Fixed spread in the tester hides the costs that matter most

[Amanhasnoname]

A tester set to a fixed 1 pip spread assumes costs never change. In real trading, spreads widen at rollover, around news, and during thin sessions, which is exactly where many scalping signals fire.

Strategies that look profitable with a fixed spread can collapse when variable spreads are applied. Trades that entered at 5 pm New York or just before a release suddenly cost three or four times as much.

Using historical spreads from tick data gives a more honest result. If those are unavailable, adding a spread penalty during known expensive windows is a decent approximation.

It also reveals which hours the strategy truly works. After applying realistic spreads, many systems turn out profitable only in the busiest two or three hours of the day, and the rest is a drag.

Filtering out trades when the spread exceeds a threshold, in both the test and live trading, keeps the two aligned.

Was a variable spread used in testing, and how much did it change the results?