We ran 30 strategy backtests twice — once paying the real spread, once with it set to zero. Everything else was identical. 8 of them are profitable only in the version where trading is free.

The result

Profitable paying the spread	16 of 30
Profitable if the spread were free	24 of 30
Pooled result, as traded	-7,408 pips
Pooled result, spread ignored	+6,455 pips
Total cost of the spread	13,863 pips

The same set of strategies is a +6,455 pip winner in a world with no costs and a -7,408 pip loser in this one. Nothing else changed — same entries, same exits, same bars.

The eight that only work when trading is free

These cross from profitable to losing purely because of the spread. If you have ever backtested something, liked the curve, and then watched it bleed in a live account, this is the most likely reason.

Pair	Strategy	PF as traded	PF if the spread were free
EURUSD	rsi	0.9	1.0
USDJPY	macd	0.99	1.06
AUDUSD	bollinger	0.99	1.17
GBPJPY	ema_cross	0.9	1.07
GBPJPY	rsi	0.85	1.02
GBPJPY	macd	0.83	1.0
GBPJPY	supertrend	0.86	1.06
USDCAD	supertrend	0.77	1.01

Where it hurts most

Pair	Strategy	Trades	Spread	PF as traded	PF if free	Cost
GBPJPY	rsi	714	2p	0.85	1.02	1,692 pips
GBPJPY	macd	631	2p	0.83	1.0	1,474 pips
GBPJPY	bollinger	649	2p	0.81	0.95	1,227 pips
GBPJPY	supertrend	434	2p	0.86	1.06	1,138 pips
GBPJPY	ema_cross	422	2p	0.9	1.07	924 pips
USDJPY	rsi	480	1p	1.05	1.2	857 pips
USDCAD	bollinger	241	1.4p	0.72	0.96	737 pips
USDCAD	supertrend	198	1.4p	0.77	1.01	575 pips

Two things drive the cost, and they multiply. The spread on the pair — GBPJPY at 2.0 pips is twice EURUSD at 0.8. And how often you trade — a strategy taking 700 trades pays that spread 700 times.

That combination is why the worst rows above are all the same pair. A high-frequency strategy on an expensive cross is paying the spread twice over, and no amount of indicator tuning gets that money back.

The arithmetic nobody does first

Before testing anything, multiply. A strategy averaging 400 trades in six months on a 2-pip pair pays 800 pips in spread. If its edge is 3 pips a trade, the edge is 1,200 pips and it keeps a third of it. If its edge is 2 pips a trade, it has no edge at all, and no amount of parameter tuning will create one — tuning only ever changes what the trades are, not what they cost.

This is also the honest reason most scalping strategies fail. They are not wrong about direction often enough to explain the losses. They are right just often enough, and then hand it all to the spread.

What to do with this

Only backtest with costs on. A backtester that does not charge the spread is not measuring trading; it is measuring a different game that happens to use the same charts.
Count your trades before you count your pips. Trade frequency is a cost multiplier, and the strategies that look most active are usually the ones paying the most.
Prefer cheap pairs for short-term work. The same signal on EURUSD at 0.8 keeps more than on GBPAUD at 2.5.
Treat any published backtest without a spread column as fiction.

Check it yourself

Our free strategy tester charges each pair’s real spread on every trade, and the profit calculator shows what a round trip costs per lot on each pair. Neither needs an account.

What the spread actually costs you