Risk Analysis: Experience vs Data


Details
How can we asses risks with small data sets?
One of the challenges in insurance is that despite having many customers, insurance companies often have only small amounts of claims data with which they can assess risks.
Markus Gesmann will present some Bayesian ideas to analyse risks with little, or even no event data.
We will touch on ideas from Daniel Kahneman and David Spiegelhalter and play around with Bayesian Belief Networks, hierarchal models and Stan.
Bio
—-
Markus Gesmann is responsible for research and modelling at Vario Partners.
Prior to Vario, he headed up the Analysis function at Lloyd's of London for eight years, where he was responsible for market-wide analytical research and development.
Before Lloyd's, he was responsible for pricing at Libero Ventures, a start-up backed by Lehman Brothers. There he developed new technology to transfer underwriting risk into capital markets.
Markus is the maintainer of the claims reserving ChainLadder package and he is the founder of the Insurance Data Science conference. He is a regular speaker at industry conferences.

Risk Analysis: Experience vs Data