Black Swans and Hockey Riots: Extreme Value Analysis and Generalized Extreme Value Distributions

In this article, we’ll look at:

• A brief introduction to the theory of Extreme Value Analysis (EVA) and the Generalized Extreme Value (GEV) distribution for estimating the probability of outlier events
• Applying GEV fitting and analysis to a practical time series problem using the block maxima method in R Studio
• Predicting the probability of another hockey riot in Vancouver.

Links to my other articles:

1. Random forests
2. Softmax classification
3. Climate analysis

Introduction:

Nassim Taleb’s book The Black Swan: The Impact of the Highly Improbable (2007) introduced to the general public what Actuaries, Engineers and Options Traders have secretly known about and struggled with for quite a long time. Taleb’s book talks about how outlier events with extreme impacts called ‘black swans’ actually occur more often than people think, and by corollary how people fail to adequately design against or prepare for these events. For example, natural disasters such as the 9.1 magnitude 2011 Tōhoku earthquake (which caused the Fukushima nuclear meltdown) and man-made disasters such as the 2008…