Risk Models
Expert risk modelling
for insurers and risk managers
Future-proof against the new reality of climate risk
Traditional risk models for insurance and risk scoring have long relied on data from past events to predict what lies ahead. This conventional approach is not enough to give you an edge against the impact of climate change on your portfolio of business. We help you model and quantify losses on catastrophic events, smaller events returning with high frequency and new perils in new territories. We call it climate tech meets insure tech.
Forward-looking models at varying temporal scales
Mitiga’s bottom-up, physics-based risk models are built on real-world, real-time factors such as vegetation, weather dynamics, and topography. This gives us a comprehensive understanding of how various hazards behave in the physical world. We then leverage AI, HPC and probabilistic modelling techniques to create high-resolution, forward-looking models that help you prepare for the future.
Insurance
Our models are new valuable tools for insurers. They help you price or underwrite policies and close protection gaps across geographies and sectors.
Scoring & ratings agencies
Our HPC capabilities let you apply computationally-intense risk models to the analysis of large portfolios.
Risk management
Probabilistic, physics-based models using stochastic catalogues: We make the most sophisticated modelling techniques in insurance available to the wider financial sector.
Look everywhere. And forward in time.
Our proprietary Earth Science AI™ solution gives you global coverage by filling data gaps for equitable distribution of climate protection benefits across the world. We call this ''global by design'' modelling.
We pioneer the use of probabilistic modelling techniques to forward-looking natural hazard models. This gives you visibility into next season and up to 100 years into the future.
Flexible levels of downscaling according to use cases. This is done to optimize highly computationally-intense calculations and ensure affordability.
High-resolution models down to 30 meters
Flexible delivery through APIs
Mitiga Fire System
Our hazard models and outputs combine ground-level data, physics-based models and machine learning to predict both ignition risks and the spreading behavior of wildfires.
Forecasting time scales range from a few days to seasonal and long-term projections, which are computed through the layering of physics-based models with climate forecasting models.
We leverage high-performance computing to enable this new high resolution approach. Our systems will simulate millions of wildfire events yearly to enhance model validity and robustness.
Starting from our research focus on Southern Europe, we have built the machine learning transfer methodology to make these models region-agnostic and relevant worldwide.
Mitiga XTreme Weather System
Our XTreme Weather models cover severe convective storms (hail and extreme wind) and other extreme events such as extreme temperatures, precipitation and drought.
We combine physical and statistical models with machine learning to simulate the occurence and intensity of SCS.
Varied temporal scales to enable a variety of applications: seasonal and annual outlooks, historical return periods, climate projections.
Flexible targeted geographies in the United States and in Europe, powered by our proprietary transfer learning techniques. This approach was validated by our US to Eastern Europe methodological framework developed for the UN.
Mitiga Volcano System
High-resolution models for both volcanic ash and lava flow hazards.
Models are developed through the combination of observation data, physics-based models and probabilistic simulation techniques powered by HPC.
Flexible temporal scales: applications range from near-present warning systems for the aviation sector to long-term risk and loss assessment for parametric insurance products such as catastrophe bonds (CAT bonds).
Mitiga has developed the world's first volcanic CAT bond in partnership with the Danish Red Cross. The underlying CAT model for the bond uses numerous data inputs to anticipate expected losses due to explosive eruptions at 10 volcanoes worldwide.
Mitigate climate risk differently
