Catastrophe Modeling (2026/2027): Moody’s RMS vs. Verisk Touchstone, Core Modules & PML Interpretation

Updated Nov 2, 2026 — refreshed vendor lineage matrix, core metric architecture, and secondary perils discipline.

Climate load: model updates. Treaties: QS / XL. Related: Physical hazard repricing.

Vendor Comparison & Lineage Matrix (Moody’s RMS vs. Verisk Touchstone vs. Cotality)

In commercial property and reinsurance markets, three primary model lineages dominate portfolio analytics. As verified on regulatory model evaluation standards (including the Florida Commission on Hurricane Loss Projection Methodology accepted models list, Rev. October 2026), current operational standards stand as follows:

Platform & Parent	Verified Production Version	Simulation Engine & Architectural Strengths	Market Dominance & Adoption
Moody’s (formerly RMS)	Risk Modeler (2.55.0) / North Atlantic Hurricane V25.0	High Definition (HD) simulation architecture; explicit climate condition catalogs; sophisticated hydrodynamic coastal surge modeling.	Global reinsurance benchmark, Lloyd’s syndicates, European treaty panels, and Bermuda retrocessional markets.
Verisk (formerly AIR Worldwide)	Touchstone 2025 / U.S. Hurricane Model V3.0.0	Component-based engineering vulnerability functions; seamless integration with ISO property rating classes and commercial building envelope data.	Dominant among U.S. commercial admitted carriers, state regulatory rate filings, and North American insurance-linked securities (ILS).
Cotality (March 2025 rebrand of CoreLogic)	Risk Analytics Suite / Standalone Peril Catalog	Granular parcel-level hazard scoring, structural replacement cost integration, and dynamic wildfire perimeter and fuel-loading databases.	Mortgage origination risk, banking compliance, primary personal lines underwriting, and commercial real estate portfolio screening.

What catastrophe modeling does

Cat models simulate thousands of plausible years built from stochastic event catalogs, not a replay of last season’s headlines. Each synthetic year produces insured loss after physics-based or statistical hazard, damage functions, and policy terms. The output is a distribution: average year, bad year, and tail years that drive reinsurance and capital.

That distribution feeds underwriting, pricing, accumulation control, and regulatory filing narratives. It also feeds the reinsurance treaty conversation—what you retain, what you cede, and at which loss threshold.

The four modules (and where data goes wrong)

Hazard

Earthquake ruptures, hurricane tracks, hail footprints, wildfire spread—each peril has its own event set and correlation rules. Climate-sensitive perils now get explicit frequency and severity review in vendor releases; treat vendor release notes as directional, not gospel. Carrier loading of those shifts shows up in renewal pricing and attachment points long before the next model release.

Vulnerability

Damage ratios tie construction, occupancy, year built, and mitigation features to mean loss and spread. A wood-frame coastal dwelling and a reinforced concrete high-rise do not share one curve. Field verification still beats assumed ISO construction codes.

Exposure

The model is only as clean as the geocoded book: limits, deductibles, coinsurance, business interruption, and secondary modifiers. Bad geocodes and stale valuations dominate many portfolio errors more than hazard disagreement.

Financial

Policy forms, sublimits, facultative certificates, and treaty attachments turn ground-up damage into retained loss. This is where PML management and accumulation control meet accounting reality.

Core Metric Architecture: How to Read AAL, OEP, AEP, and TVaR

Technical catastrophe risk quantification relies on four primary statistical outputs. Confusing these metrics during underwriting or board presentations leads to severe capital misallocations:

Average Annual Loss (AAL)

AAL is the expected value of loss per year, calculated by averaging modeled losses across all simulated years in the stochastic catalog. It represents the “pure risk premium” required to cover catastrophe losses over the long run, excluding capital charges, operational expenses, and profit margins. Underwriters track AAL drift across underwriting quarters to ensure premium volume stays ahead of exposure inflation.

