Data and models
For the Lloyd’s market, profitable underwriting depends on continually improving the tools and techniques used to select and assess risks, manage risk portfolios and optimise the claims handling process. Data is the lifeblood of the underwriting process and, with an ever-increasing range of sources available, we want to make sure we’re looking beyond our core data sets to power smart decision-making in the market.
We’d like to hear from start-ups who can find and help us tap into alternative data sets to paint a more accurate picture of risk and help the market underwrite more profitably. Ideas could range from finding new data sources that highlight high risk policies for more accurate underwriting, to predicting future claims based on search engine trends, or aggregating existing data sets in new ways to reveal hidden trends.
We are also interested in how new algorithms, models, and statistical techniques can help us to create a more bespoke experience, gain a deeper understanding of risk, and address the protection gap. Rather than just seeking new insights from new data, we are also seeking new insights from existing data.
Solutions could include:
- A proprietary or aggregated data source(s) that could be of interest to help us understand risk for existing products or develop new ones. – Data sharing platforms which would benefit a regulated market like ours.
- Economic or company data that would assist with rating and reduce the burden on customers to provide data to us.
- Ways of extracting data from legacy systems.
- Pre-competitive pricing and risk models for existing or new risks.
- Models to help us understand specific scenarios; such as extreme cyber attacks, natural hazards, liability exposures or damage to intangible assets.
- Data enrichment services
Case Study: Insurdata
Insurdata specialize in the creation and augmentation of peril-specific exposure and risk data via its Exposure Engine platform. Launched in 2017 to address the lack of accurate property-related data available to the re/insurance market, Insurdata’s platform enables re/insurers to create high-resolution, peril-specific data in real-time. By generating accurate geocode information and precise building attribute data, underwriters and portfolio managers can achieve the level of granularity necessary to capitalize fully on today’s advanced modelling capabilities.
Insurdata came into the Lab wanting to understand the state of property exposure within the market. They believe that there is a fundamental mismatch between property data and modelling resolution which has a material impact on the way risk is managed, mitigated, priced and accounted for. Over the 10-weeks Insurdata decided to focus their time on analysing syndicate property data in order to educate syndicates on their property exposure and show how accurate their Exposure Engine is compared to the currently used geocode providers.
Their mentor, Emma, helped send out an email to Heads of Exposure Management within the market, explaining what Insurdata were looking to do and asking them to send a sample of property exposure data for Insurdata to analyse. Insurdata ended up receiving data from 21 syndicates. Their analyses of 25,000 US properties across 19 syndicates found that 73% of geocodes were displaced by 10m+, 33% by 100m+ and that Annual Average Loss (AAL) for flood moved on average by 60% per location. Insurdata ran sessions with each individual syndicate to feedback their analysis and findings.