Can a shared market forecast the scale of wildfire destruction across California?
The Rhine example settles on a river-gauge condition. This example settles on reported damage: CAL FIRE’s annual statewide “All Structures Destroyed” count, covering homes, businesses, outbuildings and other structures. Increasing market thresholds would show how traders price the scale of that outcome.
An insurer or reinsurer could compare those prices with its own forecast of statewide destruction, or examine cover tied to the same count. One property can burn in a low-count year or survive a high-count year, so this is not a forecast of loss at an individual address.
One shared set of YES/NO contracts asks how many structures CAL FIRE will report destroyed statewide in the year. Different buyers could combine the same contracts for different payouts. The chart shows an illustrative equal-weight basket.
Use CAL FIRE’s annual All Structures Destroyed total. It includes residences, commercial buildings, outbuildings and other structures. A YES question pays when the count is at least its threshold.
Move the slider to test a count. Counts above 25,000 retain the maximum payout.
Simulated price of the equal-weight basket: $0.2575 per $1 of maximum payout. $25,750 at this limit, before fees. $65,000 hypothetical payout on a $100,000 basket.
| YES question · structures | Simulated YES price | Historical frequency | Selected outcome | Equal basket weight |
|---|---|---|---|---|
| At least 625? | $0.400 simulated | 13 of 20 years | YES | $5,000 of payout |
| At least 1,875? | $0.385 simulated | 7 of 20 years | YES | $5,000 of payout |
| At least 3,125? | $0.370 simulated | 6 of 20 years | YES | $5,000 of payout |
| At least 4,375? | $0.355 simulated | 4 of 20 years | YES | $5,000 of payout |
| At least 5,625? | $0.340 simulated | 4 of 20 years | YES | $5,000 of payout |
| At least 6,875? | $0.325 simulated | 4 of 20 years | YES | $5,000 of payout |
| At least 8,125? | $0.310 simulated | 4 of 20 years | YES | $5,000 of payout |
| At least 9,375? | $0.295 simulated | 4 of 20 years | YES | $5,000 of payout |
| At least 10,625? | $0.280 simulated | 4 of 20 years | YES | $5,000 of payout |
| At least 11,875? | $0.265 simulated | 2 of 20 years | YES | $5,000 of payout |
| At least 13,125? | $0.250 simulated | 2 of 20 years | YES | $5,000 of payout |
| At least 14,375? | $0.235 simulated | 2 of 20 years | YES | $5,000 of payout |
| At least 15,625? | $0.220 simulated | 2 of 20 years | YES | $5,000 of payout |
| At least 16,875? | $0.205 simulated | 1 of 20 years | NO | $5,000 of payout |
| At least 18,125? | $0.190 simulated | 1 of 20 years | NO | $5,000 of payout |
| At least 19,375? | $0.175 simulated | 1 of 20 years | NO | $5,000 of payout |
| At least 20,625? | $0.160 simulated | 1 of 20 years | NO | $5,000 of payout |
| At least 21,875? | $0.145 simulated | 1 of 20 years | NO | $5,000 of payout |
| At least 23,125? | $0.130 simulated | 0 of 20 years | NO | $5,000 of payout |
| At least 24,375? | $0.115 simulated | 0 of 20 years | NO | $5,000 of payout |
Historical frequencies describe twenty published years; they are not current-year probabilities. Simulated prices are independent examples, not live quotes or insurance premiums.
| Structure threshold | Hypothetical payout once reached | Years at or above |
|---|---|---|
| At least 625 | 5% | 13 of 20 |
| At least 1,875 | 10% | 7 of 20 |
| At least 3,125 | 15% | 6 of 20 |
| At least 4,375 | 20% | 4 of 20 |
| At least 5,625 | 25% | 4 of 20 |
| At least 6,875 | 30% | 4 of 20 |
| At least 8,125 | 35% | 4 of 20 |
| At least 9,375 | 40% | 4 of 20 |
| At least 10,625 | 45% | 4 of 20 |
| At least 11,875 | 50% | 2 of 20 |
| At least 13,125 | 55% | 2 of 20 |
| At least 14,375 | 60% | 2 of 20 |
| At least 15,625 | 65% | 2 of 20 |
| At least 16,875 | 70% | 1 of 20 |
| At least 18,125 | 75% | 1 of 20 |
| At least 19,375 | 80% | 1 of 20 |
| At least 20,625 | 85% | 1 of 20 |
| At least 21,875 | 90% | 1 of 20 |
| At least 23,125 | 95% | 0 of 20 |
| At least 24,375 | 100% | 0 of 20 |
The California FAIR Plan secured $750 million of wildfire reinsurance through the Golden Bear catastrophe bond. Aon reports that investors supplied $1.10 billion of single-peril California wildfire risk capacity in 2025. Swiss Re reports that many excess-of-loss reinsurance covers were triggered by the January 2025 fires. Existing wildfire risk therefore draws protection buyers, reinsurers that carry claims, and investors willing to bear losses. A reinsurer seeking to offset matching exposure could buy YES on a CAL FIRE threshold; a speculator willing to take that event risk could buy NO. Whether either would trade these proposed structure-count markets at useful depth remains untested.
