How Can Slippage Assumptions Be Measured?

What “slippage assumptions” mean in measurement

Slippage usually means the difference between the price you expected to get when you planned an execution and the price you actually received when it happened. A “slippage assumption” is a specific, stated method (and numeric input) that turns that uncertain difference into something measurable for backtesting, simulation, or analysis.

To measure slippage assumptions, you first separate the idea into two parts:

The measurable fields: which expected price, which executed price, and whether you include fees/spreads/commissions.
The comparison rule with timing: how you match planned orders to fills and which timestamps define “expected” vs “executed.”

If either part is unclear, the measured slippage may be an artifact of your methodology rather than the market or execution behavior.

A measurable framework: fields and timestamps

A practical measurement approach defines these fields for every comparable event:

Planned/expected price (P_expected): the price used at decision time. Examples include a reference quote (bid/ask/mid) at the planning timestamp, or a model-projected fill price.
Executed/actual price (P_exec): the fill price recorded when execution completes (or when you take the simulated fill).
Direction: buy vs sell, because slippage is typically computed so “adverse movement” increases the cost.
Slippage in price terms:
- For buys: Δ = P_exec − P_expected
- For sells: Δ = P_expected − P_exec
Slippage in cost terms (optional but common): Δ multiplied by position size, and optionally net of transaction costs—only if you consistently model the same cost components in both expected and executed prices.

Then you define timestamps:

T_planned: when the strategy decides and selects P_expected.
T_fill_request (optional): when the order is submitted.
T_fill: when the fill price is determined.

Measurement is only meaningful when the “expected” price and the “executed” price are tied to clear timestamps and an explicit matching rule (for example: one planned order maps to one executed fill, or you aggregate partial fills).

Two comparison options (and what to document)

You can measure slippage assumptions with either of these comparison patterns. The key is to use the same pattern consistently across your dataset.

Option A: Quote-to-fill (expected reference quote to executed fill)

Choose a reference at T_planned (e.g., bid for buys or ask for sells, or a mid-based convention).
Compute slippage from that reference to the realized fill.
Document the exact quote convention and whether the reference is bid/ask or mid.

Option B: Model-to-fill (expected model fill to executed fill)

Instead of using a raw quote, compute P_expected from a rule (for example, a conversion rule that estimates an effective fill price).
Compute slippage from the model fill to the realized fill.

Important similarity requirement: both options must define what “expected” means. If Option A uses mid and Option B uses spread-adjusted prices, comparisons across methods are not apples-to-apples.

Evidence through examples: what “measured” looks like

A measurement output should be more than one number. Common outputs are:

Distribution of Δ (mean, median, percentiles) rather than only an average.
Frequency of adverse vs favorable slippage (for buys vs sells, sign conventions matter).
Dependence on context: group by volatility regime, time of day, or spread width.

A simple example logic (no live data assumed):

For each planned event, record P_expected and T_planned.
When execution occurs, record P_exec and T_fill.
Compute Δ using the direction-aware formula.
Summarize Δ by groups that reflect execution difficulty (for instance, wider spread periods).

The “evidence” is that you can reproduce the same Δ values from the stored fields and timestamps.

Material limitations and failure modes

Several limitations often break slippage measurement and make assumptions look more precise than they are:

Timestamp mismatch: If P_expected is based on a quote that was visible later (or earlier) than intended, you can introduce systematic bias. 2) Spread and cost double counting: If your expected price already includes spread effects, but you also subtract or add spread again in cost terms, measured slippage becomes a bookkeeping error. 3) Execution mapping errors: Partial fills, order cancellations, or multiple fills per planned order can distort event-level pairing. 4) Non-stationarity: Historical slippage relationships often change with market conditions and technology. A method that fits one period may not fit another.