Time Series Modelling for Risk Analysis in Frequency Containment Reserves Market

This work investigates extreme events in the reserve market of balancing services. Extreme events have a low probability of occurrence, yet can have severe impact. Understanding the dynamics and determinants of volatility, correlation and risk in energy markets is essential. Previous investigations have investigated similar markets, yet the balancing services market is not investigated. In this study the balancing reserves in Swedish and Danish markets are analysed. When trading two energy markets, the trader needs to check the past data from these markets to establish if they are uncorrelated, or further, even independent. In this work it is observed through the correlation analysis of time series data that they Swedish and Danish reserve markets are dependent, as they have a non-neglible correlation in both the prices themselves and in the price changes in each market. The mutual coefficients of the time series are derived using linear regression. Of interest here are the residuals, which are the difference between the observed and the predicted values. They display heavy tails, and a very reminiscent of a t distribution with df=2, which leads to an interest in extreme value theory. Extreme value theory is also of interest in the interplay between these two markets since the tail behaviours of the Danish and the Swedish markets differ, and are of type Gumbel and Weibull, respectively.