Loss development monitoring every week, actual vs expected, with the claims behind every move
Adverse development seen in weeks, not at the next quarterly review.
See one case, screen by screen ↓Watching how reported losses develop between reserve reviews, by comparing actual with expected development for every segment and accident year. Loss Development Monitoring does it every week from the last review’s selected patterns, and flags an accident year at the evaluation its curve leaves its expected range — with the claims and notes behind the move.
How to choose AI solutions for insurance →
Updated
Reserves are set on quarterly triangles. After quarter-end the data is triangulated, the actuary adds the new diagonal, compares actual with expected development by segment and accident year, runs chain ladder and Bornhuetter-Ferguson, and selects. Between reviews, nobody looks at development systematically — large-loss reports go to claims leadership, not to the reserving actuary.
But the claim file keeps moving. Experts price repairs, attorneys send letters of representation, suits are served, demands land, and the case reserve follows weeks later. By the time the triangle shows it, the quarter is booked, and someone has to rebuild the story claim by claim from the notes — while an AvE total that looks fine can hide one accident year running hot behind another running cold.
Estimated split for one quarterly review of a casualty book, by hand and with the solution.
Six agents read every claim change and note each week and mark where a segment breaks; the actuary decides.
Tamsett Casualty Group watches eight commercial casualty segments every week. Contractors GL in Texas and Louisiana, accident year 2024, sat exactly on its pattern at the Q2 review. Here is what happened next, screen by screen, in the working solution.
Monday’s run went automatically at 05:40 on the August evaluation. Nadia opens the contractors GL triangle — ten accident years, $559.5M indicated ultimate — with every cell shaded by how far it sits from expected and the development curves drawn over their own ages. Five of 80 accident years across the eight segments are outside their expected range.
“18,412 paid and case reserve transactions since Mon 31 Aug, 8 segments” — and 2,906 notes, 61 large-loss reports and 44 counsel letters read.
The evaluation scrubber replays the triangle as it stood at each month-end. At the June evaluation, accident year 2024 was on its pattern: actual minus expected flat inside its band, and no segment outside its range.
“Every accident year inside its range. At the June evaluation no segment had left its pattern — this is what the Q2 review saw.”
At the July evaluation, AY 2024 reported 50.6% of its indicated ultimate against an expected range of 45.3%–50.2% — +$1.9M unexpected, A/E 3.6×. The comparator marked the break on the curve and flagged it on Mon 3 Aug, eight weeks before the Q3 quarter-end on 30 Sep.
“Left its pattern · July 2026 evaluation · 31 mo” — “flagged Mon 3 Aug — 8 weeks before quarter-end”
A cell that is not hot opens its numbers: AY 2021 at 48 months reported $33.4M. From 36 to 48 months it was expected to add +$5.4M and added +$5.0M — 63.3% reported, inside its 61.0%–65.8% range. Expected is the Bornhuetter-Ferguson emergence from the selected pattern.
“Inside its expected range — no claim moved more than $100,000 against its expected in this period.”
The hot cell opens the flag. The bridge walks from $31.9M reported at 30 Jun, through +$1.5M expected emergence and +$0.3M from the other 474 claims, to +$4.6M from twelve claims — $38.3M at 31 Aug. Below it, what it could mean for the ultimate: $68.4M at the Q2 review, $73.3M Bornhuetter-Ferguson with the unexpected added, $78.4M chain ladder on the new diagonal.
“For you to select — the agents never pick.”
Each of the twelve claims shows what happened, its case reserve at 30 Jun and 31 Aug, the signal in the notes and how many days the note led the reserve. Three expert reports, three demands, three suits served or amended, three letters of representation: twelve signals, a median 15 days ahead of the reserve.
Every signal cites its passage — the note reader never infers a signal the text does not state.
Linden Bayou Residences Owners Assn. v. Keller-Dane Builders, Harris County. The adjuster’s 16 Jul note records the plaintiff’s expert: water intrusion in 212 of the 240 units, repair estimate $6.8M plus $0.9M relocation. Defense counsel recommended an indemnity reserve of $1,700,000, and the case reserve went from $150,000 to $1,700,000 on 29 Jul.
Margin note: “Signal — the adjuster records the reserve as inadequate on the day of the report.”
Two evaluations outside its range, so the agents propose the reserve review — the choice is Nadia’s. She marks it for the reserve review and keeps her reason with the flag: construction defect expert reports and Harris County suits, not in the Q2 a priori; bring both indications to the committee.
“AY 2024 queued for the Q3 reserve committee · 15 Oct · range +$4.9M to +$10.0M”
The weekly development brief to Theo Marchetti is drafted from the run and Nadia’s decision: across the eight segments, +$45.8M reported against +$36.3M expected, half of the unexpected in one accident year. Net of the $700,000 ceded on Linden Bayou, AY 2024 is +$4.2M. Seven sources, each numbered in the text.
The brief writer never selects an ultimate or recommends a reserve, and only Nadia sends it.
AY 2024 is first on the Q3 reserve committee list: twelve claims, nine cited notes, the range for the ultimate, Nadia presenting and Theo deciding. Year-end questions wait beside it — re-selecting the contractors GL pattern from 24 to 48 months, and construction defect as its own segment.
The same accident year, seen by the five people who carry it — what their quarter looked like, and what it looks like now.
Six agents read every claim change and note each week and mark where a segment breaks; the actuary decides.
Reads every paid, case reserve and recovery transaction each night and rolls them up to cumulative reported by segment, accident year and month.
Reads adjuster notes, large-loss reports and defense counsel letters and marks each development signal — expert report, letter of representation, suit, demand, life-care plan, surgery, verdict — with its date, amount and passage.
