Not everything the calendar prints in red is a volatility event. The Fed decision hour averages 73.2 pips of EUR/USD range — 5.21 times a normal bar at the same hour — while weekly unemployment claims, the most frequent high-impact release on the calendar, manages 1.21x. This page measures 3,963 US high-impact releases from 2007-2026 against bid-side bars, and finds that the size of an NFP surprise does not predict the size of the move.
Key takeaways
- The Fed rate decision hour averages 73.2 pips of EUR/USD range — 5.21x a normal bar at the same hour (n=91). It is the most violent scheduled hour of the FX week by a wide margin.
- NFP averages 63.6 pips, 2.69x a normal 12:00/13:00 UTC bar (n=180). CPI averages 49.5 pips, 2.08x (n=82).
- The size of the NFP surprise does not predict the size of the move. Sorted into terciles by how far actual missed forecast, EUR/USD release-hour range comes out at 60.7, 64.6 and 65.5 pips, with the median flat near 57 pips.
- Weekly unemployment claims moves EUR/USD 1.21x a normal hour, gold 0.97x and US100 0.49x — for some instruments the release hour is quieter than normal.
- On the pooled set, the first 15 minutes' direction still holds four hours later 63.5% of the time.
vs a normal hour
EUR/USD range
measured
intact after 4h
Release-hour volatility, EUR/USD
The average range of the hourly bar containing the release, against the average range of that same clock hour on weekdays with no high-impact US release in it.
| Release | Releases | Avg range pips | Median | vs a normal hour |
|---|---|---|---|---|
| Fed rate decision | 91 | 73.2 | 65.4 | 5.21x |
| Non-farm payrolls (NFP) | 180 | 63.6 | 56.6 | 2.69x |
| FOMC meeting minutes | 99 | 36.8 | 33.0 | 2.67x |
| CPI (inflation report) | 82 | 49.5 | 39.6 | 2.08x |
| Retail sales | 175 | 37.3 | 33.8 | 1.59x |
| ISM manufacturing PMI | 151 | 36.9 | 32.0 | 1.46x |
| ISM services PMI | 129 | 34.4 | 29.5 | 1.35x |
| GDP (advance) | 54 | 31.9 | 27.3 | 1.35x |
| Durable goods orders | 112 | 31.1 | 28.1 | 1.33x |
| PPI (producer prices) | 161 | 30.9 | 27.8 | 1.31x |
| Consumer confidence (CB) | 132 | 32.2 | 28.8 | 1.27x |
| Weekly unemployment claims | 98 | 28.5 | 23.8 | 1.21x |
| Any high-impact US release (pooled) | 2,279 | 38.8 | 31.6 | 1.82x |
The Fed's 5.21x needs explaining, because it does not mean the Fed matters five times more than NFP. Part of it is when the Fed prints. A 14:00 ET decision lands at 18:00 or 19:00 UTC, and those hours are otherwise nearly asleep: the non-event EUR/USD baseline is 14.3 pips at 18:00 and 13.0 at 19:00, against 22.6-26.7 pips through 12:00-15:00. NFP lands in an hour that is already among the busiest of the day, so it has less headroom to multiply. In absolute pips the two are much closer (73.2 against 63.6) than the multipliers suggest. Both readings are true and they answer different questions — absolute range for position sizing, the multiple for “is this hour unusual”.
