FX Execution Quality: Measuring Costs Beyond the Quoted Spread

View the accompanying illustration: FX Execution Quality: Measuring Costs Beyond the Quoted Spread

Research edition: October 8, 2026

A narrow displayed spread is only one component of FX execution quality. The final result also depends on the price when the decision was made, the time taken to execute, the quantity filled and any explicit charges. Evaluating these components separately helps explain whether a trading idea lost value during implementation. It also prevents a favourable fill on one order from being mistaken for evidence of consistently superior execution.

This October 8, 2026 research edition develops a practical transaction-cost framework. It does not rank a broker or claim observed performance for any venue. All prices, quantities and costs used below are hypothetical.

Define the benchmark before collecting results

A decision-price benchmark measures execution against the market when the trading decision was formed. An arrival-price benchmark uses the market when the order reached the execution process. These timestamps answer different questions. A delay between them may explain part of the result, so substituting one for the other after the trade can obscure a genuine cost.

The Global Foreign Exchange Committee's algo and TCA templates encourage standardised information for assessing algorithmic execution. Their role is to support transparency and comparability. For an individual analytical workflow, the corresponding principle is to define the dataset and benchmarks consistently before judging performance. No template eliminates the need to understand the actual instrument and execution terms.

Use a consistent sign convention

For a fully executed purchase of a linear base-currency quantity Q, a simplified decision-price cost is Q multiplied by the difference between the average fill price and the decision benchmark, expressed in quote currency. For a sale, reverse the sign so that an adverse execution still produces a positive cost. Add explicit fees using the appropriate currency conversion.

Suppose a hypothetical EUR/USD purchase of EUR 100,000 has a benchmark of 1.1000 and an average fill of 1.1002. The difference is two pips, equivalent to USD 20 for that quantity. A hypothetical USD 7 commission increases the measured total to USD 27. This calculation assumes complete execution and compatible price conventions; it does not estimate an actual provider's costs.

If the benchmark is a mid-price, crossing the spread contributes to the difference. If the benchmark is already an executable ask for a purchase, the spread treatment is different. Do not add a spread charge mechanically after calculating a benchmark difference that already incorporates it. Document what is embedded and what is separately charged.

Account for partial and missed execution

An order that fills only a favourable fraction can appear inexpensive if the unfilled amount is ignored. For a specified end-of-measurement price, a simplified implementation-shortfall calculation for a purchase combines the cost on executed quantity with the price change on the unexecuted target quantity, then adds fees. The end price and cancellation rule must be chosen consistently.

Consider a target of EUR 100,000, half executed two pips above the decision price, with the remaining half unfilled when the market stands five pips above that benchmark. The simplified executed cost is USD 10 and the opportunity component USD 25, before fees. Reporting only the first number would omit a material part of implementing the original decision.

Opportunity cost is model-dependent because the counterfactual trade was not executed. Present it as an analytical estimate rather than an observed cash payment. Explain how cancellations, target changes and strategy-driven decisions are treated so that a legitimate change in the investment decision is not automatically attributed to execution failure.

Distinguish observable costs from causal claims

Market movement during execution is observable; the amount caused by a particular order is harder to identify. A post-trade price move may include unrelated news, general market flow or the strategy's information content. Calling every adverse move market impact overstates what the data establish. Use precise descriptions and identify the assumptions needed for causal attribution.

Compare results across matched conditions: currency pair, order size, session, volatility, urgency and execution method. A provider handling larger urgent orders cannot be assessed fairly against one executing small passive orders using an unadjusted average. Include enough observations to describe the distribution and show dispersion, not only a single mean.

Build a usable execution dataset

Record intended quantity, decision time, arrival time, order type, applicable bid and ask, acknowledgements, fills, rejects, cancellations and explicit charges. Use consistent clocks and retain the instrument specifications. Clock errors can make an apparent latency statistic meaningless, while inconsistent currency conversion can distort monetary comparisons.

For an EA, compare intended orders with actual broker responses. Separate strategy decisions from order-routing behaviour. A trade absent from the fill history could reflect a deliberate signal cancellation, a rejected request or a connectivity problem. Those outcomes have different operational implications and should not be combined into one unexplained category.

Review costs against the strategy's edge

Relate implementation costs to the expected scale and holding period of the strategy. A small cost may be significant for a frequent short-horizon strategy and less material for a slower position. Stress the assumptions under wider spreads, reduced liquidity and larger order sizes. Historical average costs are not a guarantee that future execution will remain similar.

The final review should answer three questions: what was measured, how comparable the observations were and which changes might improve implementation without changing the trading objective. A disciplined TCA process creates an evidence trail; it does not replace risk controls or establish a universally best execution method.

Source: Global Foreign Exchange Committee, algo and TCA templates. The calculations and data-review framework are original educational illustrations.

Educational analysis only. Execution quality and strategy profitability are related but distinct assessments, and both remain uncertain in future markets.