Reducing size when the spread is wider than usual

[Fairman]

Reducing size when the spread is wider than usual

Most people treat the spread as a cost to minimise and use a filter that blocks trades when it's too wide. That's sensible, but there's a middle ground that's less discussed: trading at reduced size when the spread is somewhat wider than normal, but not extreme.

The reason is that the spread is a fixed cost per trade in pips. If your typical target is 6 pips and the spread is normally 0.5, the spread is a small part of the trade. If the spread goes to 1.5, that's a quarter of your target gone at entry. The expected value of the trade drops, even if the setup is the same.

Reducing the position size when expected value is lower is one way to handle that. You still take the trade, since the setup is valid, but you risk less on it because the odds are slightly worse.

My rough rule: normal spread, normal size. Spread up to double the normal, half size. Beyond double, no trade. It's a simple step function and easy to follow.

Some might say if the trade is worse, don't take it at all. That's reasonable. But skipping too many trades has its own cost, and in some sessions the spread is often a bit elevated. A size reduction is a compromise between skipping and full exposure.

[Fairman]

Wider spreads often mean the market is less liquid, and that affects more than the entry cost. Slippage on the stop is more likely too, and the exit spread is also wider. So the real cost increase is often more than the spread you see at entry. That's another argument for reducing size, since the total cost of the round trip has gone up. The exit side often gets overlooked. You might enter at a normal spread and then find it widened by the time the trade closes, especially near rollover or news. Planning for that is part of why size reduction makes sense in those conditions.

[Fairman]

The normal spread should be defined for each pair and session. A spread of 1 pip on GBPJPY during London is normal. The same on EURUSD would be high. A table of typical spreads per pair and hour, built from your own data, makes this rule work properly. Using one fixed normal value for all conditions leads to the rule triggering too often or not enough. A spreadsheet with spreads recorded at each entry over a few weeks is enough to build this. Once you have it, the normal value for each hour is clear and the rule becomes easy to apply.