METHODOLOGY

How we measure the figures on our homepage

Our homepage shows three figures. This page explains where each one comes from, what “accurate” means, and what the figures don't tell you.

Last reviewed: 16 October 2026

Where the accuracy data comes from

The two accuracy figures come from FX Foresight's own tracking accounts, not from customers. These accounts hold sample foreign-currency payments, called requirements, each with a currency, amount and due date. When a requirement is entered, we record the mid-market rate and the timing score shown at that moment. Every day after that we save a snapshot of the rate and score. On the due date the requirement is complete and is graded once. No customer data is used or shown.

What counts as “accurate”

Each completed requirement is graded by comparing the timing score shown at entry with what the market did by the due date:

• Score of 55 or above (leaning toward acting now): accurate if waiting until the due date would have given a worse rate.
• Score of 45 or below (leaning toward waiting): accurate if the rate on the due date was better than the entry rate.
• Score between 46 and 54 (neutral, no strong call): counted as accurate, because no directional call was made. This is a limitation, and it raises the percentage compared with grading only directional calls.

“Better” and “worse” depend on direction. For a payment you're making, a lower cost in your home currency is better. For money you're receiving, a higher amount is better. Accuracy is about direction only. It does not measure how big the move was.

“1–3% typical cost difference”

This figure is a general illustration of how far a currency rate can typically move over a 30-day payment window. That range is the difference between paying on a good day and a poor day. It is not calculated from our tracking accounts and is not a measure of FX Foresight's results. The actual range for any payment depends on the currency pair and market conditions and can be smaller or larger.

“Timing accuracy across payments held longer than eight days”

This is the share of completed requirements graded accurate, counting only those where more than eight days passed between entry and the due date. Requirements across all currency pairs in the tracking accounts are included, and the figure is shown to one decimal place. Very short windows are left out because there is little time for a timing decision to matter.

“Currency-pair timing accuracy across N completed requirements”

This is the same accuracy measure for a single currency pair, using that pair's completed requirements at every window length. The number of completed requirements shown on the card is the real sample size behind the figure. For example, “89% EUR/GBP timing accuracy across 26 completed requirements” means 26 completed EUR/GBP requirements were graded. The percentage is rounded to the nearest whole number. Only major pairs (EUR/USD, GBP/USD, USD/JPY, USD/CAD, AUD/USD, EUR/GBP) are eligible, and the card rotates daily among the pairs that currently meet our evidence standard.

Our evidence standard

We only publish an accuracy figure once it covers at least 10 completed requirements and is at or above 83%. The figures are recalculated as requirements complete. If a later result falls below 83%, the homepage keeps showing the most recent result that met the standard rather than the lower one. If no result has met it, the card shows a dash instead of a number. In other words, the homepage shows the best recent qualifying result, not always the latest raw figure. Please read it with that in mind.

Market data

Rates are mid-market rates, the midpoint between wholesale buy and sell prices, sourced from an independent third-party market-data provider. Mid-market is a neutral reference. The rate your bank or payment provider quotes will usually be less favorable because it includes their spread and fees.

Limitations

See also our disclaimer and the timing calendar.