Forecast Accuracy Tracking for FX Hedging: A CFO's Guide

To track forecast accuracy, save every version of your forecast with its creation date and target period. Compare each version with actuals once the period closes. Report both the size of the error (MAPE) and its direction (bias). In FX hedging this matters because your hedge ratio is only as good as the exposure forecast behind it. If you routinely forecast 20% too much foreign-currency revenue, you will over-hedge by about the same margin, and the over-hedged amount is an open speculative position.

This guide is for finance directors, treasurers and risk managers who forecast foreign-currency cash flows. It covers what to measure, how to set up the tracking, what benchmarks are reasonable, and how to turn forecast error into hedging decisions.

TL;DR

• Most teams cannot track accuracy because they overwrite old forecasts. Keep a dated, version-controlled history first, and the metrics follow.
• Measure two things together: accuracy (MAPE or WAPE) and bias (mean percentage error or tracking signal). A forecast can be close on average and still be consistently too high.
• Use forecast error to set hedge ratios by horizon. Hedge more of the near-term, high-confidence exposure and less of the distant forecast.
• Check whether human adjustments improve on the baseline forecast. If they don't, they add bias rather than insight.
• Review monthly, log your assumptions, and report error alongside exposure in the same pack.

Table of Contents

• Why forecast error is a hedging problem, not just a planning problem
• Step one: stop overwriting your forecasts
• Step two: choose metrics that answer different questions
• What counts as a good number?
• Turning error into hedge ratios
• Build the habit: a monthly routine
• Common mistakes to avoid
• Our perspective
• Put the tracking into one place
• FAQ
• Sources

Why forecast error is a hedging problem, not just a planning problem

In most departments, a forecast miss means a revised budget. In a company with foreign-currency flows, a miss has a second effect: the hedges were sized on the forecast. Two cases show the damage.

Over-forecast. You expected EUR 4 million of sales and hedged 80%, which is EUR 3.2 million. Actual sales come in at EUR 3 million. You now hold EUR 200,000 of forward cover with no underlying exposure. If the pound strengthens against the euro, you lose on that excess with nothing to offset it.

Under-forecast. You expected EUR 4 million and hedged EUR 3.2 million, but sales reach EUR 5 million. The unhedged share is far bigger than your policy intended, and your margin carries the currency move.

There is also an accounting angle. Under IFRS 9, a cash flow hedge can cover a highly probable forecast transaction, and the standard does not itself define what counts as highly probable. A documented track record of forecast accuracy is the most practical evidence you can show your auditors that your forecasts deserve that label. Talk to your auditor about how your own policy applies.

The search phrases "forecast error in hedging" and "forecast error hedging" point at the same thing: error in the exposure forecast is the main reason a hedge programme drifts away from its policy. Our own list of common FX risk mistakes puts weak exposure data near the top for the same reason.

Step one: stop overwriting your forecasts

The commonest reason organisations cannot measure accuracy is mundane. According to Onetribe's guide to measuring forecast accuracy, most teams overwrite earlier forecast versions with new ones. When March's number replaces February's, there is nothing left to compare with what actually happened.

The fix is a version-controlled repository, which can be a well-organised spreadsheet or a dedicated platform. Each saved record needs:

1. Currency and exposure type, for example EUR receivables or USD supplier payments.
2. Target period: the month or quarter the cash flow is expected to land in.
3. Forecast date: when this version was created.
4. Horizon: the gap between forecast date and target period (one month out, three months out, twelve months out).
5. Owner: who produced or adjusted it.
6. Actual, filled in after the period closes.

The horizon field is the one most teams omit, and it matters most. A forecast made one month ahead should be far more accurate than one made nine months ahead. If you only compare final forecasts with actuals, you hide the error that your longer-dated hedges actually faced.

Step two: choose metrics that answer different questions

No single number describes forecast quality. Use a small set where each metric has a job.

Accuracy and bias are different things. As the Onetribe guide notes, practitioners separate how close a forecast is from whether it is systematically high or low. A forecast with a modest error can still be dangerous if the error always points the same way. For hedging, that is the worse case: random error partly washes out across a portfolio, but bias compounds in one direction.

Tracking signal in practice

The tracking signal is the sum of forecast errors divided by the mean absolute deviation. One master's thesis on forecast monitoring describes it as the main way to detect model drift and gives an acceptable range of -4 to +4. Inside that band, errors look like noise. Outside it, the forecast is leaning one way and needs recalibrating.

