# California wildfire: provisional shared structure-count ladder

This illustration reuses the frozen CAL FIRE 2025 Wildfire Activity Statistics snapshot and its annual **All Structures Destroyed** series for 2006–2025. The earlier 100-market definition remains archived in `definitions.json`; its thresholds and historical results are not rewritten. The source, correction cutoff, missing-report and dispute provisions in that definition remain the proposed settlement rules. Historical readings are later-archive reconstructions, not original-date settlement records.

The observation is CAL FIRE's published annual total of all structures destroyed, including residences, commercial property, outbuildings and other structures. It is neither a count of homes alone nor an insured-loss amount. Do not sum potentially overlapping incident records. A contract with threshold *t* resolves YES when the published count is **at least** *t*. If the report or total is unavailable, its result remains unresolved; a missing count is never zero. The fixed 2025 report is the source snapshot for the research history. Its PDF page 54 / printed page 49, Figure 10 supplies the values, checked against independent OCR and the overlapping 2024 report.

## Why twenty markets

The design reference is a linear payout from 0% at zero destroyed structures to 100% at the inherited 25,000-structure cap: `100 × min(count / 25,000, 1)`. This is a payout-design reference, **not a fitted relationship to insured losses**. Limit a single market to at most a 5% payout jump. Equal 5% weights therefore require at least 20 markets. Split 0–25,000 into twenty intervals of 1,250 structures and put a threshold at each midpoint: `625 + 1,250 × (i − 1)` for *i* from 1 to 20. The published integer thresholds are 625, 1,875, 3,125, 4,375, 5,625, 6,875, 8,125, 9,375, 10,625, 11,875, 13,125, 14,375, 15,625, 16,875, 18,125, 19,375, 20,625, 21,875, 23,125 and 24,375. They govern settlement.

A complete count activates exactly the thresholds it reaches. Each YES adds 5% of the maximum hypothetical payout. The staircase differs from the linear reference by at most 2.5 percentage points at any nonnegative integer count. At or above 24,375 structures all twenty YES contracts pay; the linear reference reaches 100% at 25,000. The midpoint approximation may therefore overpay the linear reference by up to 2.5 points. This payout resolution is a declared product choice; neither the cap nor the 5% step is calibrated to insured cost. An exchange must test whether twenty books can attract useful depth.

The 2025 reconstructed count is 16,627, activating thirteen contracts and a hypothetical 65% payout on the equal-weight basket. The 2018 count of 22,868 activates eighteen and pays a hypothetical 90%. These are calculated examples, not observed insurance payments or trading returns. A buyer may use different weights on the same listed questions, but that changes its basket and comparison target.

## Historical frequency overlay

At any nonnegative count *x*, historical exceedance frequency is `number of 2006–2025 annual counts ≥ x ÷ 20`. It is a descriptive share of these twenty published years. The chart plots it on the same structure-count axis as the payout, with a distinct line and label. It is **not** a modeled forecast, a bond-implied probability or a market price. The downloadable `wildfire-exceedance.csv` reports the exact share for each of the twenty contract thresholds; `wildfire-ladder-history.csv` reports each annual outcome.

A California FAIR Plan catastrophe bond provides a real example of modeled insured wildfire loss and capital-market risk transfer. Its per-occurrence, FAIR Plan covered-loss measure differs from CAL FIRE's statewide annual count. Its attachment probability, expected principal loss and spread cannot price or calibrate these structure-count contracts without a separately validated mapping. Simulated prices on the page remain editable examples, not probabilities, premiums or executable quotes.

Reproduce the JSON and CSV files with `python3 apps/peril-standard/scripts/peril_standard.py export` from the frozen input pack; run `verify` to check exact replay. The JSON records methodology, source identity, history and observed threshold frequencies. CAL FIRE source-use notices and other rights information are in `rights.md`.