For cat-bond issuers + sponsors

Audit-defensible calibration for OM disclosures.

Catastrophe-bond offering memoranda disclose modeled probabilities of attachment and expected loss. Today these probabilities are sourced from RMS, Verisk AIR, or Munich Re NATCAT — none of which publish pre-registered probabilistic forecasts with subsequent Brier scoring. Investors take the modeled numbers on faith. As ILS allocators professionalize, the absence of an audit-defensible calibration credential becomes marketing friction during placement.

See live ratings Read methodology v1.0

What Forecast Registry provides for the offering process. A neutral, third-party credential supporting the modeled probabilities in your offering memorandum. SHA-locked at lock, OTS-anchored on Bitcoin before resolution, scored under the standardized methodology. Citable in the OM as an external validation reference.

Use cases at placement

OM external-validation citation

Reference the registry credential in the modeled-loss section of the offering memorandum. Demonstrates independent calibration reference beyond the structuring model vendor.

Investor roadshow defensibility

Present the registry's per-peril leaderboard to ILS allocators during pre-placement diligence. Reduces "your model vendor said so" friction.

Spread-narrowing argument

An independent calibration credential is grounds for arguing tighter coupon spreads on perils where the credential's track record demonstrates the structuring model's risk assumptions are consistent with publicly-verifiable forecast accuracy.

Trigger-mechanism defensibility

For parametric or industry-index triggers, cite the registry's resolution-source linkages (IBTrACS, PERILS, CAL FIRE, NIFC, USGS) directly in the trigger definition. Eliminates basis-risk arguments built on disputed resolution sources.

What we are asking

One reviewable conversation with structuring counsel on whether the registry credential can be cited in your next OM.
Participation in the v1.0 public comment — cat-bond market practitioner perspective on trigger-definition mappings.
Optional: contribution of historical (anonymized, SHA-locked) modeled probabilities for retrodictive scoring against IBTrACS / PERILS. Your structuring vendor would be the obvious contributor; you would be the sponsor.

Vendor disclosure framework

Methodology §4.1 documents the publicly-citable out-of-sample miss patterns of the three dominant vendor cat models (RMS / Verisk AIR / Munich Re NATCAT). Methodology §4.2 documents the retrodictive proof of concept on Atlantic hurricane (n=20 IBTrACS-scored records). The quantitative head-to-head with vendor models requires regulator-compelled disclosure under NAIC Model Bulletin 30 §3(D); cat-bond issuers benefit from this regulatory engagement because the vendor-disclosure regime, once in place, also serves OM-disclosure standards.

Reference forecaster calibration

Founding contributor's reference forecaster: 1,963 SHA-locked forecasts, 82 resolved, Brier 0.0399 on n=18 HC in-headline forward. Atlantic basin retrodictive backfill: n=20, mean Brier 0.2002, BSS +0.199 vs climatology. Per-record breakdown →

Contact

Addie Conner, founding contributor. hello@forecast-registry.org (provisional).

Last reviewed: 2026-06-05.