Latency arbitrage detection is an investigation into whether trading activity repeatedly benefits from price information reaching one part of the execution chain before another. For a forex brokerage, the challenge is to distinguish a credible stale-price pattern from ordinary fast trading, volatile markets or incomplete records.
A profitable account or a short holding time cannot answer that question alone. The dealing team needs a reconstructable sequence of quotes, requests and executions, with enough context to explain the apparent advantage. This checklist sets out the evidence to collect before drawing a conclusion.
Define the suspected timing advantage
Write down what the alert is claiming. Which price may have been stale, relative to which reference, and at what point in the trade lifecycle? Distinguish a concern about quote delivery from a slow response, a bridge delay or an execution outcome. These hypotheses need different evidence.
Specify the instruments, accounts and observation window. A broad label such as “latency trader” is less useful than a precise claim that can be tested and, where appropriate, rejected. Preserve the rule version and the observations that caused the case to be opened.
Reconstruct the sequence from source records
Collect the available quote, request, server, bridge and fill records. Document what each timestamp represents, its precision and the clock that produced it. Do not calculate a cross-system delay as though two clocks were synchronised when that has not been established.
Retain the original identifiers and the relationships between records. MT5 orders, deals and positions are different entities; a request and its eventual executions should not be treated as one indistinguishable event. MetaQuotes explains the basic trading-record relationships.
If you have a fill time but no reliable record of the price available when the request was received, you may be able to review execution outcomes. You cannot automatically reconstruct the precise quote state from that fill alone.
Choose a comparable price benchmark
State the source, side and time of the reference price. A buy execution compared with a midpoint is not the same comparison as a buy execution compared with an available ask. Check instrument specifications, price precision and whether the reference was accessible within the relevant execution path.
Keep the benchmark stable across the sample and label any substitutions. Differences between feeds can reflect aggregation, spreads or timing. An apparent discrepancy should lead to a source review rather than automatically becoming evidence against the account.
Look for recurrence across comparable conditions
Compare repeated sequences, not just the most striking trades. Review the direction of subsequent price movement, activity around quote changes, holding times and the distribution of results. Keep market conditions, order types and trade sizes visible so dissimilar trades are not pooled into a misleading average.
Examine related accounts where supported by evidence. Shared devices, network relationships or coordinated timing may add context, but none independently proves an execution exploit. Use the linked-account investigation guide to keep relationship analysis separate from execution findings.
Measure the operational impact separately
Review the relevant book allocation and actual hedge execution before estimating the brokerage effect. Client gains, retained market P/L and external execution differences describe different parts of the economics. Avoid counting a price difference twice as both a trading loss and an execution cost.
Record the size and coverage of the reviewed sample, including missing records. If a conclusion depends on an assumed reference price or timing adjustment, make that assumption explicit and test whether it changes the result.
Turn the investigation into a reviewable decision
- Confirm the suspected mechanism and the evidence required to test it.
- Check clock alignment, timestamp meaning and data gaps.
- Use an appropriate and consistently applied benchmark.
- Review recurring patterns and plausible alternative explanations.
- Record the finding, owner and reasons for any operational response.
Separate an alert from a confirmed policy breach. Escalation and account controls should follow the brokerage’s review process, with a record of both supporting and contradictory evidence. Feed dismissed cases back into threshold reviews to improve the quality of future alerts.
The BrokerCNC Toxic Trading Detection workflow connects activity, behavioural signals and incident review. Ask our team which quote and execution records can be linked in your setup. Start with the broader toxic trading guide, then book a live demonstration around your investigation process.
Common latency-arbitrage questions
Does fast or profitable trading prove latency arbitrage?
No. It is a reason to inspect the relevant context, not a substitute for a demonstrated timing and price pattern.
Can every case be resolved from trade history alone?
No. Some hypotheses require historical quotes, reliable timestamps or execution-path records that a position history does not contain.
See the workflow in BrokerCNC.
Discuss your MT4/MT5 servers, operating priorities and the evidence your team needs.
Book a Live Demo →