The same releases, other instruments
| Release | EUR/USD | GBP/USD | USD/JPY | Gold | US100 (Nasdaq) | US500 (S&P) |
|---|---|---|---|---|---|---|
| Fed rate decision | 5.21x | 4.59x | 4.66x | 3.27x | 2.19x | 2.33x |
| Non-farm payrolls (NFP) | 2.69x | 2.28x | 3.33x | 2.25x | 1.9x | 2.03x |
| CPI (inflation report) | 2.08x | 1.93x | 2.66x | 1.58x | 2.68x | 2.64x |
| FOMC meeting minutes | 2.67x | 2.28x | 2.37x | 1.74x | 1.1x | 1.19x |
| Retail sales | 1.59x | 1.42x | 1.75x | 1.2x | 1.01x | 1.19x |
| ISM manufacturing PMI | 1.46x | 1.34x | 1.55x | 1.06x | 0.98x | 1.09x |
| Weekly unemployment claims | 1.21x | 1.14x | 1.23x | 0.97x | 0.49x | 0.67x |
| Pooled high-impact | 1.82x | 1.65x | 1.84x | 1.31x | 1.12x | 1.23x |
Three patterns are worth pulling out. USD/JPY is the most NFP-sensitive instrument we carry at 3.33x, ahead of EUR/USD. CPI is the one release the Nasdaq cares about more than FX does (2.68x on US100 against 2.08x on EUR/USD) — an inflation print is a rates story and an index is a duration asset. And several “high impact” releases are not events at all for some instruments: ISM manufacturing runs 0.98x on US100 and weekly claims 0.49x, meaning the release hour is calmer than an ordinary hour of the same clock time.
The surprise does not predict the move
The intuitive model — a bigger miss against forecast produces a bigger move — is not what 180 NFP releases show.
| Surprise tercile | Miss vs forecast | Median miss | Mean range pips | Median range |
|---|---|---|---|---|
| Smallest third | 0-30K | 17K | 60.7 | 56.8 |
| Middle third | 32-82K | 56K | 64.6 | 55.8 |
| Largest third | 82K+ | 114K | 65.5 | 56.9 |
A 114K median miss produces a median move of 56.9 pips; a 17K miss produces 56.8. The mean drifts up by about 8% across the entire range of surprises, which is close to nothing given how widely the underlying data is spread. Beats and misses are also symmetric: 100 beats averaged 62.0 pips of range, 78 misses averaged 66.5.
The reasonable interpretation is that the scheduled uncertainty moves price, not the number. The market clears its positioning in the first minutes either way, and the unemployment rate and average hourly earnings print in the same instant, so a headline beat can arrive alongside a soft internal. If you have been sizing news trades on how big you expect the miss to be, this is the finding to argue with.
Fade or follow?
Using M15 bars. The M15 files are rolling 100,000-bar windows, so this section covers roughly 2022 onwards and the samples are far smaller than the hourly tables above — read the release count column before the percentages.
| Release | Releases | Release-bar range 15 min, pips | 60-min displacement | First 15 min direction holds 4h |
|---|---|---|---|---|
| Pooled high-impact | 327 | 29.5 | 23.2 | 63.5% |
| NFP | 36 | 55.4 | 37.9 | 72.2% |
| Retail sales | 31 | 29.2 | 23.7 | 74.2% |
| ISM manufacturing | 22 | 26.5 | 22.1 | 77.3% |
| CPI | 18 | 75.4 | 55.5 | 50.0% |
| Fed rate decision | 12 | 41.8 | 43.2 | 66.7% |
63.5% persistence on the pooled set is a real edge over a coin flip and a poor foundation for a strategy on its own. It says nothing about how far price travelled against you first, and the 60-minute displacement (23.2 pips) is smaller than the release bar's own range (29.5 pips), which is the signature of a spike that partly retraces. CPI's 50.0% on n=18 should not be read as “CPI reverses”; it should be read as n=18.
Methodology
- Calendar: the 95,799-row economic calendar the simulator ships, 2007-2026, carrying currency, impact rating, event name, actual, forecast and previous.
- Source defect 1 — missing release times. 24,699 rows (25.8%) carry a midnight placeholder instead of a clock time. For releases issued at an immovable US Eastern time we reconstruct the moment from the date (08:30 ET for BLS/BEA/Census prints, 10:00 ET for ISM and the survey houses, 14:00 ET for the FOMC) under US Eastern daylight-saving rules, and validate that reconstruction against the rows that do carry a time — it is used only for events where it lands in the correct UTC hour on at least 95% of validation rows. Non-farm payrolls validated at 100.0% of 59 timed rows and PPI at 100.0% of 48. Events that failed are used from timed rows only — the Philly Fed index at 19.1% (its release time moved from 10:00 to 08:30 ET), the UoM sentiment surveys at 55.6%, and Federal Funds Rate at 88.1% (the FOMC published at 14:15 ET before 2013).
