Two accounts quote EURUSD at the same average spread on their marketing pages. On one of them, a trader taking four positions a day around the London open pays roughly what the page says. On the other, the average is dragged down by the quiet Asian hours and the mornings when nothing happens, and the trader who works the open pays visibly more. The advertised number is not false. It is answering a question nobody asked.
An average is a summary of something you did not choose
A spread is the distance between the price you can sell at and the price you can buy at. It moves constantly, because it is a product of how many liquidity providers are quoting and how much size they are showing, and both of those change through the day. The mechanics are covered in the guide to the bid ask spread.
What an average hides is the shape of the distribution. Two brokers can show the same mean while one holds a tight quote for twenty two hours and blows out for two, and the other sits slightly wider all day and barely reacts to news. If your entries cluster in the calm hours, the first is cheaper. If you trade releases, the first is much more expensive, and no monthly average will ever tell you that.
The fix is not a better source of averages. It is a log with a timestamp on every reading, taken on your account, on the symbols you actually trade.
What a monitor records
The simplest version is a script on the trading platform that reads the current bid and ask on every tick and writes symbol, time and spread to a file. On MetaTrader that is a short indicator or expert advisor pulling the platform's spread value in points; on other platforms it is usually a few lines against the quote feed. Left running on a small virtual server for a fortnight it produces something no marketing page can: your broker's quote history at minute resolution.
From the raw log the useful views come out quickly. Average spread per symbol per hour of the day. Maximum spread per hour, which is where the damage lives. Count of readings above some threshold, say three times the normal value, which tells you how often the quote is unusable. A separate line for the daily rollover window and for the minutes around scheduled releases.
There are ready made options too. Public comparison pages, including the live spread listings published by Myfxbook, sample multiple brokers at once and are useful as a sanity check on whether your quotes look normal for the market. They are not a substitute for your own log, because they measure their own connections and their own account types, and neither may match yours. Treat them as a second opinion on the market, not as a reading of your account.
| Cost component | Where it hides | How to measure it |
|---|---|---|
| Spread | Averaged across hours you do not trade | Tick log, grouped by hour of day and by symbol |
| Commission | Quoted per side or per round turn, per lot | Read it off the contract specification, convert to pips for your lot size |
| Slippage | Difference between requested and filled price | Compare order price with fill price in the trade history |
| Rollover widening | A few minutes each day, often outside chart attention | Filter the log to the rollover window and read the maximum |
| Financing | Charged nightly, invisible on the ticket | Position size times the published swap rate times nights held |
The comparison that decides anything
Once the log exists, convert everything to one currency amount per round turn on the position size you normally trade. A raw spread account with a commission and a standard account with the cost inside the quote are directly comparable only in that form, and the answer flips depending on whether you trade one lot or a tenth of one. The way brokers assemble these charges is set out in the piece on spreads, markups and commissions.
Then add the holding cost. A position kept for four nights pays financing four times, and for some instruments that dwarfs the entry cost entirely, which the guide to swap rates works through. Traders comparing brokers on spread alone while holding swing positions are optimising the smaller number.
Measure on the live account, not the demo. Demo servers frequently quote a synthetic feed with a fixed or flattering spread, and comparing a demo log against a live one tells you about the two servers rather than about the market. If you cannot log live, log a small live account.
Reading the results without fooling yourself
A fortnight of data covers a fortnight of conditions. If those two weeks contained no central bank decision and no holiday session, the log will understate what a quiet month looks like when one arrives. Keep the collector running permanently and rebuild the summary monthly, and mark the days when something unusual happened so they can be included or excluded deliberately.
Be careful about the conclusion you draw as well. A wider spread during a data release is not misconduct, it is what the underlying market does when liquidity providers pull back. What is worth noticing is the comparison against other brokers at the same minute, and the pattern of when your fills arrive worse than the quote you saw. That belongs in the trade record rather than in the spread log, which is one reason a trading journal that captures requested and filled price earns its keep.
For anyone taking many small trades a day, this measurement is the difference between a method that pays and one that does not, since cost per trade is the largest fixed drag on a short holding period. The arithmetic behind that is spelled out in the scalping guide. Leveraged trading carries a high risk of loss, and a cost you have never measured is a cost you cannot manage.
"A pricing page tells you what the market looked like when somebody wrote it. Your own log tells you what you paid last Tuesday at 16:31. Only one of those shows up on the statement."
— Alex Onta, Executive Director, SINGUARD
Key Takeaways
- An advertised average spread is computed over the whole session, so it rarely describes the hours a given trader actually works.
- A script logging bid and ask on every tick for two weeks produces average and maximum spread per hour, which is the view that matters.
- Compare account types on all-in cost per round turn at your usual position size, then add financing for anything held overnight.
- Collect the data on a live account, keep collecting after the decision is made, and mark unusual sessions rather than deleting them.
Frequently Asked Questions
What is a spread monitoring tool?
It is any tool that records the gap between bid and ask over time instead of showing only the current value. That can be an indicator or script on a trading platform writing each reading to a file, a panel that displays the minimum, average and maximum for the session, or a public comparison page that samples several brokers at once. The purpose is the same: replace a single advertised number with a distribution measured over your own trading hours.
Why does the spread widen at rollover?
Around the daily rollover, liquidity providers step back while positions are rolled and books are squared, so fewer quotes are available and the gap between the best bid and best ask grows. Thin holiday sessions and the minutes around a scheduled data release produce the same effect for the same reason. Orders that reach the market during those windows are filled at whatever price exists, which is why stop orders placed near the current price can be triggered by a widening that never reflects a real move.
Is a zero spread account cheaper?
Not automatically. A raw or zero spread account moves the cost into a commission per lot, so the fair comparison is the all-in figure: measured spread plus commission plus any typical slippage on entry. For a trader holding positions for days, financing charges usually outweigh both. Comparing account types on one line of a marketing page rather than on total cost per round turn is how most traders end up on the wrong one.