Golden Bear pays on insured-loss terms. The proposed markets settle on CAL FIRE’s annual statewide structure count; bond loss estimates do not price these questions.
An insurer or reinsurer with a matching statewide parametric obligation could buy YES to offset some of its payment. A business or person exposed to wildfire disruption might also seek supplemental cash. The contract pays from the statewide reported count, so it can pay without that buyer losing property or fail to pay when that buyer does lose property.
A speculator could buy NO when the price seems to overstate the chance of severe statewide destruction. NO pays if the annual count stays below its threshold. A wildfire outcome may add a different risk to a portfolio, but the return and its correlation with other assets depend on entry prices and market conditions; neither is established for these proposed contracts.
The same public price curve could move from an independent check to a model input and, if markets become deep enough, to risk transfer. Each step needs evidence of useful forecasts and executable prices.
An insurer could compare a market price with its own statewide structure-count forecast and investigate disagreements. Prediction-market research supports testing traded forecasts in other settings; price-interpretation research shows why a price is not automatically a physical probability. The comparison could initially add work.
If prices become deep and demonstrate accuracy, an insurer could test them as an input to its aggregate destruction forecast. A Federal Reserve rate-market study illustrates the comparison method, not wildfire performance. California’s public-model program shows the much broader property-loss modeling job that a statewide count cannot replace.
With enough executable depth, a buyer of cover tied to the CAL FIRE count could buy YES contracts to offset part of that obligation. Existing insurance-linked securities show capital-market risk transfer on different covered-loss terms. Basis risk, trading costs, collateral and legal treatment would still need assessment.
Compare any avoidable forecast cost with market access, integration, monitoring, validation, fallback modeling and liquidity support. Public evidence does not yet support a saving for this exact annual-count forecast.
These values were reconstructed from a later frozen archive. They do not show what was available at the original settlement date.
| Year | Structures destroyed | YES markets | Payout share |
|---|---|---|---|
| 2006 | 431 | 0 of 20 | 0% |
| 2007 | 2,895 | 2 of 20 | 10% |
| 2008 | 992 | 1 of 20 | 5% |
| 2009 | 121 | 0 of 20 | 0% |
| 2010 | 94 | 0 of 20 | 0% |
| 2011 | 137 | 0 of 20 | 0% |
| 2012 | 248 | 0 of 20 | 0% |
| 2013 | 495 | 0 of 20 | 0% |
| 2014 | 650 | 1 of 20 | 5% |
| 2015 | 3,194 | 3 of 20 | 15% |
| 2016 | 1,274 | 1 of 20 | 5% |
| 2017 | 10,868 | 9 of 20 | 45% |
| 2018 | 22,868 | 18 of 20 | 90% |
| 2019 | 703 | 1 of 20 | 5% |
| 2020 | 11,116 | 9 of 20 | 45% |
| 2021 | 3,560 | 3 of 20 | 15% |
| 2022 | 946 | 1 of 20 | 5% |
| 2023 | 157 | 0 of 20 | 0% |
| 2024 | 1,837 | 1 of 20 | 5% |
| 2025 | 16,627 | 13 of 20 | 65% |
Twenty equal 5% steps are the fewest that meet the declared maximum 5% jump. Put each threshold at the midpoint of a 1,250-structure interval from zero to the inherited 25,000-structure reference cap. The payout stays within 2.5 percentage points of a linear count-based payout. This is a design choice, not a curve fitted to insured losses. Twenty order books still require a liquidity test.
The authoritative reading is the CAL FIRE annual Wildfire Activity Statistics report, using its published All Structures Destroyed series rather than adding incident reports. The 2025 report, PDF page 54 / printed page 49, supplies the frozen 2006–2025 history. A missing report or missing total leaves contracts unresolved, not at zero.
The proposed source cutoff is September 30, Y+1 at 23:59 UTC. Latest unambiguous official observation-year Redbook snapshot archived no later than cutoff. Use its All Structures Destroyed entry for year Y. Do not sum incident records or substitute a later year report. Preserve publication evidence and completed retrieval time. Only eligible pre-cutoff source selection, transcription and rule-application corrections during dispute window. Archive subsequent provider revisions separately; do not reopen resolved contracts. Cite the frozen report checksum, page, year and disputed calculation. Publish decisions and snapshot history. Unresolved source or definition issues remain unresolved. The venue must specify how unresolved positions are treated. These are proposed terms, not provider guarantees.