Occurrence Exceedance Probability (OEP)

Occurrence exceedance probability (OEP) represents the probability that the single largest catastrophic event in a given year will exceed a specified loss threshold. OEP curves govern per-occurrence excess of loss treaty design: if a carrier’s 1-in-250 year OEP loss is $250 million and its net capital retention is $50 million, it must secure a $200 million per-occurrence reinsurance limit.

Aggregate Exceedance Probability (AEP)

Aggregate exceedance probability (AEP) measures the probability that the sum of all catastrophic losses across all events throughout an entire policy year will exceed a threshold. While OEP evaluates peak severity, AEP captures event frequency. Carriers operating in convective storm corridors or active hurricane basins rely on AEP curves to structure aggregate stop-loss treaties that protect against cumulative capital depletion.

Probable Maximum Loss (PML)

Probable maximum loss (PML) is a return-period snapshot (often 1-in-100 or 1-in-250 year) extracted from either the OEP or AEP curve. PML definitions vary widely—ranging from ground-up damage to gross loss and net retained loss after reinsurance. Clear contract wording must always specify whether a quoted PML is gross or net of treaty inuring.

Tail Value at Risk (TVaR / Tail-VaR)

Tail value at risk (TVaR) measures the expected loss conditional on the loss exceeding a specified threshold (e.g., TVaR 99.5% corresponds to the conditional average of all losses beyond the 1-in-200 year threshold). Reinsurers, rating agencies, and ILS investors mandate TVaR metrics because standard PML figures obscure the steepness of catastrophic tail severity.

Secondary Perils (SCS, Hail, Wildfire) & Statement of Values (SOV) Scrutiny

In modern commercial property underwriting, catastrophe modeling has expanded far beyond headline tropical cyclones and earthquakes. The surge in severe convective storms (SCS), straight-line wind, large hail, and wildland-urban interface fires has fundamentally transformed portfolio risk evaluation:

• Statement of Values (SOV) Hygiene: Commercial underwriters reject schedules that rely on outdated building valuations. Carriers enforce strict coinsurance penalties or margin clauses when reported values fail to reflect post-inflation replacement costs.
• Granular Secondary Modifiers: Modern models require exact secondary engineering modifiers: roof covering material (e.g., UL 2218 Class 4 impact resistance), roof age (>10-year roofs face aggressive depreciation schedules), wall construction, roof-to-wall anchoring, and window protection.
• Percentage Deductibles & Cosmetic Exclusions: In hail-prone corridors, flat $25,000 or $50,000 deductibles have been widely replaced by 1% to 2% calendar-year wind/hail percentage deductibles, paired with cosmetic damage waivers for architectural metal surfaces. For in-depth peril breakdown, see Secondary Perils in Cat Modeling: SCS, Flood, and Wildfire.
• Commercial Coverage Endorsements: For structural gaps and policy limitations, review Commercial Property Insurance Exclusions, Gaps & Endorsements.

How carriers and reinsurers actually use vendor output

Primary insurers run vendor models for rate filings, portfolio monitoring, and facultative submissions. Large carriers blend multiple vendors, apply custom vulnerability or hazard overlays, and stress-test against recent industry loss years. None of that appears on your policy jacket.

Reinsurers rebuild the loss distribution with their own exposure ingestion, treaty modeling, and historical loss experience. A cedent’s RMS print and a reinsurer’s internal view routinely diverge on the same zip codes. Negotiation happens on net loss curves, reinstatements, and hours clauses—not on whose logo is on the PDF.

State regulators increasingly expect documentation when filings lean on cat models. The NAIC maintains a catastrophe models (property) resource page and trains regulators through its Catastrophe Risk Management Center of Excellence. In 2026 the Pre-Disaster Mitigation and Risk Modeling Working Group advanced a draft Residential Mitigation Grant Program Model Act (comment period closed September 8, 2026)—a voluntary framework for state DOIs to stand up retrofit grant programs tied to recognized mitigation standards and premium discount expectations.