Tracks suits served or amended, venue, counsel, demands and mediation dates for every open liability claim.
Compares expected with actual development for every segment and accident year at each evaluation, draws the expected range from 80% of the past ten years’ variation, and marks the evaluation where a curve leaves it.
Matches every moving claim to its treaty layer, shows the unexpected gross, ceded and net, and lists claims past the notice line.
Drafts the weekly development brief and the evidence for each reserve review item, with Bornhuetter-Ferguson and chain ladder indications side by side.
Decides every flag. The agents propose; a named person decides.
The actuary can ask in plain words — or say where a flag should go. Every answer carries its source.
Why is contractors GL accident year 2024 flagged?
Because it reported +$6.4M in July and August against +$1.5M expected from the Q2 pattern — +$4.9M unexpected, A/E 4.3×. Twelve claims moved +$4.6M of it. The largest is Linden Bayou Residences v. Keller-Dane: the plaintiff’s expert priced repairs at $6.8M plus $0.9M relocation, and the case reserve went from $150,000 to $1,700,000 thirteen days later.
When did AY 2024 leave its pattern?
At the July evaluation (31 months): 50.6% reported against a range of 45.3%–50.2%. At the Q2 review on 30 Jun it sat exactly on its pattern, 46.6%. It was flagged on Mon 3 Aug, eight weeks before quarter-end.
How much of the unexpected is ceded?
$700,000 on contractors GL: Linden Bayou at $1.7M is above the $1M retention of the 2024 casualty excess of loss, so AY 2024 is +$4.2M net. On umbrella AY 2022 the Tovar verdict is 75% ceded under the quota share: $1.5M of $2.0M. Five more claims are past the 50% notice line, with notices sent.
Send AY 2024 to the Q3 reserve committee with the range
Done. Contractors GL AY 2024 is marked for the reserve review and is first on the Q3 committee list for Thu 15 Oct, with both indications: $73.3M Bornhuetter-Ferguson and $78.4M chain ladder against $68.4M at Q2. The selection stays with you and Theo.
13 screens from the working solution, on its sample data. Pick one to see it large.
The range, the triggers and the schedule are settings — every change saved as a new version, applied from the next run.
| Setting | Default | Choose from |
|---|---|---|
| Range around the selected pattern | 80% of past variation | 70% · 80% · 90% of past variation, from the last ten years at each age |
| Flag when the unexpected is over | $250,000 | $100,000 to $1,000,000, in $50,000 steps — smaller moves outside the range are listed, not flagged |
| Propose “for the reserve review” after | 2 evaluations | 1 · 2 · 3 consecutive monthly evaluations outside the range |
| Patterns in use | Q2 2026 review | The selected patterns as signed by the Chief Actuary |
| Claim transactions | Every night | Every hour · every night · every week |
| Full monitor run and brief | Mondays 05:00 | Mondays 05:00 · daily 05:00 · Fridays 17:00 |
| Reinsurance notice line | 50% of retention | 40% · 50% · 60% of retention |
| Signals read in the notes | 8 signals | Each switched on or off — from expert report to closure below reserve |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample casualty book · “estimated” = our estimate · “typical” = published figures (S&P Global Market Intelligence via Carrier Management / Insurance Journal, Apr 2026 — $3.9B adverse development on AY 2021–2023 other liability occurrence in 2025; CAS and Triple-I, legal system abuse, 2025 — 27.4%–34.0% of other liability occurrence losses). People, companies and claims named on this page are fictional — characters and sample data in the working solution.
Watching how reported losses develop between reserve reviews, by comparing actual with expected development for every segment and accident year. Loss Development Monitoring does it every week from the last review’s selected patterns, and flags an accident year at the evaluation its curve leaves its expected range — with the claims and notes behind the move.
Expected is the Bornhuetter-Ferguson emergence from the signed selected pattern and indicated ultimate: unreported at the start times the share of the pattern due in the period. The expected range is drawn at each age from 80% of the past ten years’ variation, and both are settings.
No. The agents are read-only on claims, reserves, patterns and indications. They show a range of indications — Bornhuetter-Ferguson and chain ladder side by side — but never select an ultimate or recommend a reserve.
It attributes the unexpected to the claims that moved and reads their adjuster notes, large-loss reports and counsel letters for development signals — expert reports, letters of representation, suits, demands, surgeries, verdicts. Each signal links to its passage, with how many days it came before the case reserve change.
Yes. Every moving claim is matched to its treaty layer, so each flag is shown gross, ceded and net, and claims past the notice line are listed for the reinsurance accountant, who sends the notices.
Yes. The reserving actuary marks every flag watched, explained or for the reserve review, with the reason kept in the audit trail. The agents can propose the reserve review after two evaluations outside the range; the Chief Actuary decides what the committee books.
Yes. The range around the pattern, the dollar threshold for a flag, how many evaluations before the reserve review is proposed, the run schedule, the reinsurance notice line and the signals read in the notes are all settings, versioned with who and when.
The Agentic Solution Engine builds and deploys it from your requirements — your segment definitions, last review’s selected patterns, treaty structures and a feed from your claim system — and it goes live once every quality gate has passed. We will walk you through it on your own triangles first.
An auditor's question answered in minutes, with the original support attached.
1-min demoPremium & Claims ReconciliationBreaks found and explained on day one of close, each tied to its source records.
1-min demoStatutory, GAAP & IFRS 17 ReconciliationOne explained walk from statutory to GAAP or LDTI to IFRS 17 results.
1-min demoWe’ll run Loss Development Monitoring on your own segments and last review’s patterns.