VortexDNA Shares $77 Per Policy Improvement at NZ Society of Actuaries Conference

Christchurch, NZ October 27, 2009. Attendees at the New Zealand Society of Actuaries General Insurance Seminar last week became aware of a dramatic development in underwriting: the ability to reduce residual error by more than 7.5%, resulting in underwriting improvements of more than $77 per policy.

The case study, detailing VortexDNA's actual experience with a leading US auto insurer, was presented by Martin Burley, VortexDNA Head of Research, and Olav Jordens, VortexDNA Head of Analytics. The process for calculating the error was straightforward yet rigorous: first, a small sample of the client's customers (approximately 2,000 people) provided seed data through an online game. This seed data was then mapped to attributes like age or zip code. From these attributes, VortexDNA profiles could be inferred for the remaining population.

Importantly, although the profiles are inferred from existing attributes, VortexDNA data is entirely new and not just a rehash. The existing attributes simply provide a vehicle for the seed data to be extrapolated across the entire population.

Once the profiles had been inferred, VortexDNA regressed the genome profiles for a model set of customers against the residual error of the client's predictions. The analytics team made predictions on multiple test sets containing data on customers that were not included in the model set.

To put it in lay terms, the process allowed VortexDNA profiles to be generated for both customers and claims events, and then compared with each other to determine the likelihood of a given customer making a given claim.

The improvements were spread across four claims types: Bodily Injury, Property Damage, Collision and Comprehensive, with the largest reduction in error being posted in Property Damage: a whopping 9.36%. The test set results were independently validated by the client.

Ultimately, VortexDNA data and algorithms provide a new segmentation opportunity that easily integrates with client data to significantly improve underwriting ratios.

Click here to download the presentation in PDF format.