- Source defect 2 — bad daylight-saving labels. The calendar's local UTC offsets drift and are wrong on a minority of rows, which lands a release in the neighbouring bar. Releases whose UTC hour is not one of the hours holding at least 15% of that event's own history are dropped.
- Resolved sample: 3,963 USD high-impact release rows — 3,367 from the source clock and 596 reconstructed.
- Price data: bid-side H1 bars, EUR/USD from 2010-07-15 (gold from 2009-11-12, indices from 2013-05-22); M15 bars are rolling 100,000-bar windows reaching back to about 2022-07-25.
- Baseline: the mean range of the same clock hour on weekdays containing no high-impact US release at all, per instrument. Bars where high equals low, and weekend bars for non-crypto instruments, are excluded from both the baseline and the event set.
- Families: releases that print simultaneously are collapsed to one release moment before measuring, so the CPI and core-CPI lines count once rather than three times.
- Caveat that matters most here: ranges are bid-side and exclude the spread, which is exactly the wrong assumption for news trading — the spread widens hardest at these moments. Treat every pip figure as a ceiling on what was capturable, and see the spread cost study.
- What this is not: a strategy test. Range is not profit, and none of these figures account for slippage, which is the dominant cost at a release.
What to do with this
If you flatten before news, flatten before the Fed rather than before the calendar. One event on the list runs above 3x, three run above 2x, and the median high-impact release runs 1.82x. Stop forecasting the number — across 180 releases the size of the NFP miss explained almost nothing about the size of the move. Size on absolute range, not the multiple, because the Fed's headline figure is partly a statement about how quiet 18:00-19:00 UTC normally is.
Then check your own instrument: weekly claims at 0.49x on US100 means an event the calendar prints in red is, for that market, a below-average hour. The simulator marks these releases on the chart and lets you step through any of them bar by bar, so you can watch a spike hit an actual position — see trading high-impact news and how to use an economic calendar.
Cite or republish this data
This research is free to quote, screenshot, or republish — in an article, a video, a newsletter, or a course — as long as you credit FxBacktest with a link back to this page. Copy the attribution below.
Frequently asked
How many pips does NFP move EUR/USD?
Across 180 releases, the hourly bar containing the release averaged 63.6 pips of range with a median of 56.6 — about 2.69x a normal EUR/USD bar at the same hour. USD/JPY moves more in relative terms, at 3.33x.
Which economic release moves forex the most?
The Fed rate decision. Its release hour averages 73.2 pips of EUR/USD range, 5.21x a normal bar at the same clock hour, ahead of NFP at 2.69x and CPI at 2.08x.
Does a bigger NFP surprise mean a bigger move?
Not measurably. Splitting 180 releases into terciles by how far actual missed forecast gives mean EUR/USD release-hour ranges of 60.7, 64.6 and 65.5 pips, with a flat median near 57 pips. The scheduled uncertainty appears to matter more than the number itself.
Does the initial news move continue?
On the pooled set of high-impact US releases, the direction of the first 15 minutes is still intact four hours later 63.5% of the time. That beats a coin flip and says nothing about how far price ran against the position first.
Is weekly unemployment claims worth trading?
On this evidence, rarely. It is the most frequent release the calendar flags high impact and it moves EUR/USD 1.21x a normal hour, gold 0.97x and US100 0.49x.
Where does this release data come from?
The 95,799-row economic calendar the FxBacktest simulator ships, covering 2007-2026 with actual, forecast and previous values, joined to the same Dukascopy bid bars the simulator replays. 24,699 calendar rows carry no release time and are either reconstructed from the fixed US Eastern release schedule and validated, or dropped.