A worked example for a EUR receivables forecast, in EUR thousands:

The cumulative error is +100. The mean absolute deviation is 25. The tracking signal is +4, right at the edge of the acceptable range. Each month's miss looks small, between 5% and 9% of actuals, but the pattern is a consistent over-forecast. On this evidence you would trim the hedge ratio on the more distant months now, before the problem shows up in a quarterly report.

A note on very small or lumpy flows

MAD-based and percentage measures can mislead on low-volume or lumpy data. Stephan Kolassa's research at Lancaster University argues that for low-count data it is better to evaluate the whole predicted distribution with proper scoring rules than to rely on point-forecast errors. For a company with a handful of large invoices a year in one currency, that suggests treating the forecast as a range, such as a base, low and high case, rather than a single figure. Then test whether actuals fall inside the range as often as you said they would.

What counts as a good number?

Benchmarks vary by industry, horizon and how volatile your customers are, so treat any figure as a starting point. One Onetribe benchmark for mid-sized firms puts MAPE below 5% as exceptional and 5 to 10% as good. It calls 10 to 20% acceptable and says anything above 20% points to a structural problem in the process.

Two cautions apply before you adopt these bands for FX:

• They are general financial-forecast benchmarks, not FX-specific ones. Cash-flow forecasts in a second currency, built from customer orders and payment behaviour, are often noisier than a budget line.
• Your own history is a better yardstick. Track for two or three quarters, then set targets per currency and per horizon. Improvement against your own baseline is what matters.

A public-sector reference point: the UK Treasury's budget holder forecasting handbook says poor performance shows when a department is not within 1% of its eventual outturn by the end of the sixth month of the financial year. Departmental spending is far more predictable than export sales, so do not copy that threshold. The principle transfers: set an explicit accuracy target at a defined point in the cycle and report against it.

If your team is unsure how to read these numbers or interpret them for statutory reporting, an outside accountant familiar with your structure can help. For internationally structured groups, this checklist of warning signs that a business needs accounting help is a useful reference for deciding when to bring one in.

Turning error into hedge ratios

Tracking is only worth the effort if it changes decisions. Here is how forecast error should feed a hedging policy.

Hedge in layers by confidence

Most corporate hedging policies already step the ratio down as the horizon lengthens, for example a high share in months one to three and a lower share in months seven to twelve. Your accuracy history tells you where to set those steps. If your three-month MAPE is 6% and your nine-month MAPE is 22%, a flat 70% ratio across both is hard to defend. The first is comfortably below your error and the second is not.

A simple rule: do not hedge more than the portion of the forecast you are confident will materialise. If the typical miss at a given horizon is 15%, a ratio at or below 85% leaves room for the downside case.

Hedge to the low case

For volatile exposures, size the hedge to the low end of your forecast range and treat the remainder as an unhedged layer to be added as visibility improves. This keeps the hedge from running ahead of the underlying cash flow, and it works naturally with the way the tracking signal flags over-forecasting.

Connect forecast error to your risk metrics

The uncertainty in your cash-flow forecast and the uncertainty in the exchange rate are two separate sources of risk. A value-at-risk view that assumes the exposure number is exact understates the real range. We explain how exposures feed a VaR-based approach on our page about hedging based on value at risk, and how to test such models in our note on backtesting rules for FX VaR. Treating exposure error as an input to those models, even in a simple way such as a stressed low and high case, makes the output more honest.

Build the habit: a monthly routine

The UK government's East Midlands growth hub advises a monthly check of actuals against forecast with the senior team. That cadence works for FX exposure too. A practical routine:

1. Close the month. Load actuals for each currency and exposure type into the repository.
2. Calculate error by horizon. Fill in the one-, three- and six-month-ahead forecasts for the period that just ended.
3. Update MAPE, bias and tracking signal. Use a rolling window of 6 to 12 periods rather than a single month.
4. Flag breaches. Any currency whose tracking signal leaves the -4 to +4 band, or whose MAPE crosses your own threshold, goes on the agenda.
5. Review adjustments. Compare the unadjusted baseline with the final figure to see whether sales or regional input improved accuracy.
6. Adjust the hedge programme. Change ratios, tenors or the forecast method where the evidence supports it.

