Analysis of extremes in Financial Time Series

Specialeforsvar: Isse Hassan Mohamud

Titel: Analysis of extremes in Financial Time Series: Investigating the interplay between heavy tails and dependence

Abstract: This thesis will look into time series analysis, focusing particularly on stationarity and log-returns, using models such as autoregressive moving average (ARMA), autoregressive conditionally heteroscedastic (ARCH), and generalized autoregressive conditional heteroskedasticity (GARCH). The thesis argues that to effectively handle data in real-life financial time series analysis, it is essential to adapt or modify the ARMA, ARCH, and GARCH models. Traditional versions of these models assume stationarity,
which can limit their applicability to real-life data that are often not stationary. The thesis will therefore explore the limitations of these models in handling dependencies and heavy tails in time series data, and presenting improvements to address these challenges. Through analysis of data from companies like Novo Nordisk and A.P. Moller Maersk, the thesis showcases the effectiveness of these models in capturing dependence structures in financial time series when developed. 

Vejleder: Thomas Mikosch
Censor:    Anders Rønn-Nielsen, CBS