Loss Run Analysis

What it extracts

15 extraction fields

Portfolio Summary

A one-look roll-up of the whole submission - lines, carriers, periods, valuation dates, claim counts by status, and paid/reserves/incurred - so a reviewer can orient before reading claim detail.

Normalized Claims Register

Every claim from every carrier normalized into one row-per-claim register with a common set of fields, so mixed-format loss runs finally line up and can be compared.

Claims Summary by Policy Period

Loss experience per policy period - counts by status, paid, reserves, gross and net incurred, and averages - correcting the common error of summing the outstanding column as if it were incurred.

Loss Summary by Line of Business

Losses grouped by coverage line (GL, property, auto, workers' comp, professional) so an underwriter can see which line drives the account's incurred dollars.

Large and Shock Losses

The individual claims an underwriter must read - those above the large-loss threshold, and isolated shock losses well above the rest of the book - with the open reserves that could still develop.

Loss Distribution by Size Band

The severity distribution across incurred size bands, showing at a glance whether the book is frequency-driven (many small claims) or severity-driven (a few large ones).

Cause of Loss Analysis

Claims grouped by cause of loss or cause of injury with counts and incurred, revealing the perils actually driving losses across whatever line the loss run covers.

Frequency and Severity Trend

The per-period frequency and average-severity trend an underwriter uses to judge whether the risk is improving or deteriorating, noting that recent periods are immature and develop upward.

Late-Reported Claims

Claims reported well after the loss date, which delay reserving and drive incurred-but-not-reported (IBNR) uncertainty; ordered by reporting lag.

Claim Development and Reserve Adequacy

Open exposure and reserve adequacy by period - open-claim ratio and reserve per open claim - flagging books that could develop adversely after binding.

Geographic and Jurisdiction Distribution

Claims by state or jurisdiction, so a reviewer can see where the exposure sits and account for venue-driven severity in multi-state books.

Data Quality and Normalization Flags

The integrity checks an analyst runs before trusting a loss run - per-claim footing (incurred = paid + reserves), totals footing, policy-period gaps/overlaps, mixed valuation dates, duplicates, and coding gaps - each flagged Pass, Fail, or Review.

Notable and Unusual Losses

A narrative flag of claims that are atypical for the insured's line and would prompt an underwriter follow-up, from assaults and litigation to catastrophe-linked or clustered losses.

Insured Operations vs Loss Pattern

A web-researched read on the insured's actual operations versus the loss pattern, surfacing misclassification, undisclosed operations, or exposures the class code does not capture.

Underwriting Summary

The underwriter-facing synthesis that sits at the top of the file - overall experience, frequency vs severity, the large losses, reserve adequacy, data caveats, and the follow-up items to request - drawn from the columns above.

Where it fits

What it needs

What it produces

Review model

An underwriter reviews the flagged large and shock losses, the data-quality flags, and the cited claims register before the loss history is trusted for pricing.

Who uses it

Volume fit

Grounded in

• Incurred loss identity: Incurred = Paid + Outstanding Reserves (case reserves); calendar-year incurred also reflects change in IBNRverified as of 2026-07-22
• Loss ratio = (Incurred Losses + Loss Adjustment Expense) / Earned Premiumverified as of 2026-07-22
• ACORD data standards for insurance data exchangeverified as of 2026-07-22
• NCCI Experience Rating Plan - primary/excess split point (workers' compensation)verified as of 2026-07-22
• State loss-run disclosure requirementsverified as of 2026-07-22

Changelog

• July 2026