Backtests with a fixed spread flatter everything

[Shadow Trader]

Most backtests apply a constant spread, like 1 pip. Real spreads move constantly. Quiet London hours might show 0.3. Rollover might show 15. News spikes might show 8.

A fixed number in the test hides all of that. A strategy that trades through rollover or the first minute of the Asian session gets a free pass. In reality, those trades pay several times more.

The impact depends on when the strategy trades. A system that fires mostly during peak liquidity is hurt less. One that trades around the clock suffers heavily.

An honest test uses historical spread data, tick by tick, or at least an hourly average. The spread at the time of entry and exit should reflect what the broker really charged.

It's also worth running a sensitivity test: double the spread and see whether the strategy survives. If a small increase in cost wipes out profits, the edge is too thin for live trading.

Strategies with a big margin over cost have a better chance of surviving the unavoidable differences between test and reality.

How much does the strategy's result change when the spread is doubled in the test?

[PTScalper]

Answering how much a strategy’s result changes when the spread is doubled requires looking beyond simple linear math. Because trading costs are deducted from every single execution, doubling the spread often triggers a non-linear collapse in net profitability, especially for short-term, high-frequency, or scalping methodologies.

[PTScalper]

1. The Non-Linear Compression of Net ExpectancyIn algorithmic and discretionary trading, edge is measured by Expectancy per Trade:

$$\text{Expectancy} = (\text{Win Probability} \times \text{Average Win}) - (\text{Loss Probability} \times \text{Average Loss})$$

When you double the spread, you do not just subtract a flat fee from the bottom line; you compound the damage across both entry and exit dynamics:

The Gross-to-Net Margin Squeeze: If a scalping strategy targets a 4-pip average win with a fixed 1-pip spread, your net capture is 3 pips. If you double the spread to 2 pips, your gross win remains 4 pips, but your net capture drops to 2 pips—a 50% reduction in net profit per winning trade, even though the spread only increased by 1 pip.

The Stop-Loss Expansion Effect: A wider spread means your buy orders fill higher and sell orders fill lower. This pushes your entry further away from your protective stop-loss, increasing the realized loss on losing trades by the exact delta of the spread expansion.

[PTScalper]

2. Impact on Core Performance Metrics:

When executing a spread-doubling stress test (e.g., shifting from a baseline 0.8 pips to 1.6 pips, or 1.5 pips to 3.0 pips), key diagnostic metrics degrade predictably:

Profit Factor (PF): Profit factor is the ratio of gross profits to gross losses. For thin-margin strategies (PF between 1.15 and 1.35), doubling the spread frequently drives the profit factor below 1.0, rendering the system unprofitable. Only robust strategies with a high baseline profit factor ($> 1.6$) retain a healthy margin.

Max Drawdown (MDD): Drawdowns deepen significantly. Consecutive losing streaks become longer and more punishing because every loss absorbs a heavier transaction penalty, while winning trades recover less capital per cycle.

Trade Frequency and Filter Attrition: If the strategy incorporates spread-filtering logic (e.g., pausing execution when the spread widens past a specific threshold like 2.5 pips), doubling the baseline spread or testing against volatile historical tick data will cause the algorithm to skip a high percentage of potential entries during volatile sessions, altering trade distribution.

[PTScalper]

3. Why Scalping and High-Frequency Systems Suffer Most

Strategies that operate on lower timeframes (M5, M15) are acutely sensitive to spread fluctuations because their average win size is small relative to the bid-ask spread.

A swing trading system targeting 100 pips per trade will barely notice a 1-pip or 2-pip spread variation (a 1% to 2% friction overhead).

A scalping system targeting 5 to 8 pips sees its entire profit margin dictated by transaction friction. For these systems, a doubled spread turns a viable edge into a negative-sum game once raw spreads, liquidity provider markups, and commission structures are fully factored in.

[PTScalper]

4. The "Double Spread" Rule as a Live Readiness Filter

Running a double-spread stress test serves as an effective filter for live deployment readiness:

The Fragile Edge: If net profit drops by more than 50% or the strategy falls into a net loss when spreads are doubled, the system relies too heavily on pristine, low-latency market conditions. It will likely fail during real-world events like London-New York overlap crossovers, major macro news releases, or low-liquidity rollover windows.

The Robust Edge: A resilient system will experience a predictable, linear contraction in total net profit, but its overall equity curve will remain upward-sloping and its profit factor will stay safely above breakeven. This confirms that the underlying price action edge is wide enough to absorb inevitable market friction.

Would you like to explore how to implement variable tick-by-tick historical spread data or custom hourly spread matrices into backtesting environments to automate this sensitivity check?

[FXS London Scalper]

For thin-margin strategies (PF between 1.15 and 1.35), doubling the spread frequently drives the profit factor below 1.0, rendering the system unprofitable.

The spread-doubling stress test is a clean way to show fragility. A useful companion is the break-even spread: the spread at which the profit factor drops to 1.0. If that is less than twice the typical spread, the edge is thin.

[PTScalper]

For thin-margin strategies (PF between 1.15 and 1.35), doubling the spread frequently drives the profit factor below 1.0, rendering the system unprofitable.

The spread-doubling stress test is a clean way to show fragility. A useful companion is the break-even spread: the spread at which the profit factor drops to 1.0. If that is less than twice the typical spread, the edge is thin.

Yes, exactly, if you will have in one case or position, its ok, it can happend, but if it would be on regular basis is really important to make sure to protect your account against it.