Within the financial services industry, Insurance as an industry has historically been most resistant to adoption of technologies. Insurance contracts are complex and feature detailed clauses that can trigger depending on real world actions.
To illustrate this, imagine you charge Person A $10 to insure his ship. In a world where you might have sparse live information, it may be difficult or opaque to determine how you might react to changing the premium once an insurance-triggering event (catastrophe) occurs. In this sense, your response to a typhon hitting Person A’s ship is inelastic, and your loss capped at $10. You have no profit in this scenario.
When you have data, there is suddenly an ability to make more informed decisions within a smaller time frame. By having access to information, you can now offer Person B a $10 premium to insure her ship, but if that ship enters a dangerous territory (natural disasters, warzone, illegal fishing area etc) you can determine Person B has violated the terms of the insurance insurance and enact clauses to either increase your premium to $15 or decrease payout to $5. Your profit in this scenario is $5.
As this example displays, a lot of insurance contract issues can be solved with refined, streamlined, trusted data. A lot of this data exists in the real world - for instance from governmental bodies like USGS, NGOs and proprietary data carriers like WindWard, etc. The remaining steps are to create aggregated access, generate analyses to develop this data, and connect with Insurance firms. Applicants should be comfortable with text mining, data science workflows -- or be interested in learning!
The (eventual) expected outcome is to form a startup and bring this product to market.
I can bring industry experience to the table, with connections to underwriters (who write and manage insurance contracts), as well as having worked in data science at the UK market leader Lloyd’s of London. During that time, we had trials from data providers seeking to provide data around maritime risk. I know how to position this product in the market, because I have sat across the table, making decisions on data providers. For context, in 2017, Windward was unsuccessful at landing more than a trial version at the company, and their mistakes will pave the way for a better product.
Data Science (data cleaning, text mining, web scraping, analysis) or willing to learn!
Project Duration (in Months)
3 months with scope for more
Number of openings
Today, in the UK there is a launched product called City Risk Index (http://cityriskindex.lloyds.com/explore/) that designates 300 cities around the world of importance with respect to insurance, and it contains analytics regarding the likelihood of various disasters and their impact to GDP. I worked on this project in its infancy, and much of the information is based on 10 year projections and not live-updating information - resulting in a constantly decreasing usefulness over time. We aim to beat this benchmark.
What are your research interests?
Tell me about your 2 favourite projects you have worked on.