Skip to content
Learning path Bronze Understand and protect

Slippage: difference between benchmark and executed price

Slippage is the difference between a stated reference price and the actual execution price. It can be favourable or adverse and is not automatically the same as spread or market impact.

In plain terms — Slippage measures how far the average price actually obtained differs from the price chosen as a reference. Unless the benchmark price and time are specified, the number cannot be verified.

Slippage is an execution difference. For a purchase, it is often expressed as average executed price minus reference price; for a sale, the sign is reversed so that a positive value represents a cost.

buy slippage  = average executed price − reference price
sell slippage = reference price − average executed price

Slippage can be adverse or favourable. An order may receive price improvement, while in a fast market it may execute at a worse price. Slippage therefore does not necessarily mean a loss, broker error or manipulation.


Which reference price?

Common measures use different references:

Benchmark Question answered
bid/ask quote on receipt how far does the fill differ from the liquidity displayed when the order arrived?
midpoint how far is the fill from the centre of the quote?
arrival price how much did implementation change from the decision or arrival time?
signal price what did it cost to turn a recorded signal into a trade?
interval VWAP how does execution compare with trades during the period?

No benchmark is universally best. A backtest often uses the bar or signal price; a transaction cost analysis may use more granular timestamped quotes. The comparison must retain the feed, time zone, latency and rounding rules.


Why the price can change

The market continues to evolve after an order leaves the interface. The observed difference may reflect several factors:

  • available quantity smaller than the order size;
  • quote changes during transmission and routing;
  • execution across several levels of the order book;
  • volatility and liquidity cancellations;
  • delay between signal, submission and receipt;
  • rules of the order type or venue;
  • impact generated by the order itself.

A market buy order, for example, may receive fills at several asks. Its average price must be weighted by quantity. Once a stop order is triggered, it takes on the behaviour of the child order specified in its definition; the trigger is not a guaranteed execution price.


Slippage, spread and market impact

The terms describe different objects.

Term What it describes
distance between the best bid and best ask at a given time
Slippage observed difference between benchmark and executed price
component of price movement attributable to the order itself
Fee explicit charge by a service or venue

Crossing the spread may appear in slippage when the benchmark is the midpoint. Book walking may be predictable from displayed depth and, at the same time, contribute to the order's impact. To avoid double counting, the Transaction costs entry separates the components before adding them.


Measurement with partial fills

For an order executed in several parts:

average price = sum(fill price × fill quantity) / total executed quantity

The unexecuted quantity does not disappear from the analysis. It may represent an opportunity cost or residual risk, but should not be converted arbitrarily into price slippage. Cancellations and rejections must be recorded as distinct outcomes.

It is useful to segment results by:

  • instrument and venue;
  • time band and session;
  • order type;
  • size relative to depth or volume;
  • volatility and market conditions;
  • broker, routing path and feed.

An aggregate average can hide a tail of highly adverse executions. Median, percentiles and regime-specific distributions help describe risk without pretending it is stable.


Possible controls and their trade-offs

  • A limit order controls the worst acceptable price if it executes, but may remain partly or wholly unexecuted.
  • Reducing size may consume less depth, but changes the desired position.
  • Splitting the order may contain visible impact, but increases exposure time to market movement.
  • Avoiding unstable windows reduces some risks, but may forgo moments essential to the strategy.
  • Improving infrastructure and routing may reduce controllable delays without eliminating price risk.

No procedure can therefore eliminate slippage, non-execution probability and market risk at the same time.

Common mistake — Applying a fixed percentage of slippage to every instrument and regime in a backtest without comparing it with actual executions. It is a modelling assumption, not a constant property of the market.


Sources