JobsJudo World

Methodology

Where every number on the World Dashboard comes from, how the derived metrics are computed, and what we are not claiming.

DashboardMethodology

Latest Quarter

2026-Q1

Most recent dashboard period represented by the JSON snapshots.

Event Ledger

28,052

Rows available to the World event and panel snapshots.

Freshness

<24h

Green badges mean the backing snapshot refreshed in the last day.

Sources

Data sources

fresh<24hstale1-7dold>7d

BLS Current Employment Statistics (CES)

Monthly

Total nonfarm employment, sector employment for 12 supersectors, sector-level average hourly earnings.

Source: BLS Public Data APILast refreshed: Apr 30, 2026, 11:21 AM UTCfresh4h ago

BLS Job Openings and Labor Turnover Survey (JOLTS)

Monthly

Total + sector openings, hires, quits, layoffs/discharges. Both absolute levels and per-worker rates.

Source: BLS Public Data APILast refreshed: Apr 30, 2026, 11:21 AM UTCfresh4h ago

BLS Current Population Survey (CPS)

Monthly

Headline unemployment rate, U-6 underemployment, labor force participation, prime-age employment-population ratio, demographic unemployment cuts (race, education, age).

Source: BLS Public Data APILast refreshed: Apr 30, 2026, 11:21 AM UTCfresh4h ago

BLS Local Area Unemployment Statistics (LAUS)

Monthly (~6 week lag)

Unemployment rate, unemployment level, employment level for the top 25 U.S. metros plus Austin and Raleigh.

Source: BLS Public Data APILast refreshed: Apr 30, 2026, 11:21 AM UTCfresh4h ago

Federal Reserve Economic Data (FRED)

Daily / weekly / monthly

S&P 500, NASDAQ, Treasury yields (10Y, 2Y, 3M), Fed funds rate, CPI, GDP, jobless claims, Consumer Sentiment, OECD Composite Leading Index, Phila Fed Coincident Index, OECD Business Confidence, St. Louis Fed Financial Stress.

Source: FRED CSV endpoint (no key required)Last refreshed: Apr 30, 2026, 01:54 PM UTCfresh1h ago

Atlanta Fed Wage Growth Tracker

Monthly

Overall wage growth (3-month and 12-month moving averages), Job Switchers, Job Stayers. Derived from CPS matched-pair data.

Source: Direct xlsx pull from atlantafed.org (FRED dropped the series)Last refreshed: Apr 30, 2026, 01:54 PM UTCfresh1h ago

Indeed Hiring Lab

Daily

U.S. national job postings index (seasonally adjusted), 37 sectors, 593 metros, 51 states. Baseline 100 = February 1, 2020.

Source: Public GitHub repo (hiring-lab/job_postings_tracker)Last refreshed: Apr 30, 2026, 01:54 PM UTCfresh1h ago

Anthropic Economic Index v2

Snapshot (v2 release)

Per detailed-SOC observed_exposure score (share of Claude conversations matching that occupation’s tasks), aggregated to SOC major group.

Source: Hugging Face dataset (Anthropic/EconomicIndex)Last refreshed: Apr 30, 2026, 01:54 PM UTCfresh1h ago

BLS Occupational Employment and Wage Statistics (OEWS)

Annual (May 2023 vintage)

National employment headcount per SOC major group. Joined with the AEI exposure scores for the AI-exposure × headcount overlay.

Source: Pre-extracted in the Anthropic AEI bundle (BLS direct downloads are bot-blocked)Last refreshed: Apr 30, 2026, 01:54 PM UTCfresh1h ago

JobsJudo world events table

Continuous

Polymorphic event ledger of layoffs, WARN filings, funding rounds, hiring sprees, M&A, IPOs, AI launches, office changes, policy/civic events. Fed by a fleet of news scrapers running through Serper, ScraperAPI, and Groq Llama 3.3 70B for extraction.

Source: Internal (intelligence_world_events DB table)Last refreshed: Apr 30, 2026, 01:54 PM UTCfresh1h ago

Computation

Derived metrics

Beveridge ratio (V/U)

Computed two ways. Levels: total_job_openings divided by unemployment_level, both from BLS in raw counts. Rates: job_openings_rate divided by unemployment_rate, both from BLS as percentages. The chart uses the levels-based version, the canonical one quoted in macro coverage.

Sector wage growth YoY

Derived from CES03 sector-level average hourly earnings, comparing each month to the same month a year earlier. Computed inside fetch_bls_macro.py once the underlying series have loaded.

AI exposure × headcount overlay

Each detailed SOC’s observed_exposure from Anthropic AEI v2 maps into its SOC major group (first 2 digits of the SOC code). Mean exposure per major group is then joined with BLS OEWS May 2023 employment counts. Headline summary metrics: total US employment covered, weighted-mean exposure, and the count of workers in groups whose mean exposure crosses each threshold.

CPI year-over-year

Derived from the monthly CPI index (CPIAUCSL), comparing each month to the same month a year earlier. Quarterly average of the monthly YoY values.

Limitations

Caveats and known limitations

OEWS vintage

The employment-headcount side of the AI exposure overlay uses BLS OEWS May 2023 data, the latest available in the Anthropic AEI bundle. BLS publishes OEWS annually. Direct downloads of oesm{yr}nat.zip are blocked by Akamai bot mitigation, even with a browser User-Agent. A ScraperAPI proxy could refresh this.

Anthropic AEI exposure interpretation

observed_exposure is the share of Claude conversations matching the occupation’s tasks. It is not "share of jobs that Claude can do." A 0.38 mean exposure for Computer and Mathematical Occupations means roughly 38 percent of Claude conversations sampled matched tasks in that group, weighted across detailed SOCs. We treat it as a proxy for "how much AI is touching this occupation today," which is defensible but not the only plausible interpretation.

Indeed metro names

Indeed Hiring Lab metro labels do not perfectly match BLS LAUS metro labels. The chart-card matching in the dashboard is best-effort by substring. Where the join fails the metro is omitted from the supply-and-demand pairing.

No real PMI substitute

We tried to add the ISM Manufacturing PMI Employment subindex but ISM withdrew it from FRED in 2014. The OECD Composite Leading Index, Business Confidence, and yield-curve series are reasonable substitutes but they survey different things. None of them are direct measures of "hiring intentions."

Event extraction confidence

Event rows from tier-1 publishers (Reuters, Bloomberg, AP) carry high numerical accuracy. Press-release-only sources carry lower confidence. Each event row in the underlying database has a confidence field that downstream queries can filter on.

This page is live data

Latest quarter on dashboard: 2026-Q1. Event ledger size: 28,052 rows. Each "Last refreshed" timestamp and freshness badge above is computed from the JSON snapshot itself, not hand-edited. If a source went stale, you would see the badge flip from green to amber to red.