Min, Median, p90, At Capture: One Spread, Four Statistics · Tanzania
Real spreads we recorded on FxPro’s own MetaTrader 5 Raw+ feed — 6 instruments, 3,135,851 ticks sampled, last captured 2026-08-19. The spread you actually trade on, not a marketing ‘from 0.0’.
Open FxPro Account →This page publishes four different statistics about the same quantity, and its first job is to say which is which. The spread on each instrument is recorded on a live FxPro MetaTrader 5 Raw+ account by an in-terminal logger that writes down the bid and ask of every tick; from that stream the table reports the tightest quote observed, the median, the level exceeded about a tenth of the time, and the single live quote at the moment of the last capture. Four questions, four answers: what is possible, what is normal, what to budget for, and what the market happened to be doing when the reading ended. The hourly view underneath is the same data cut by hour across the last twenty-four hours, and the cost table converts the typical spread into money for one standard lot and adds the $7.00 Raw+ round turn. Every block carries the moment it was captured, because a spread quoted without one cannot be checked by anybody.
This is the live, hour-by-hour measured spread feed (refreshed daily). For the Standard vs Raw+ cost comparison and fees, see our spreads & costs page.
Measured Raw+ spreads (pips)
| Instrument | Best (min) | Typical (median) | Busy market (p90) | At capture | Ticks sampled |
|---|---|---|---|---|---|
| EUR/USD | 0.2 | 0.2 | 0.2 | 0.2 | 363,112 |
| GBP/USD | 0.6 | 0.6 | 0.6 | 0.6 | 488,161 |
| AUD/USD | 0.2 | 0.4 | 0.8 | 0.2 | 424,664 |
| USD/CAD | 0.1 | 0.4 | 0.5 | 0.4 | 423,444 |
| USD/JPY | 0.3 | 0.3 | 0.5 | 0.3 | 486,619 |
| XAU/USD (Gold) | 15 | 15 | 19 | 15 | 949,851 |
Best = the tightest quiet-market quote we saw; Typical = the median you usually trade; Busy market = the wider spread to expect about 10% of the time (news, rollover, thin liquidity). ‘At capture’ is the live spread at the last reading. Metals such as XAU/USD use a different contract size, so their cash cost is on our gold page. Server FxPro-MT5 Demo, feed 2026.08.19 08:54:02.
Spread through the trading day (measured, last 24h)
| Instrument | Tightest (avg) | Widest (avg) | Worst spike | Through the day |
|---|---|---|---|---|
| EUR/USD | 0.2 (03:00) | 1.541 (00:00) | 7.8 (23:00) | |
| GBP/USD | 0.6 (03:00) | 4.909 (00:00) | 15 (00:00) | |
| AUD/USD | 0.293 (12:00) | 4.474 (00:00) | 15.2 (23:00) | |
| USD/CAD | 0.294 (18:00) | 5.465 (00:00) | 20.2 (23:00) | |
| USD/JPY | 0.3 (08:00) | 10.063 (00:00) | 16 (23:00) | |
| XAU/USD (Gold) | 15 (08:00) | 175 (00:00) | 175 (00:00) |
Table hours are FxPro server time (about UTC+3 / EET); the highlighted guidance above is shown in EAT. Average pip spread by hour over the last 24 hours, with the worst single-tick spike. Spreads run tightest in the peak London–New York overlap and widen around the 00:00 server rollover and the thinner Asian hours — the sparkline is each instrument’s daily shape.
What it costs you per lot (Raw+)
| Instrument | Typical spread | Spread cost / lot | Commission (round turn) | All-in / lot | All-in (pips) |
|---|---|---|---|---|---|
| EUR/USD | 0.2 pips | $2.00 | $7.00 | $9.00 | 0.9 pips |
| GBP/USD | 0.6 pips | $6.00 | $7.00 | $13.00 | 1.3 pips |
| AUD/USD | 0.4 pips | $4.00 | $7.00 | $11.00 | 1.1 pips |
| USD/CAD | 0.4 pips | $2.88 | $7.00 | $9.88 | 1.37 pips |
| USD/JPY | 0.3 pips | $1.88 | $7.00 | $8.88 | 1.42 pips |
| XAU/USD (Gold) | 15 pips | $15.00 | $7.00 | $22.00 | 22 pips |
All-in round-turn cost for one standard lot (100,000 units): typical spread × pip value, plus the $7 Raw+ commission ($3.50 per lot per side ($7.00 round turn) on Raw+ and cTrader accounts). On a Standard account you pay a wider spread instead of that commission — see the full spreads and costs page.
