Cost assumptions in scalping backtests: what's realistic?
Posted: Tue Oct 27, 2026 10:08 am
Most scalping backtests I see posted use either the default spread from the tester or a fixed number that looks optimistic. Since costs decide whether a scalper is profitable, I've tried to set up realistic assumptions.
What I use for EURUSD on a raw account, based on my live data:
What I've seen several times: a strategy at 1.5 profit factor with default costs drops to around 1.1 with realistic ones, and below 1 in the stress test. That's the typical scalping trap.
For gold I use a wider stress, because live variation is much larger.
What cost assumptions do you use? Has anyone compared their backtest assumptions with actual live costs over a few months?
What I use for EURUSD on a raw account, based on my live data:
- Spread: real tick data with variable spread where possible. If fixed, 0.2 pips for London/NY and 0.5 for Asia.
- Commission: the real round-turn rate, $7 per lot in my case.
- Entry slippage: 0.1 pips on market orders.
- Stop slippage: 0.2 pips average, plus an extra 1 pip on any stop inside a news window.
- Target slippage: zero (limit orders).
What I've seen several times: a strategy at 1.5 profit factor with default costs drops to around 1.1 with realistic ones, and below 1 in the stress test. That's the typical scalping trap.
For gold I use a wider stress, because live variation is much larger.
What cost assumptions do you use? Has anyone compared their backtest assumptions with actual live costs over a few months?