Rare events are typically difficult to model due to the lack of historical data. In fact, the events we are typically concerned about, e.g., catastrophic events, may never have happened before, and will hopefully never ever happen. On the other hand, frequent occurrences can be easily characterized by statistical models learned from data. Inevitably, such statistical models are bound to give us a false sense of security about never-seen-before situations.
Even though we may not have any actual observations, we can still speculate and hypothesize about possible rare events, i.e., we can reason on theoretical grounds as to what could possibly go wrong.
The objective of this webinar is to present Bayesian networks as a framework to merge machine-learned knowledge from data with theoretical knowledge from domain experts in order to produce a joint probability distribution that includes common and rare events at the same time.
The case study we present is addressing some of the challenges of Modern Portfolio Theory and was inspired by Rebonato & Denev's book, Portfolio Management Under Stress.
- Presentation Slides (PDF, 32 MB)
- Optimization Network (XBL, 9.1 MB)
- Shipping Company Returns (CSV, 3.8 MB)