Observation window Mar 30–Apr 5, 2026 → Aug 10–16, 2026
The system publishes its forecasts and its misses on the same page.
104 forecasts scored against what actually happened.
Forecast accuracy →
75%
across 104 scored forecasts
Direction called right
72%
up, down or flat
Confidence reliability
Good
The system claimed more certainty than it earned, so confidence has been lowered.
Evidence base →
2,783
job postings analyzed · 40 entities tracked
How a number gets here
Every conclusion is inspectable
EvidenceTrendForecastConfidenceRecommendation
This week (Aug 10–16, 2026)
Rising
This week (Aug 10–16, 2026)
Falling
Standing caveat
What this can't tell you (and why that's okay)
- We count job posts, not hires — postings move faster than actual hiring, so treat this as an early signal, not a final number.
- We only see the sources we're connected to — "not in our data" isn't the same as "not happening."
- Our forecasts are straight-line projections — they're built to miss sudden turns, and we count those misses against ourselves in the accuracy score above.