The heart of insurance is the risk model, the mathematical assessment of the likelihood that a customer will make a claim and for how much. In a price competitive marketplace the need for ever greater levels of granularity and fine tuning is essential to ensure that you are not selected against. We are experts not just in building the models themselves, but also in understanding how to build up a peril structure and combine all the different elements together to form a cohesive collective.
- Better peril structures lead to greater predictive accuracy which leads to better loss ratios.
- Employing advanced modelling techniques where they are required leads to better loss ratios and lower resource needs.
- Greater confidence in your risk pricing leads to better results in your retail pricing.
- Going beyond traditional perils to head of damage level to get under the skin of claims performance.
- Splitting injury into real-world limits, separating pedestrian strikes from other injuries, and building claimants per claim predictors.
- Selecting the right interactions, and rebuilding severity model into baskets of propensity models for damage.
- Combining GLM and GBM models to cover the weaknesses of each other, and combine accuracy with consistency.
- Highly advanced model builders in both python and other software systems allows us to build best in class models.
- Proven track record of success, building risk models for motor, home, and pet for our clients.
- The first step is always to ensure that your data schema is fit for purpose and to apply feature engineering.
- The second step is to ensure your peril structure is fit for purpose so that each model is doing the right thing in the right place.
- Only then do we rebuild the models and have them all aligned to fit the final vision.