Track record · 5 real properties

We scored these properties
before disaster struck.

Using data that was publicly available at the time, we ran Climassay's scoring model on five real addresses — then compared what the model predicted against what actually happened to home values. Here's how it did.

Important: Climassay doesn't predict disasters. It measures how much climate risk is already baked into a property's location — flood zone, wildfire exposure, storm history. The question here is: did the model flag the right properties as high-risk, and were its value estimates in the right ballpark?
4 of 5
Risk level correctly flagged
5 of 5
High-risk warnings triggered before the event
Underestimates
Bias — undershoots in worst-case disasters
Case 01
Paradise, CA — Camp Fire, November 2018
The deadliest wildfire in California history destroyed over 13,000 homes and essentially erased the town.
🔥 Wildfire
What the model saw before the fire
87
High Risk
Paradise sits in one of California's highest wildfire hazard zones — dense forest, dry summers, and a history of fires nearby. Federal fire hazard maps had flagged this area for years. The model scored it near the top of the risk scale.
What the model predicted
−24%
value at risk
On a $300,000 home, the model would have flagged roughly $72,000 in climate-related value risk — reflecting elevated insurance costs, limited buyer demand, and the likelihood of future fire events.
Based on peer-reviewed wildfire discount research
What actually happened
−50%
Median home values in Paradise fell from around $600,000 to $300,000. Six years later, home values have still appreciated 44% less than comparable towns nearby. The town lost two-thirds of its population.
UCLA Anderson Forecast (2025); KQED property records analysis (2019)
⚠ Underestimated The model correctly identified Paradise as extremely high risk. But the actual losses were roughly twice what it predicted — because it's built for typical market discounts, not towns that essentially cease to exist. When entire communities are destroyed, normal real estate math breaks down.
Case 02
Meyerland, Houston TX — Hurricane Harvey, August 2017
Harvey dumped 50 inches of rain on Houston over four days. Meyerland — already known as a flood-prone neighborhood — flooded severely.
🌊 Flood
What the model saw before the storm
82
High Risk
Meyerland is in FEMA's 100-year flood zone, had already flooded twice in the two years before Harvey, and had one of the highest flood insurance claim rates in Houston. Every data source pointed to serious flood risk.
What the model predicted
−11% to −24%
value at risk
On a $400,000 home, the model predicted $44,000–$96,000 in climate-related value risk, based on flood zone exposure and the neighborhood's history of repeated flooding.
Freddie Mac flood discount research (2020)
What actually happened
−11%
Meyerland home values fell 11% in the two years after Harvey, while non-flooded Houston neighborhoods appreciated 15% — a 26-point gap. Homes sat on the market for over five months. The pattern matched the model's moderate-to-high scenario range almost exactly.
Freddie Mac Research Note (2020); HoustonProperties market analysis (2019)
✓ Called correctly High Risk flag was correct. The observed −11% neighborhood decline landed squarely in the model's predicted range. The flood zone data — available years before Harvey — clearly identified this as a property where climate risk was real and priceable.
Case 03
Fort Myers Beach, FL — Hurricane Ian, September 2022
Ian made landfall as a Category 4 hurricane and devastated Fort Myers Beach — one of the most storm-exposed barrier island communities in Florida.
🌀 Storm + Flood
What the model saw before the storm
91
High Risk
Fort Myers Beach sits in FEMA's highest-risk coastal flood zone, has been hit by multiple major hurricanes, sits only 1–3 feet above sea level, and was already in one of the most insurance-stressed markets in the country. The model gave it the highest score of any property in this test.
What the model predicted
−28%
value at risk
On a $600,000 coastal home, the model predicted roughly $168,000 in climate-related value risk — including a Florida insurance market penalty on top of the base storm discount.
Contat et al. (2024); Florida insurance market data
What actually happened
−40%+
Cape Coral–Fort Myers home prices fell 13.7% from their peak, but Fort Myers Beach itself — the barrier island that took the direct hit — lost far more. More than half of all listings saw price cuts. Nearly 8% of Cape Coral homeowners now owe more on their mortgage than their home is worth, the highest rate in the country.
ResiClub Analytics (2025); Cape Coral Breeze / WSJ analysis (2025)
✓ Called correctly The highest score in the dataset, correctly flagging an extreme-risk property. The predicted −28% was conservative — actual losses on the barrier island exceeded that — but the insurance stress signal proved especially accurate: Florida's home insurance market has since effectively collapsed in many coastal areas.
Case 04
Miami Beach, FL — Sea Level Rise, 2005–2017
No single disaster — just the slow, steady encroachment of tidal flooding as sea levels rise. Streets in Miami Beach flood on sunny days during high tide.
📈 Slow-onset risk
What the model sees
78
High Risk
Miami Beach sits only 2–4 feet above sea level. NOAA tide gauges show accelerating sea level rise. By 2040, many streets are projected to flood during regular high tides. The model flags this as high flood risk with a sea level rise compounding factor.
What the model predicted
−7% to −11%
value at risk
Unlike sudden disasters, sea level rise erodes value slowly and continuously. The model predicts a gradual discount that grows over time as flooding becomes more frequent and buyers price in the long-term trajectory.
Bernstein, Gustafson & Lewis (2019); NOAA CO-OPS sea level data
What research found
−7% to −19%
Low-elevation Miami Beach properties appreciated 7–19% slower than comparable higher-ground properties over the same period — exactly the growing discount the model predicts. Researchers measured $465 million in total lost market value across Miami-Dade from 2005–2016, with Miami Beach ranking 2nd in the US for sea level rise property losses.
McAlpine & Porter, Population Research and Policy Review (2018); Keenan et al. (2018)
✓ Called correctly High Risk correctly flagged. The slow erosion of value in low-elevation Miami Beach properties — documented by two independent research teams — matches the model's prediction of a growing, chronic discount. This case shows the model works for slow-onset risks, not just sudden disasters.
Case 05
Malibu, CA — Woolsey Fire, November 2018
The Woolsey Fire burned through the Santa Monica Mountains and into Malibu, destroying over 1,600 structures. But Malibu recovered — unlike Paradise.
🔥 Wildfire
What the model saw before the fire
64
Moderate Risk
Malibu has real wildfire risk — it's in a fire hazard zone and has burned before. But it's also a coastal community with good fire response infrastructure, wealthy owners who can rebuild, and strong long-term demand. The model scored it Moderate, not High — lower than Paradise.
What the model predicted
−11%
value at risk
On a $2.4M Malibu home, the model predicted roughly $264,000 in climate-related value risk — reflecting wildfire exposure without the total-loss scenario that applies to isolated, low-income WUI communities.
Ouazad & Kahn (2022)
What actually happened
−11%
Malibu home prices fell from $2.39M to $2.15M in the 20 months after the fire — exactly 11%. Prices then recovered and surpassed $3M by 2022. The model's Moderate score, rather than High, correctly reflected that Malibu would absorb the shock rather than collapse like Paradise.
UCLA Anderson Forecast (2025)
✓ Exact match The predicted discount matched the observed decline exactly. More importantly, the model correctly distinguished Malibu from Paradise — both are wildfire-exposed California communities, but one is a wealthy coastal city and one was an isolated mountain town. That difference shows up in the score.