Keep an assumptions log

The same Treasury handbook stresses evidence-led assumptions and advises keeping a detailed log. In practice, record the assumptions behind each forecast: the sales pipeline, expected payment terms, price changes, the budget rate. When you miss, the log tells you whether the assumption was wrong or the execution was. That distinction is what lets you fix the cause rather than just widen your tolerances.

Test whether human adjustments help

Forecast value added is the simple test of whether a manual overlay beats the baseline. Onetribe and Skillset Lab both argue that CFOs and risk managers should check whether planner, sales or marketing adjustments genuinely improve accuracy or only introduce personal bias. In FX terms: if the regional sales team keeps lifting the euro forecast and the tracking signal keeps rising, the overlay is the problem. Say so with data, not opinion.

Common mistakes to avoid

• Measuring only the final forecast. It is flattered by the fact that it was made close to the period end. Always split by horizon.
• Reading MAPE alone. Without a bias measure you cannot tell random noise from a one-way lean.
• Pooling all currencies. A large, stable USD flow can mask a poor forecast for a smaller currency. Report by currency pair or exposure type.
• Ignoring timing error. A payment that slips one week into the next month looks like a miss in two periods. Decide whether you measure by month, week or settlement date, and stay consistent.
• Never feeding results back. If accuracy figures live in a spreadsheet nobody reads, the hedge ratios will not change. Put them in the same pack as the exposure and hedge report, the way our risk reporting checklist suggests.

For more on building the forecast itself, see our guides to improving cash forecasting and to currency forecast methods.

Our perspective

We see forecast accuracy as the least glamorous and most useful part of a hedging programme. Teams spend weeks choosing instruments, then size them from a number nobody has ever checked. The instrument is rarely the weak point. The input is.

The approach we favour is plain. Keep every forecast version. Report error by horizon. Pair MAPE with a bias measure. Let the history, not the most confident person in the room, set your hedge ratios. When exposure, hedges and forecast error sit in one view, the conversation with the board shifts from "why did the rate move against us" to "was our hedge sized to what we knew". That is the question a finance team can answer and improve.

We are also expanding to Poland and Sweden, and the same principle applies in each market: whichever currencies you hold, the discipline of tracking the forecast behind the hedge does not change.

— Bartas

Put the tracking into one place

If your exposure forecasts, hedges and reports are scattered across spreadsheets, accuracy tracking is the first thing to slip. The Corphedge platform brings together currency position tracking, hedging strategy simulation, Value at Risk analysis, portfolio reporting and automated notifications, so exposure and hedge cover can be viewed together. If you want to build the underlying skills in your team first, our FX hedging course is a one-off purchase at 220 EUR. For help applying any of this to your own exposures, see operational expert advice.

FAQ

How do you track forecast accuracy for currency exposures?

Save every forecast version with its date, target period and horizon, then compare each with actuals when the period closes. Calculate MAPE for size of error and a bias measure for direction, by currency and by horizon. Review monthly on a rolling window.

What does tracking signal mean in forecasting?

It is the cumulative sum of forecast errors divided by the mean absolute deviation. A value between -4 and +4 is generally read as stable, according to one academic study of forecast monitoring. Outside that range, the forecast is drifting in one direction and needs recalibrating.

What is a good forecast accuracy level?

One benchmark for mid-sized firms treats a MAPE under 10% as good and above 20% as a sign of structural problems. FX cash-flow forecasts vary a lot by business, so set targets from your own history by currency and horizon.

How does forecast error affect a hedge?

The hedge is sized from the forecast, so over-forecasting leaves you holding cover with no underlying exposure, while under-forecasting leaves more of the exposure open than your policy intended. Setting lower hedge ratios for longer horizons, where error is larger, limits both outcomes.

Sources

• Hedge Accounting (IFRS 9), IFRS Community
• Jak mierzyć trafność prognoz finansowych, Onetribe
• Měření přesnosti forecastu, Onetribe
• Tracking Signal Usage for Time Series Forecast Monitoring, master's thesis
• Evaluating Predictive Count Data Distributions in Retail Sales Forecasting, Lancaster EPrints
• Budget holder forecasting handbook, GOV.UK
• 5 top tips on financial forecasting, East Midlands Combined County Authority

Recommended

• Quantitative FX Assessment Tips for CFOs: 2026 Guide
• FX risk management for CFOs: proven strategies to protect profits
• A CFO's Guide to Cash Flow Hedging Programs That Actually Work
• Why Use Options in FX Hedging: A Guide for CFOs