Just want to read it?
No download needed. Every trade has a plain-English page with the wages, margins, and benchmarks laid out in tables.
Benchmarks at a glance
The Level CFO operating benchmarks in the dataset, on screen. Each row is a real distribution across the blended contractor pool (n disclosed), not a single average. Full sources are in level-benchmarks.json.
Machine-readable files
For analysts and AI assistants. Plain JSON, versioned with the 2026 edition.
Public-company prices, market caps, returns, margins, and valuation multiples are a historical reference dated 2026-07-23, not live quotes or current valuation comparisons. All 69 company records remain available with their fundamentals periods, while current_comparison_eligible: false keeps crawlers and analysts from treating them as current.
Building-stock provenance
The building-stock file aggregates public state and county tax-assessor and property records. Level analyzed an aggregated compilation of those public records and does not claim that it downloaded every record directly from each government website. The source was a stacked weekly snapshot history, deduplicated to one record per parcel before aggregation.
The file covers 50 states plus the District of Columbia by construction-era band and commercial or industrial use group. Its properties are assessor parcels, not individual buildings, and 39.4% have unknown construction era. It is therefore separate from Level CFO's FEMA/ORNL structure-footprint dataset, which counts physical serviceable buildings inside state and metro polygons.
Core files
Per-trade files
One full data object per trade: scores, sub-scores, ~40 research topics (labor, market structure, unit economics, cash cycle, AI exposure, channel, lock-in, entry barriers, M&A, demand) with citations, research-pass counts, and confidence tiers, geography (all 50 states + top-50 metros), market structure, unit economics by revenue band (under $1M / $1M to $5M / $5M to $20M / over $20M), reputation, firmographics, Level CFO benchmarks, and public-company comps. Path pattern: /data/trades/<slug>.json
Open data & license
Trade Economy Index original analysis, scores, and compilation are published under CC BY 4.0. Cited third-party material remains subject to its original source's rights and terms. Aggregates never identify an individual company.
Required attribution:
Source: TradesIndex by Level
When republishing the index's original analysis or compilation online, include the source line above with a clickable link to https://tradesindex.org. Quoting specific wording is not required.
Cite this dataset:
Yang, S. (2026). The Trade Economy Index, 2026 Edition [Data set]. Level CFO. https://doi.org/10.5281/zenodo.21762674. Trade AI Score (TAS) = composite of AI-Resilience and AI-Leverage.
The dataset carries a permanent DOI, 10.5281/zenodo.21762674, and the same tables are mirrored one row per observation on Hugging Face and GitHub (CSV and JSON, with citations, research-pass counts, and confidence metadata retained where available). The mirrors carry the same tables and version. Cite the DOI if you need a reference that survives a move.
Level CFO aggregate operating benchmarks are published under CC BY 4.0 (last updated 2026-Q3). Public-source figures retain their source citation; some citations are text-only rather than clickable URLs. Machine agents: see /llms.txt for a structured guide.
Machine agents: see /llms.txt for a structured guide to this dataset.