Non-farm payrolls

NFP counts jobs added or lost, excluding farm workers and a few other categories. It is released monthly and moves FX, indices, and rates simultaneously. Three numbers come out at once:

The headline — net jobs added.
The unemployment rate — from a separate household survey. It can move opposite to payrolls because the surveys differ.
Average hourly earnings — the wage inflation component. In an inflation-focused regime, this sometimes matters more than the headline.

Why one print is nearly meaningless

Payrolls are revised heavily and the monthly noise is large. The standard professional approach is to look at the three-month and six-month averages. A single 12k print in a storm month tells you nothing; three consecutive weak prints tell you the labour market is cooling.

Initial jobless claims: the high-frequency read

Claims are weekly and noisy week-to-week, but they are the timeliest labour signal available. Never read one week — read the four-week moving average, which is what the release itself emphasises. A sustained rise in the four-week average is one of the earliest reliable signs of labour market deterioration.

The level that matters

There is a "breakeven" pace of job growth — the rate needed to hold unemployment steady given population and participation. Above it, unemployment falls; below it, unemployment rises. This level changes over time with immigration and demographics, so treat any fixed rule of thumb as dated.

How to actually trade it

Because the surprise is what matters and the reaction is fast and violent, the realistic retail approaches are: stand down during the release, or trade the aftermath once spreads have normalised. Trying to catch the first second is a spread-cost problem, not an analysis problem.

How to check this yourself

Put the last six payroll prints in a column and average them. Do the same with the last six initial-claims readings. Compare each smoothed view against the most recent headline. The raw print is noisy enough that a single month tells you almost nothing about the trend; the six-month average usually tells you what you need.