Open FxPro Account →Advertised ‘from 0.0’ vs what we measured
FxPro markets Raw+ as spreads ‘from 0.0 pips’ — a best-case floor. Across our sample the tightest EUR/USD quote we recorded was 0.2 pips and the typical was 0.2 pips. That is normal: the ‘from’ figure is a floor you rarely trade on, so judge a Raw+ account by its typical spread and how far it widens under load (the p90 column), not the headline number.
How we measured this
- Recorded on FxPro's own MetaTrader 5 Raw+ account — the broker's real pricing feed, not a third-party estimate.
- Captured in-terminal by an MQL5 expert advisor that logs every tick's bid and ask, so the spread is exactly what the platform shows.
- 3,135,851 ticks across 6 instruments; the figures refresh on a schedule.
- Demo and live Raw+ share the same pricing feed, so these spreads are representative of a funded account.
Spreads are variable and widen around high-impact news and the daily rollover. Past readings do not guarantee future spreads. Last updated 2026-08-19.
Four columns, four different questions
The temptation with a spread is to reduce it to one number, and the reduction is where most cost arguments go wrong. A spread changes with every tick, so any single figure is a choice about which part of the distribution to report, and that choice ought to be visible rather than buried.
So the table reports four. The tightest quote observed answers what is possible. The median answers what is normal. The level exceeded about a tenth of the time answers what to budget for when a release lands or liquidity thins. The quote at the moment of capture answers what the market happened to be doing when the reading ended, and it is the only one of the four that is a quote rather than a summary.
Read across a row rather than down one column and the instrument describes itself. A row where all four figures sit close together is a steady instrument. A row where the busy-market figure is several times the median behaves very differently under load, and what it costs to trade depends on when you trade it.
The sample, the window and the stamp
Behind each row is a tick log: an in-terminal recorder that writes the bid and ask every time the platform receives a new quote, so the sample is the quote stream itself rather than a periodic snapshot of it. The number of ticks that went into a row is printed with the row, because the size of a sample is part of the result.
The window is the last twenty-four hours, refreshed daily. That is short by the standards of published statistics, and deliberately so: the figure is meant to describe the market of the day you are reading it, and a longer average would trade that away for smoothness. The cost of the choice is that one turbulent session stays in the numbers until it rolls out of the window, which is why the capture moment is printed rather than implied.
The stamp is the part that makes everything else checkable. A number with a moment attached can be disagreed with; a number without one can only be believed or not. Everything measured on this site carries one, including the contract fields on our trading conditions page, where the read date sits under each block.
What this page does not claim
A measured floor is not a guarantee. The tightest quote in the sample was available for as long as it lasted, and nothing about it promises the same quote will be there when an order arrives.
The busy-market column is not a worst case either. It is the level exceeded about a tenth of the time, which means it is exceeded, and the spike column beside the hourly rows exists precisely to show how far that can go.
The cost table, finally, prices one round turn at the typical spread. It says nothing about what accrues while a position stays open, which is measured separately on our swap rates page, and nothing about the fill itself. Stating those edges is not a disclaimer, it is what stops a measurement from being read as a promise.
What each column answers
| Column | What it is | Question it answers | What it does not say |
|---|---|---|---|
| Best (min) | The lowest quote in the sample | What is possible | How often it happens |
| Typical (median) | The middle of the sample | What is normal | What the loud moments look like |
| Busy market (p90) | Exceeded about a tenth of the time | What to budget for | The worst case |
| At capture | One live quote at the last reading | What it was doing just then | Anything about the rest of the day |
| Ticks sampled | The size of the sample | How much the row rests on | Whether the day was typical |
| Worst spike | The widest single quote in the hour | How far it can go | How long it stayed there |
Four of these summarise a day of quotes and one is a quote. Comparing a screen against the wrong one is the usual reason a feed looks wrong.