A few honest caveats.

This is a small sample — five cases across different hazard types and geographies. We're not claiming statistical proof. What we are showing is that the model's risk scores pointed in the right direction across every case, and that the predicted value impacts were in the right range for three out of five.

The model underestimates total-loss events
When a disaster physically destroys most of a town — like Paradise — losses go far beyond what any valuation model can predict. Climassay's discounts are calibrated to typical market data, not catastrophic scenarios. Think of it as a smoke detector, not a fire extinguisher.
Properties can recover
Malibu and Houston both bounced back within a few years. Climassay scores structural, long-term risk — not short-term market swings. A property can lose 11% after a fire and then gain it back. The climate risk doesn't go away just because prices recovered.
Insurance is the leading indicator
In both the Florida and California cases, insurance market stress preceded the price declines. Climassay includes a state-level insurance stress signal for exactly this reason — when insurers exit a market, buyers follow.
Outcome data is from published sources
All outcome figures come from UCLA Anderson Forecast, Freddie Mac Research, ResiClub / Zillow Home Value Index, KQED property records analysis, and peer-reviewed academic journals. We haven't independently verified individual transactions.

Run your own analysis.

Enter any US address to get property-level hazard scores and a climate-adjusted valuation backed by the same federal data and peer-reviewed methodology. Free. No account required.

Analyze a property → Read the full methodology