Brought to you by:

Reinsurance startup Kettle uses AI to price wildfire risk

Facebook Twitter LinkedIn Google

Insurtech Kettle is applying advanced technology to optimise returns on cover for catastrophic risks.

So far it has used deep learning and unique algorithms in a wildfire reinsurance product in California, and it has plans to use the same weather and satellite data in its models to predict flood patterns and risks, and wind and hurricane risks.

Kettle says its underwriting platform is “built for the modern world” and outperforms the industry with machine learning algorithms that use more than seven billion lines of weather and ground data.

“We aim to protect the world against climate change using deep learning and reinsurance,” said Kettle, which has just raised $US25 million ($34.51 million). “We use ground-breaking technology to deliver better protections for people and more stable returns for the insurance industry.”

Founded by Andrew Engler and Nathaniel Manning, Kettle is structured as a reinsurance Managing General Agent and is setting up its own risk-bearing entity. Mr Engler was formerly VP of digital at Argo Group.

In 2020, Kettle's model predicted that the fourteen largest fires, which accounted for 98% of the damage, were in the top 20% of areas most likely to burn across California's hundred plus million acres.

In 2021, Kettle's model predicted the areas consumed by the Dixie and Caldor Fires as some of the most dangerous parts of California.

Kettle says its Genesis Model & Contagion Model has an accuracy score of 89% after it divided California into 320,000 micro grids of half a square mile each. It uses these millions of grid-level, high resolution simulations to generate wildfire ignition and spreading patterns, accurately capturing the distribution of wildfire risk and property damage.

It trains “deep convolutional neural networks” on factors influencing wildfires predicts wildfire propensity. Kettle's neural networks run upward of 140 million model parameters to calculate probabilities of fire damage at the half square mile resolution across Calilfornia.

“We use the accurate results of Kettle’s wildfire simulation model to develop a pricing model and a portfolio optimization algorithm,” it said. “We use the pricing model and algorithm to create optimal portfolios of risk for our clients.”