average time in the market), not the “standard” or “raw” label.
• Measure the spread yourself on a demo account, during the hours your robot trades and for at least a week: the advertised spread is the best one of the day.
• Ask for the swap in writing for longs and shorts on XAU/USD, including the triple-charge day: even between reputable brokers it can vary by a factor of more than two.
• Check the slippage on your first live trades at the minimum lot size: it should be small and show up in both directions.

And if you’re not sure why gold behaves differently from a currency pair, read how to trade gold (XAU/USD) first.

How to see what you really paid in MT5

MetaTrader 5 records everything, but in columns that are hidden by default. Here’s how to turn them on:

1. Open the Toolbox (Ctrl+T) and go to the History tab.
2. Right-click to choose the period and the Positions view, which groups each complete trade.
3. Right-click the table header and turn on the Commission and Swap columns.
4. The summary row at the bottom adds up commission, swap and profit for the period: your real toll, not counting the spread, which is built into the entry price.
5. Right-click and choose Report, and MT5 exports the history with totals as HTML or as a spreadsheet.

The spread has no column: for a few days, note down its value when the robot opens each trade and convert it to dollars using the contract size. With the estimated spread, the commission and the swap, you have the real cost of your system with your broker. Repeat it every month: that way you’ll notice in time if a broker has changed its conditions.