Taming the Tails: Mastering Extreme Risk with Machine Learning and Python
I remember vividly the market crash of 2008. As a junior analyst, I watched in horror as seemingly stable portfolios crumbled overnight. Traditional risk models, relying heavily on historical data and normal distribution assumptions, had failed to predict the magnitude of the downturn. This experience sparked a deep interest in understanding how to better prepare for the unexpected, the black swan events that reside in the tails of the distribution.
This is where the powerful synergy of machine learning and extreme value theory comes into play. Machine learning excels at uncovering complex patterns within vast datasets, while extreme value theory provides the tools to model and predict events that lie outside the realm of “normal” market behavior.
Table of Contents
- Data Acquisition and Preprocessing for Financial Time Series: Wrangling real-world data, handling missing values and feature engineering for optimal model performance.
- Extreme Value Theory in Action: Exploring distributions beyond the ordinary — the Generalized Pareto Distribution and techniques for threshold selection.
- Machine Learning for Tail Risk: Training models to predict extreme events using algorithms like Support Vector Machines, Random Forests and…
