How to read a broker cost table without being misled

Every broker publishes a cost table, and every one of them is arranged to be read favourably. Not by lying, usually, but by choosing which column to show and which convention to use. Here is how to read one so it tells you something.

Six questions to ask any cost table

1. Which spread convention is this? Minimum, average, or typical? Averaged over what window? A minimum from one broker against an average from another is not a comparison. If the page does not say, treat the number as unverified.

2. Is commission in the total, or in a footnote? A raw or zero account shows a very tight spread precisely because the cost was moved into commission. The only figure worth comparing across account types is spread plus commission, expressed in the same unit.

3. What unit is the number in? Pips, points, currency per lot, percentage. A table that mixes units between rows is doing the reader no favours, and converting them yourself is where mistakes get made.

4. Does it say when it was measured? A cost figure without a timestamp is a claim, not a measurement. Spreads move continuously, so a number with no time attached cannot be checked and cannot be wrong.

5. Which instruments are in the table? A broker competitive on major pairs may be ordinary on gold or indices. A single headline number covering everything describes nothing you actually trade.

6. Is swap anywhere on this page? Usually not, because it is per night rather than per trade. If you hold positions overnight it belongs in your comparison.

The comparison most tables quietly avoid

Put spread and commission in one column, converted to the same unit, for the instrument you actually trade, measured at the hour you actually trade it. Almost no broker page presents this, because it is the comparison where account types stop looking different from each other and start looking like a ranking.

That is the table we built. Our [spread comparison](/compare/spread) samples live from broker servers every five minutes and marks any reading older than ten minutes as stale rather than presenting it as current. The [cost calculator](/tools/calculator) then puts spread, commission, swap and rebate into one figure for a position size you choose.

Two traps worth naming

A missing number is not a zero. When a table shows a blank or a dash, that means not measured, not free. Treating unmeasured as zero systematically flatters whichever broker reported least, which is exactly backwards.

A total is only comparable if the components match. Two totals built from different assumptions about volume, holding time or instrument are two different questions with one answer each, not two answers to one question.

What to do with all this

Pick the instrument you trade most and the hours you trade it. Compare spread plus commission across account types at that instrument and those hours. Add swap if you hold overnight. Subtract rebate, which offsets the entry cost and not the holding cost.

That single arithmetic, done once with real numbers, will settle account choice better than any published table.

Next

• [What a lot really costs, in three components](/bai-viet/real-cost-of-one-lot)
• [Advertised spread and the spread you actually get](/bai-viet/advertised-vs-actual-spread)
• [What overnight financing costs on a position you hold](/bai-viet/what-swap-costs-you)