Definition and scope of slippage assumptions
Slippage is the difference between an assumed execution price (what a model, backtest, or order plan expects) and the actual execution price (what is filled). A “slippage assumption” is the set of numerical rules and inputs that convert this difference into something you can use in calculations (for example, how much adverse movement you expect between quote time and fill time, and how costs combine with it). To assess slippage assumptions, first state the exact price basis and timing window you are using:
- Assumed price: mid-price, bid/ask, last trade, or a specific reference (and whether it’s taken at decision time).
- Executed price: fill price, including whether it reflects partial fills and how they are aggregated.
- Timing: the delay between the reference price timestamp and the fill timestamp.
- Direction: adverse movement differs for buys vs sells.
Without these definitions, two datasets or models can produce “slippage” numbers that are not comparable.
Core data inputs to collect
You generally need data in four groups: execution details, market conditions, cost components, and metadata for assumptions.
- Execution and fill data (to measure realized slippage)
- Fill price per execution event (or aggregated fills with a clear aggregation rule).
- Fill timestamp(s), ideally with enough resolution to represent latency and reorder timing.
- Trade direction (buy/sell) and instrument identifiers.
- Order behavior: whether orders can be partially filled, canceled, or modified.
- Reference price data (to compute assumed vs realized differences)
- The reference price series used by the backtest or assumption rule (e.g., bid/ask spread, mid, or last).
- The timestamp alignment rule: what exact time the reference price is sampled.
- Whether the reference price is “quote time” (when the decision is made) or “model time” (when the calculation assumes a fill).
- Market condition data (to understand why slippage changes)
- Volatility proxies over the relevant horizon (e.g., ranges or return variability computed from the same reference feed).
- Liquidity proxies such as spread width and changes in spread.
- Regime indicators that are derived from price/quote behavior (not future information), used to test whether the assumption holds across conditions.
- Cost and friction data (to avoid double counting)
- Explicit costs: commissions, fees, or platform charges if they are part of your modeling framework.
- Spread treatment: whether your slippage assumption already includes spread crossing or whether spread is handled separately.
If you do not separate “price impact-like movement” from “spread and fees,” you can end up with an internally inconsistent assumption.
Provenance, timeliness, and quality checks
Assessing slippage assumptions requires checking not only what data you have, but where it came from, how old it is relative to decision/fill timing, and how reliable it is.
- Provenance: record the source type for each dataset (exchange/quote feed, broker execution reports, platform exports). Provenance must be documented so another person can reproduce the mapping from reference to fill.
- Timeliness: ensure timestamps can be aligned. If timestamps are coarse or inconsistent, slippage can be misattributed to “market movement” versus “measurement delay.”
- Consistency: the reference price series used to compute assumed slippage should come from the same environment as the execution timestamps (or at least be aligned via a documented mapping).
- Data quality: inspect missing data, timestamp gaps, duplicated events, and outliers. Outliers may represent real execution events (e.g., thin liquidity) or measurement errors.
A practical checklist is to verify that your computed slippage distribution is stable when you slightly adjust non-material settings (such as resampling frequency) and that it changes in expected ways when you segment by volatility or spread.
Evidence or example of what “assessment” means
A common approach is to test whether the assumed slippage rule can reproduce key empirical properties of realized slippage under the same reference and timing definitions. For example:
- Compare realized slippage (computed from fill vs reference) to the assumed distribution (mean, median, and tail behavior).
- Segment by direction, spread width, and volatility regime, then check whether the assumption systematically underestimates adverse outcomes.
- Confirm that the assumption’s implied cost per trade aligns with how costs and spread are modeled.