A single latency number rarely tells a broker whether prices are fresh or clients are receiving consistent execution. The useful audit follows a quote from source to platform and measures the order path independently.
Follow the evidence
Trace how the event could reach markets, then inspect a competing explanation.
Compare explanations
Switch lenses to see what each account explains—and what remains uncertain.
Log source and receipt timestamps, sequence gaps, stale-quote intervals, update frequency and the spread delivered to each platform. For orders, track acknowledgement and fill times, rejects, requotes and slippage separately. Compare medians and high-percentile results rather than reporting only an average. Segment tests by instrument, session, order size and market condition. Run them through more than one route and validate that system clocks are synchronised; otherwise, timestamp comparisons can mislead.
Log source and receipt timestamps, sequence gaps, stale-quote intervals, update frequency and the spread delivered to each platform. For orders, track acknowledgement and fill times, rejects, requotes and slippage separately. Compare medians and high-percentile results rather than reporting only an average. Segment tests by instrument, session, order size and market condition. Run them through more than one route and validate that system clocks are synchronised; otherwise, timestamp comparisons can mislead.
Does the published figure describe a quote’s journey, the broker’s internal processing, or a client order’s complete path? Was it measured during routine trading or around volatile releases? Which instruments, endpoints and percentile were included? Lower latency can improve a system’s responsiveness, but it does not eliminate slippage, widen or narrow spreads on demand, or guarantee best execution. Request reproducible test conditions and logs before comparing vendors.
Market data and order execution are different journeys
A broker receives prices from liquidity sources or an aggregator, processes them, and distributes quotes to its trading platform. An order then travels through a separate route for validation, risk checks, routing or internalisation, and confirmation. Network delay is only one part of each path.
Feed delay and execution delay should not be collapsed into one ‘speed’ claim. A fast quote can still lead to a slow fill, and a fast server cannot guarantee a particular price when the market has moved.
Compare distributions, not a best-case average
Log source and receipt timestamps, sequence gaps, stale-quote intervals, update frequency and the spread delivered to each platform. For orders, track acknowledgement and fill times, rejects, requotes and slippage separately. Compare medians and high-percentile results rather than reporting only an average.
Segment tests by instrument, session, order size and market condition. Run them through more than one route and validate that system clocks are synchronised; otherwise, timestamp comparisons can mislead.
Ask what the performance claim actually measures
Does the published figure describe a quote’s journey, the broker’s internal processing, or a client order’s complete path? Was it measured during routine trading or around volatile releases? Which instruments, endpoints and percentile were included?
Lower latency can improve a system’s responsiveness, but it does not eliminate slippage, widen or narrow spreads on demand, or guarantee best execution. Request reproducible test conditions and logs before comparing vendors.