How to Trade NFP: The First 5 Minutes on Six Real Reports

How to trade NFP has no honest one-line answer, and this page gives no entry, stop or target. The jobs report lands at 08:30 New York, and its first five minutes have run 167 pips, pointed the way for a whole day, or settled nothing. Below are six real reports, and what our numbers do and do not support.

All times are New York, a candle is named by the time it opens, and every price is Dukascopy's bid, from our own charts.

What is NFP, and when does it come?

NFP is nonfarm payrolls: the number of jobs US employers added outside farming, and the headline of the Employment Situation report. The same release carries the unemployment rate, average hourly earnings and revisions to earlier months. Traders read it as a guide to interest rates, so in forex it arrives through the dollar.

It comes at 08:30, and it is called a first-Friday event. Take the date from the official schedule instead: on our calendar 8 May 2026 was the second Friday of its month, and 2 July was a Thursday. How the release sits beside CPI and the Fed is in the guide to news trading.

Our read of this release is On NFP. Below, each candle is set against levels already on that day's chart, and the payroll number hardly comes up. The wider idea is News-Induced Liquidity: a scheduled release brings extra orders in at one known minute.

The first five minutes on USD/JPY, 7 August 2026

Most write-ups treat the first minute as the event. USD/JPY's biggest minute was its fifth. London had done almost nothing: every trade from the open to 08:30 stayed inside the Asian Box, 36 pips from 158.216 to 158.573. Thursday's low sat under it at 157.558.

That low was 89 pips through Thursday's low. It held for the rest of the day, and nobody could have known that at 08:34.

The 08:30 candle closed at 157.259, 59 pips off its low and still 30 under Thursday's low. The first 15-minute close back above that level, the Close Back, was the 09:45 candle at 157.739, which closed an hour and a half after the release. It did not settle much. Price reached 157.988 at 11:00, the 11:30 candle closed back under the level at 157.510, and the 12:15 and 12:30 candles traded 157.305 and 157.322. The closes stayed under the level until the 14:00 candle, then went either side of it, and the day ended at 157.761.

The first candle pointed the way: GBP/USD, 5 June 2026

On this Friday the fade never came. London had already run 47 pips over the box (1.34156 to 1.34359), reaching 1.34831 at 06:30, the high of the day and 21 pips over Thursday's high of 1.34624. The 08:15 candle closed back under that high at 1.34620, by less than a pip.

The report, 172,000 jobs, beat its forecast, as Reuters reported it. The 08:30 minute fell 22 pips from its open of 1.34615. The first five minutes reached 1.34352 and ended at 1.34396, still 22 under the open. The candle closed at 1.34310, 33 pips from top to bottom.

The 08:45 candle went through Thursday's low of 1.34112 and closed at 1.34042. No candle closed back above that low for the rest of the day, the nearest being 1.34109 at 09:00. The low came at 15:15 at 1.33298, 101 pips under the jobs candle's close. Anyone reading "the first spike always fades" was on the wrong side of the whole afternoon. EUR/USD's minutes that morning are on the guide to news trading, and the other markets on the markups of 1 June 2026.

A report that settled nothing: EUR/USD, 8 May 2026

Some releases barely register. From 03:00 to 08:30 London had covered 32 pips, 1.17416 to 1.17734. The jobs candle then covered 23.

Its first minute went 8 pips under the open of 1.17640 and 5 over it, and closed at 1.17642, where it started. The next four minutes rose, and the fifth closed at 1.17775. The candle's high of 1.17787 at 08:39 traded under a pip past Thursday's high of 1.17783, and it closed under that high at 1.17768.

The 09:30 candle closed over Thursday's high at 1.17820, and the 10:00 candle closed back under it at 1.17734, with the University of Michigan survey and the ECB's president both on our calendar at 10:00. The high of the day came at 16:45 at 1.17880. A big red or green candle at 08:30 is not a given.

Six reports on one market: did the first five minutes say anything?

