When Spread Assumptions Behave Differently: Market Conditions, Mechanics, and Limits

Direct answer

Spread assumptions behave differently when the real trading environment stops matching the simplified inputs used to estimate costs. That mismatch typically appears when market liquidity changes, volatility changes, execution becomes less predictable, or additional costs (fees, commissions, financing, or varying effective spreads) enter the picture. Instead of treating spread as a constant, it helps to think of spread as a cost that depends on conditions at the moment of execution.

Mechanics and definition

A “spread assumption” is a modelling choice that represents the transaction cost created by the bid–ask spread (and sometimes execution effects tied to it). In practice, spreads can be represented in several ways:

• Fixed/constant spread: assume the same spread for all trades and times.
• Time-varying spread: use spreads that change with time (or with observed market states).
• Effective spread / slippage-aware spread: attempt to reflect not only the quoted bid–ask gap, but also how far execution prices move from quotes.

The key idea is that spread assumptions are only as accurate as the mapping between the assumed model and the actual conditions your orders face. If the model assumes stable liquidity and tight spreads, but the market becomes illiquid or fast-moving, the cost you pay (effective spread) can deviate from the cost your model expects.

Evidence or example (conditional comparisons)

Consider the same trading logic evaluated under two different market regimes, while keeping everything else constant in the model.

Option A: “Static spread” assumption

You assume a constant spread (for example, by using a single spread value across the test period). This tends to fit situations where:

• liquidity is stable,
• spreads are consistently tight,
• and execution behaves similarly across time.

Option B: “Variable spread” (or execution-aware) assumption

You allow spreads and/or execution effects to vary (for example, using spreads observed at each time step, or adjusting for slippage). This tends to better match situations where:

• liquidity fluctuates,
• volatility changes quickly,
• and order fills become less predictable.

In both cases, the “different behaviour” is conditional: the more the real market departs from the assumptions used, the larger the cost mismatch becomes. That cost mismatch can change the historical accounting of performance and the feasibility of executing trades as planned.

Limitations and failure modes

At least four material limitations can cause spread assumptions to “behave differently” than expected:

1. Liquidity regime changes: During thin trading, the quoted spread may widen, and the effective cost can increase further due to poorer fills.
2. Volatility bursts: In fast markets, prices move quickly between observation and execution; a model using a stable spread can understate execution cost.
3. Execution quality differences: Order type, latency, and whether orders rest in the book can affect the actual fill price, making effective spread deviate from quoted spread.
4. Model scope mismatch: Some spread assumptions include only the bid–ask gap, while real trading outcomes also include commissions, financing effects, and other transaction costs. If the model omits these, results may diverge.

Because of these failure modes, historical relationships do not establish future results, even if they appear consistent under one set of conditions.

Verification and next question

To independently verify where spread assumptions differ, check whether your assumptions are aligned across three layers:

1. Market data layer: Do you use bid/ask (or spread) that matches the conditions at the moments trades are executed?
2. Execution layer: Does your model reflect realistic fill behaviour (e.g., slippage or execution latency), or does it assume idealised fills?
3. Cost layer: Does the model include all relevant costs that affect the net price, not just a simplified spread?

A helpful next question is: Which part of your model is most sensitive—liquidity, volatility, or execution fill quality? Sensitivity analysis (varying spread assumptions within plausible ranges and observing how outcomes change) can reveal whether your conclusions depend heavily on one simplifying assumption.