Skip to main content
Wildfire Insurance Study → Interactive

Portfolio Tail Risk Explorer

Adjust fire season severity, weather correlation, and portfolio size. Watch how correlated weather regimes amplify tail losses far beyond what independent models predict. The gap between the two curves is the risk your pricing model misses.

Season Severity (λ) 5

Average number of fires threatening the portfolio per season. Bad years average 10.

1 5 10 15
Weather Correlation 0.5

How much fire occurrence and severity share weather regimes. 0 = independent, 1 = fully correlated.

0.0 0.3 0.5 0.7 1.0
WUI Neighborhoods 30

Number of WUI neighborhoods in the insurer's portfolio.

5 30 50 100
Annual Expected Loss
--$M
P99 (1-in-100 Year)
--$M
Correlation Amplification (P99 Ratio)
--x
Loss Exceedance Curve
Correlated
Independent
Correlation Premium
Correlated vs. Independent Risk
Metric Correlated Independent Ratio
Mean -- -- --
P90 -- -- --
P99 -- -- --
Screening-level Monte Carlo. 1,000 simulated seasons per configuration. Fires per neighborhood drawn from regime-weighted Bernoulli trials. Loss per fire is lognormal (mean $8M, std $3M) scaled by regime severity. Correlated model shares a single weather regime across all neighborhoods; independent model draws regimes per-neighborhood. Actual wildfire loss distributions depend on fuel load, terrain, suppression response, and structure density.
Why This Matters

The Correlation Premium

Start with the default settings and note the gap between the two curves. The correlated model (rose) and independent model (blue) agree on average losses — correlation barely changes the mean. But look at the right tail. At the 1-in-100-year level, correlated weather regimes produce losses far exceeding the independent assumption.

Now push weather correlation to 1.0. The curves diverge dramatically in the tail. This is the fundamental problem with pricing wildfire risk using independence assumptions: the expected loss looks fine, but the tail — the loss that bankrupts you — is wildly understated. Every neighborhood burns in the same bad year.

Try increasing portfolio size to 100 neighborhoods. Under independence, diversification works: the tail shrinks relative to the mean. Under correlation, diversification is an illusion. More neighborhoods just means more exposure to the same bad weather regime. The P99 ratio climbs.

The Fidelity Lesson
An independent-fire model prices the average year correctly but misses the catastrophic year by 2-5x. The correlation premium — visible as the shaded gap between the curves — is the minimum additional capital an insurer needs to survive a regime-driven fire season.