Do backtest spreads on exotic currency pairs match real live execution?

The two cost numbers

Historical 90th percentile spread was 0.080964% per side. One broker snapshot was 0.209606% per side. The chart shows only those two cost numbers side by side.

Three things you must not mix

First, the 2020-2023 USDNOK simulation. It covers four years of past USDNOK prices. The trade rules stay the same in each cost case. Only the cost number changes. That lets you see what cost alone does.

Second, the historical 90th percentile. That is a high point from past spread history. Past spreads sat at or below it most of the time. It is still history. It is not a broker quote you can click today.

Third, one broker snapshot. It is one single quote for USDNOK taken at one time. It shows what one broker showed at that moment, per side. It does not tell you what you will pay on your next trade, on another day, or with another broker. I compare like-for-like cost units before judging the test.

In the snapshot-cost case, the same 2020-2023 trade rules with the snapshot cost lost 13.01% a year, rounded to 13%. The trade rules did not change. The cost number did. This 13% is a stress-test result from history. It is not what happened on a funded account.

Invented arithmetic example

Invented example, not a historical trade. Take an invented trade size of $100,000. At the historical cost the one-side cost is about $81. At the snapshot cost it is about $210. Take an invented $10,000 challenge account with $200 loss room left before the challenge loss limit. The second cost alone exceeds that room by about $10, before other price movement and fees.

What to check on your own backtest

Compare your symbol, quote time and per-side spread with the cost setting in your test. Use this short list before you pay a challenge fee.

An EA is a program that applies your rules in MT5. A soft cost guess can make an EA look ready too soon. Put the cost setting next to each result you review.

• Exact symbol. Verify the exact broker symbol and specifications for your test and your quote. A wrong symbol makes the whole cost check void.
• Quote time. Note the day and hour of your broker quote. Spreads move during the day and week. Write the time next to the number.
• Cost units. Confirm that both numbers use the same per-side unit. Both numbers here use percent per side. Convert first if your test uses another unit.
• Spread plus commission. Add both parts of the trade cost in the same per-side unit. Your test must hold the full per-side cost, not only one part.
• Rerun with unchanged rules. Keep your trade rules exactly as they are and run the same period again with the measured snapshot cost. Change only the cost number.

Cost belongs next to the result

When you judge a backtest for a challenge, keep the cost guess beside the result. A backtest replays history with a set cost, and one snapshot is one single quote. Neither tells you what your broker will show when you trade. What they can do is show how firm your result is when cost moves from history to a measured quote.

Sources and related guides

• Saved spread comparison and invented cost arithmetic