64 / 97
Non-Farm Payrolls (NFP)
Written NFP throughout.
Working definition
The monthly US employment release — published by the Bureau of Labor Statistics, typically at 8:30 a.m. Eastern on the first Friday of the month — which functions as the largest recurring scheduled volatility event in foreign exchange.
Non-farm payrolls is the headline figure of the Employment Situation report: the net monthly change in US employment outside farming, released alongside the unemployment rate and wage growth. Its market weight comes from its position in the causal chain — employment feeds directly into rate expectations — combined with genuine surprise variance against consensus forecasts. For FX microstructure, that combination makes the release less an information event than a recurring structural experiment: the same instruments, the same clock, a genuinely unknown number, every month.
What happens around the print is a compressed tour of execution pathology. Liquidity providers widen or withdraw ahead of the release, because quoting a two-sided price into an imminent binary is writing free options — spread widening here is rational, market-wide physics. Top-of-book depth thins toward a liquidity void, so the first seconds after the print produce gapped tapes and stop fills far from their triggers. Retail feeds are at their most stressed and least verifiable in the same window: quote staleness, throttled updates, and outright censoring all concentrate exactly where they are hardest to distinguish from honest chaos. The slippage a strategy experiences at NFP differs from its calm-session slippage in kind, not degree.
The release also carries a quieter lesson about data honesty: NFP is heavily revised in subsequent months, so the figure that moved the market is frequently not the figure the final series remembers. A backtest keyed to revised data trades a release that never happened — the point-in-time data problem in its most concentrated form.
Because the event recurs on a schedule, its structure is measurable rather than anecdotal. My research note How much does the range actually widen at NFP? examines the widening structure across 28 pairs and 1,820 release observations, with the per-pair figures publishing together with the content-hashed dataset that proves them.
Commonly confused with
Neighbouring concepts that get used interchangeably, and the distinction that actually separates them.
- The unemployment rate
Both arrive in the same Employment Situation report, at the same instant, from the same agency. Non-farm payrolls is the net monthly change in employment outside farming; the unemployment rate and wage growth are separate series published alongside it. Traders often attribute the move to whichever number they were watching.
- The revised figure
The print that moved the market and the number the final series remembers are frequently not the same number, because payrolls are heavily revised in subsequent months. Only the first print was knowable at the time, which makes this the point-in-time problem in its most concentrated form.
- Spread widening
The release is the occasion; widening is one of the mechanisms it triggers. Conflating them turns a question you can measure — how did this venue's widening compare with the market's — into a shrug about volatility.
- Unscheduled volatility
An unscheduled shock cannot be prepared for or studied systematically. Payrolls recurs on a known clock with the same instruments and a genuinely unknown number, which makes it less an information event than a recurring structural experiment — and the reason its microstructure is measurable rather than anecdotal.
How to measure it in your own data
A definition you cannot test is a definition you have to take on trust. This is the shortest honest route from the concept to a number you computed yourself.
- Records you need
Recorded tick data spanning the release windows across many months, the release calendar with exact timestamps, and a multi-venue reference feed over the same windows. One release is an anecdote; the schedule is what makes the structure measurable.
- What you compute
Range and quoted spread in the minutes around each print, expressed relative to a calm-session baseline in the same instrument, then aggregated across releases. Run the same computation against the reference feed so venue behaviour separates from market behaviour.
- What the answer tells you
Widening that matches the reference is physics — quoting a two-sided price into an imminent binary is writing free options, and every venue reprices it. What is worth finding is the excess, and the persistence after underlying liquidity returns. Check the feed itself in the same window before concluding anything: staleness, throttled updates and outright censoring all concentrate exactly where they are hardest to tell apart from honest chaos.
If this has already cost you
Release-window behaviour recurs on a schedule, which is what makes a venue’s conduct through it measurable rather than anecdotal.
- Feed Fidelity Assay“Is the feed my terminal shows me behaving consistently?”Will not establish: A verdict on your broker. One exported log from one terminal measures your feed as you received it — venue-side behaviour, other account tiers and intent are all outside what this data can carry.
- Outage-Window Report“What did the platform outage actually cost me?”Will not establish: What you would have done. A counterfactual is bounded, not known — the report states the range the reference market offered, not the exit you would have taken, and whether the window’s cost is anyone’s liability is a question for the firm or your adviser, with the measurement in hand.
Intake is not open yet, so none of these can be commissioned today. They are listed here so you know the measurement exists and what it would and would not settle — the launch list hears first.
Work it out yourself
Free calculators that take this concept as an input. Each shows its working, so the number it gives you can be checked rather than taken on trust.
Questions and answers
Why does Non-Farm Payrolls move currencies so much?
Two things combine. Employment sits directly in the causal chain leading to rate expectations, so the number has genuine informational weight; and there is real surprise variance against consensus forecasts, so the outcome is not already priced. A scheduled release that is both important and genuinely unknown is the rare case where everyone must reprice at the same instant.
Why was my stop filled so far from its level at the release?
Because top-of-book depth thins toward a void ahead of the print, so the first seconds after it produce gapped tapes with little resting liquidity between price levels. A stop is an instruction to trade at whatever exists when it triggers, and in that window what exists can be some distance away. The slippage differs from calm-session slippage in kind rather than degree.
Is the spread blowing out at NFP my broker taking advantage?
Not on its face — providers widening or withdrawing ahead of a binary release is rational and market-wide, and it happens everywhere. The answerable question is comparative: how this venue's widening profile compares with a multi-venue reference at the same timestamps, in magnitude, duration and whether it persists after liquidity has returned.
Why does my backtest of NFP trades not match live results?
Often because the backtest used revised payroll figures. The series is restated in subsequent months, so a simulation keyed to the final numbers is trading a release that never happened, on a surprise that nobody experienced. Only vintage first-print data reproduces what was knowable at the moment the market moved.
Related terms
Derived from the links this entry makes and the entries that link back to it.
Where the term is in build
Detector datasheets whose concepts include this term, or whose published copy uses it. Each one states the build state it has reached and the parameters it exposes, and carries no measured verdict.