Cat bonds, sidecars, and modeled triggers

Insurance-linked securities price off modeled loss indexes and contractual triggers as much as off historical experience. H1 2026 set a first-half issuance record of almost $18 billion; Artemis tracked about $17.98 billion across 83 transactions, above the prior H1 high near $17.6 billion in 2025. Outstanding cat bond market size approached $65.6 billion at mid-year by the same market data. Renewal pricing softened at the margin as maturing capital recycled, but tail peril demand remains.

Modeled loss triggers tie directly to the same hazard and vulnerability assumptions discussed above—see parametric insurance and index-based risk transfer for how index structures differ from indemnity cat covers.

Flood: where NFIP pricing diverges from commercial cat models

Commercial property cat models cover several perils; NFIP flood is a separate public program with its own engine. FEMA’s Risk Rating 2.0 has priced new and renewing NFIP policies on property-specific features since full implementation in April 2023. Annual premium increase caps still apply (18% for primary residences, 25% for other policies under current statutory limits). FEMA published NFIP manual updates effective December 1, 2026 covering application timing, provisional rating notices, and renewal documentation—operational detail for agents, but a signal that flood risk granularity keeps tightening.

Do not assume your carrier’s hurricane model output substitutes for an NFIP elevation certificate or private flood quote. Cross-check both when buying or renovating in coastal and riverine counties.

Practical workflow for owners and risk managers

Start with property risk assessment: document construction, roof, openings, elevation, and mitigation credits the model may never see unless you submit them. Align values with replacement cost, not stale market estimates. For large schedules, ask your broker which vendor and version powered the last reinsurance submission and whether net or gross PML drove the retention change.

During hard markets, carriers stress secondary perils (severe convective storm, wildfire) and tighten capacity before headline hurricane seasons. Modeled AAL moves show up as rate, deductible, and sublimit changes long before a named storm forms.

Frequently asked questions

What is catastrophe modeling in plain terms?

Catastrophe modeling is a computer-driven way to estimate how much insured loss a book of properties could take from natural disasters across many simulated years. It combines hazard science, building damage estimates, your policy details, and reinsurance to produce loss probabilities—not a prediction of next Tuesday’s storm.

What is the difference between OEP and AEP?

OEP measures the probability that total loss from a single occurrence (or event definition in the contract) exceeds a dollar amount within a period. AEP measures the probability that all qualifying events in the year combine to exceed that amount. Reinsurance hours clauses and occurrence caps make the distinction material to net loss.

Can I rely on one cat model vendor number for a decision?

No. Vendors disagree, models update, and reinsurers re-cut exposure. Use vendor output as structured input, then compare scenarios, recent industry loss experience, and your own inspection data. A single exported PML is a data point, not a verdict.

Does catastrophe modeling set my homeowners premium directly?

Usually indirectly. Personal lines carriers embed cat load in territory rates, reinsurance cost, and underwriting rules. You see the result as rate, wind/hail deductible, or non-renewal. Commercial schedules may reference modeled loss metrics in renewal terms explicitly.

Are catastrophe models the same as FEMA flood maps?

No. FEMA flood maps and Risk Rating 2.0 serve the NFIP and community floodplain management. Commercial cat models simulate multiple perils for insurer portfolios. A property can look acceptable in one system and stressed in another; use both for flood-prone sites.

What does PML mean in a reinsurance discussion?

PML is a high return-period loss estimate after policy terms and often after treaty coverage. The exact return period, net/gross definition, and peril scope must match the slip language. Mismatch between cedent and reinsurer PML definitions is a common source of capacity disputes.

Catastrophe model outputs represent stochastic approximations and probabilistic distributions. They do not constitute deterministic loss guarantees or binding underwriting terms. Consult certified catastrophe risk analysts and licensed brokers for portfolio modeling.