Good morning traders

šŸ—‚ļø On the desk today

šŸ“Š The Tape: where rates, the dollar and gold sit before 8:30.

🧭 What this week already told us, from six labour readings.

āœ‚ļø The first number isn't the real one. This year's revisions, and what gold did on the day.

šŸ“ Every payroll day of 2026, measured. Gold against the dollar.

šŸŖ™ Gold's real problem, which isn't the jobs number.

šŸ“Š The Tape

Spot gold was about $4,180 this morning.

September payrolls print at 8:30am ET today. The median forecast is 90,000 new jobs, with unemployment at 4.1%. The guesses run from 60,000 to 130,000.

For context, Fed economists now think the US needs somewhere between 15,000 and 87,000 new jobs a month just to hold unemployment steady, with immigration down so sharply. Some Fed Board staff put it under 10,000. So 90,000 is a solid forecast, not a weak one.

🧭 What this week already told us

Six readings before the big one.

→ Jobless claims, Thursday: 197,000, low.

→ Challenger, Thursday: 43,281 announced job cuts, the fewest for a September since 2022. Hiring plans the lowest for a September since 2011.

→ ADP, Wednesday: private employers added 90,000, against a Dow Jones consensus of 68,000.

→ JOLTS, Tuesday: 7.1 million openings in August, quits flat at 1.9%.

→ ISM manufacturing, Thursday: factory employment up to 52.7 from 51.2.

→ Conference Board, Tuesday: consumers saying jobs are "plentiful" beat "hard to get" by just 1.7 points, down from 4.2.

ā

šŸ’¬ The Desk Take: Nobody's firing and nobody's hiring much. Low claims, low layoffs, the weakest September hiring plans in fifteen years. That's a labour market in balance, and an awkward one for a Fed that just raised rates.

āœ‚ļø The first number isn't the real one

What the first print said, and what it became.

The number at 8:30 is a first estimate. The BLS revises every month twice, and this year the revisions have been brutal.

→ May was first reported as +172,000 against a forecast of 85,000. Two revisions later it's +63,000.

→ July was first reported as -23,000 against a forecast of 80,000. One revision later it's +21,000.

→ August came in at +162,000. Scotiabank's Derek Holt wrote that the beat was "entirely due to an abrupt shift in the seasonal adjustment factor". That's his view, and the first revision lands this morning.

The market trades the first print in full. Gold, measured by the GLD gold fund, fell 3.65% on the May beat that later mostly vanished, and rose 2.26% on the July miss that turned into a gain.

ā

šŸ’¬ The Desk Take: Read the revision line before you read the headline. If last month got cut, a "beat" this month may just be catching up.

šŸ“ Every payroll day of 2026, measured

Every payroll day of 2026. On the last six, gold and the dollar went opposite ways.

I pulled the close-to-close move for gold and the dollar index on every payroll day this year. One fell on Good Friday with markets shut, so that's eight.

→ Gold and the dollar went opposite ways on each of the last six.

→ The dollar moves more on payroll days: 0.34% on average, against 0.27% on any other day.

→ Gold barely does: 1.59% against 1.44%. Its big days this year came from somewhere else.

šŸŖ™ Gold's real problem isn't the jobs number

Gold's competition got almost a point more attractive since the record.

Gold set its record near $5,590 on 28 January. It's roughly 25% below that now.

Gold pays no interest, so its competition is what a safe bond pays after inflation, the real yield. On gold's record day the 10-year real yield was 1.90%. On Thursday it closed at 2.88%, per the US Treasury. One jobs number doesn't fix that.

The two-year at 4.78% sits well above the Fed's 3.75% to 4.00% range, so the bond market still expects more hikes, even though markets give October only about a one-in-four to one-in-three chance this morning.

ā

šŸ’¬ The Desk Take: A hot number pushes the dollar first and gold usually gives way. A cold one is gold's best chance in a while, and research going back to 2003 says a miss moves the dollar further than a beat.

ā±ļø One thing worth doing today

Write the forecast down before 8:30: 90,000 and 4.1%. Then read the revision line before you react to the headline.

If you're holding anything through the number, size it so a fill a bit worse than your stop is still a loss you planned for. The Trade Risk Calculator does that sum in about ten seconds.

If you want to see how you'd read a morning like this, the Forex quiz takes a few minutes and emails you a report.

šŸ“¬ The full version

Today's Recap on Substack has the rest: the week's closed trades with every loser on the list, and Kohl's taken apart in five numbers with a check you can run on any company. It's free to read.

If you keep your own numbers, the RB Trading Journal works the R out on every trade, 20% off with code RBT20.

If you trade gold, the spread on a morning like this matters as much as the direction. BlackBull Markets is where I get the best gold spreads I've found.

Everything in one place: start.rbtrading.site

šŸ”œ Next issue

Wednesday. What the jobs number did, and where the three gates sit for the week after it.

Education, not financial advice. Market figures are Thursday 1 October closes unless marked as this morning's: prices from Yahoo Finance, yields from the US Treasury. Some links are affiliate links, and I may earn a commission at no extra cost to you.