Those three days were picked, because they show three different answers. So here is every jobs report on our calendar since May on EUR/USD, with nothing left out. "Range" and "Net" are those five minutes, 08:30 to the close of 08:34. "After" is from that close to the day's last candle.

The rest of the day went the first five minutes' way on three reports and against them on three. Six reports on one market is a table to read and no sample at all, and a day's close is not a trade.

The longer count says the same thing another way. On the five reports from 8 May to 4 September 2026, on seven majors and gold, we asked whether the 08:30 candle held the day's high or low, session break to session break, leaving out the spread spike on the day's first candle. It held on 12 of 40 market-days, and half of those were on one Friday. It counts candles, not trades, and 2 October is outside it. On NFP has it report by report.

What were market makers doing in those minutes?

Nobody outside can see the orders, so this is our read. Stops rest past the obvious levels, a known minute brings thin quotes and a crowd of orders, and a stop that fires becomes a market order. A large player who wants to buy needs someone selling, and a run through the stops under a low supplies that selling all at once. The Three Seats split 7 August like this.

• Retail sold the break. The 08:33 close of 157.538 was 2 pips under Thursday's low, which looks like a clean breakdown, and the next minute ran 87 pips. Anyone who sold that minute's close of 156.840 was 17 pips from the low, and the 15-minute candle closed 42 pips against them. The 20 and 50 averages on the 15-minute chart had crossed down at 07:45 and did not cross back that day, so they still said sell at 11:00's 157.988. An average is built from closes already printed, which is The Trap.
• The market maker, as we read it, had two pools in reach. The stops under the box and the stops under Thursday's low could both fill inside five minutes everyone knew about. We read those stops as funding the turn. A thin market could print the same candle with no one aiming at it.
• The observer can state the order and no more. At 08:34 this morning looked much like 5 June, when no turn came. They only differ from later candles.

How to trade NFP: what do the plans say, and what did they meet?

Guides on page one offer the same menu, and we did not test most of it. Here is what the textbook's plans met on these days, as what price did and not as how any plan scored. NFP breakout, pullback or stay out compares them properly.

• Stay flat and wait 15 to 30 minutes. It avoids the widest spread. On 7 August the 15-minute close was still under Thursday's low, with the close back more than an hour away. We have no test of what comes after the wait.
• Straddle, with orders on both sides. On 8 May the first minute traded both sides of its open.
• Trade the first move. It met a whole day's run on 5 June and a 23-pip candle on 8 May.
• Fade the first move. On 7 August the candle's low was the day's low, and the 15-minute close back came on the 09:45 candle. On 5 June price ran 101 pips past the jobs candle's close.

Three facts hold whichever plan is in a guide. Our candles are bid only, so they hide the spread, which NFP spreads and slippage takes up. A stop becomes a market order when it fires, and in a minute that travels 87 pips the fill can be far from the stop, so normal size is not normal risk. And other releases behave no better, as how to trade CPI and Fed spikes both ways show.

What do our numbers support, and what do they not?

They support two plain things. The first minutes are large and very uneven, and the jobs candle holds an end of the day on a minority of market-days.

They do not support reading a direction from the first minutes: the table splits three and three. They do not support the idea that the first spike fades: on 5 June it did not. They give no best time to wait and no target. The first three days here were chosen afterwards, so they flatter any read of them, and a reversal seen afterwards is not a setup.

The one test we ran close to any of these plans lost. A program flagged each hunt as it happened on seven dollar majors and gold, on 5-minute candles with real bid and ask, from February 2021 to December 2024, with costs of the real spread plus 0.7 to 0.8 pips a round trip ($0.15 on gold). Fading a hunt printed on a high-impact release, not only NFP, made -0.12R a trade on 1,037 trades, R being the amount risked, so about an eighth of the risk was lost on the average trade. It is one rule on one data source, and it says nothing either way about a person reading the chart. The stop hunt guide has the wider test, and the NFP stop hunt goes further into the hunt itself.

Course 2 of ForexiaPRO, The Chaos Pattern, teaches this read step by step in lesson 2.3.3, NFP, CPI, FOMC: One Read, Three Calendars.