In simple terms
Latency is the time elapsed between two defined events, such as order submission and broker acknowledgement. There is no single delay: interface, network, controls, routing, venue, and response delivery form separate segments.
How the term is used
A useful measurement states its start, end, clock, unit, and percentile. An average can hide rare but important tails; unsynchronized timestamps can assign delay to the wrong segment. Where data allow, analysts separate decision-to-submit, submit-to-acknowledgement, acknowledgement-to-fill, and return of the outcome.
For metrics, diagnosis, and controls, see Execution latency.
Limit
Latency and slippage may be correlated, but they are not synonyms. A worse price may also reflect volatility, depth, routing, or size; a longer delay may leave the fill unchanged. Clock synchronization makes timestamps comparable, but does not by itself measure the full path.
Sources
- U.S. Securities and Exchange Commission, Frequently Asked Questions: Rule 605 of Regulation NMS (April 1, 2026) — addresses receipt and execution timestamps and speed statistics for covered U.S. orders; it is staff guidance, not a global end-to-end latency standard.
- EUR-Lex, Commission Delegated Regulation (EU) 2025/1155 — synchronisation of business clocks — current EU rules for entities within the MiFIR scope; Articles 11–16 apply from 2 March 2026. Synchronized clocks make timestamps comparable, but do not by themselves measure end-to-end latency.