Methodology
Short, honest notes on provenance, measured accuracy, and the limits of the extract.
What this is
SerffSurf is rate-filing intelligence built from public SERFF filings from 2020 onward: workers' compensation across NCCI and independent-bureau states, with other lines being added.
It is a market-intelligence terminal for pricing teams — not a filing repository and not actuarial advice.
How accurate it is
Each data set is scored on a holdout labelled blind against the source documents; the newest per set as of release r6.6 (Oct 12, 2026), holdouts 10 and 13:
| Data | Precision | Lower bound |
|---|---|---|
| Company LCMsholdout 10 | 99.2%124 / 125 | 96.3% |
| Deviationsholdout 10 | 95.0%95 / 100 | 89.8% |
| Class valuesholdout 10 | 100%41 / 41 | 93.0% |
| Expense provisionsform reader · holdout 10 | 94.9%600 / 632 | 93.3% |
| Expense totals and PLRsform reader · holdout 13 | 94.4%85 / 90 | 88.7% |
| Expense componentsAI-read, filings of ≤ 5 companies · holdout 13 | 98.3%288 / 293 per filing97.9% · 424 / 433 per company row | 96.4% |
Independent blind labelling against the source documents, adjudication of every disagreement, exact one-sided 95% lower bounds.
The live table, with the database's scale and freshness by state, is on the data page (access code required).
How numbers get here
Filings: public SERFF rate filings from 2020 onward, checked for new filings and letters every 3 hours; new filings reach the site with the weekly release. Each filing's main document is parsed into its filing fields, company tables, section text and correspondence letters.
Companies: names map to NAIC company codes and groups where possible; rate rows join through the resolved company.
Extraction: LCM and deviation values are read only from the filing's own text, and every accepted value carries a verbatim quote that appears in that text.
Range checks: LCMs outside 0.4–6.0 and deviations outside −90% to +300% are rejected and never shown. Values inside those bounds but outside typical actuarial ranges are shown and flagged for expert review.
Comparing LCMs across states
A loss cost multiplier is applied to the loss-cost basis that state uses. NCCI advisory loss costs and each independent bureau's pure premiums sit on different bases, so LCM levels are not apples-to-apples across those regimes — a higher median LCM in an NCCI state than in an independent-bureau state can be a basis difference, not proof that the NCCI state is more expensive.
Within one bureau or one state, LCM levels and trends are comparable. Prefer within-regime or within-state reads when ranking carriers or markets.
What to trust — and verify
Evidence quotes are the credibility layer. Prefer them over bare numbers when a figure will drive a decision.
Company-table rate rows are generally more reliable than filing-overall figures repeated across affiliates.
Table-heavy exhibits can yield values that quote the page correctly but sit under the wrong label; the holdout above measures how often that gets through.
Always verify material figures against the underlying SERFF filing.
What the site serves
Product pages, the API and the MCP tools read the same published release; API and MCP responses name it. Insights and aggregates only: no bulk export, and aggregate-tier codes never receive carrier-identified rows.