Replays · US payrolls (NFP) · 6 Mar 2026, 13:30 UK
When a weak jobs number split crypto from stocks
For the first time since the January 2021 report, the US jobs report showed a fall in payrolls. Crypto and US tech stocks read the news in opposite ways. Over the four hours that followed, every crypto market we track fell, and the Nasdaq-100 ETF rose.
A soft run, then a fall
Hiring had been slowing for months. Several first prints in late 2025 were below 100,000, and September's was just 22,000. January 2026 looked like a recovery at 130,000 (revised to 126,000 in this release). February's figure was a drop of 92,000, and unemployment rose from 4.3% to 4.4%. Two numbers in one report pointed the same way: fewer jobs and more people looking for one.
A single negative month is not a recession signal on its own. First prints are revised, and strikes or bad weather can pull one month down. But after half a year of soft numbers, a minus sign in front of the headline was the kind of figure traders had been waiting to react to, one way or the other.
Slow to start
- 1 13:30 UK: the February jobs report
- 2 14:00 UK: the decline begins
- 3 14:55 UK: the low of the first two hours, about $68,175
The first quarter hour gave little away. Bitcoin rose about 0.4% in the first five minutes and was back near the release price after fifteen. The decline started around 14:00 UK, half an hour after the release, and continued in steps: -1.5% at one hour, -1.7% at two hours and -2.6% at four. The first hour alone was 2.8 times a normal hour, #16 of 117 jobs reports in our data.
The split with stocks
US stocks did not join the selling. The S&P 500 ETF was roughly flat over its first four hours and the Nasdaq-100 ETF finished +0.6%. That is the pattern you would expect if equity investors saw weak jobs as a reason for lower rates, while crypto traders saw a weaker economy. Our data can show the split; it can't show the motive.
A week later the picture had flipped. Bitcoin was +4.2% above its price at the release and Ether +6.5%. The selling after the report turned out to be an afternoon's move rather than the start of a trend.
Sources: release time and figures from the U.S. Bureau of Labor Statistics (public domain), first-print values via ALFRED, Federal Reserve Bank of St. Louis. Crypto: Bitstamp 5-minute candles; SPY/QQQ: IEX. No consensus forecasts are shown because they are proprietary; comparisons are with the previous release. “× normal hour” compares the first-hour move with the median move in matched quiet hours. How we measure it.