This book combines two specific objectives: to take a relevant and current issue from the field of financial markets and theory, and to address it with state-of-the-art models from the field of computational finance.
The issue addressed concerns the observed non-normal behavior of asset returns. Recent market turmoil and related losses have increased awareness of stronger-than-expected price movements. Implications of this empirical finding are diverse and lead to impacts ranging from asset allocation and asset pricing to risk management, since to date the theory commonly assumes normal return distributions. The shift of paradigm from assuming normally distributed returns to recognizing non-normal return behavior gives rise to the development of new approaches that allow incorporation of the shifts' effects and exploit related insights.
Two new frameworks are derived from the related fields of signal processing and data mining and their interdisciplinary applications to the financial matter are developed. The concept of Gaussian mixture distributions as approximation for non-normal return densities is the fundamental concept underlying this work.
The main tasks performed are:
- The estimation of the entire non-normal return probability distribution for portfolios and single assets.
- The evaluation of the density estimates.
- Consideration and application of the superior densities for risk measure estimation and evaluation.
The concepts are first derived intuitively and theoretically. Using simple examples and real world data the methodologies are then implemented and evaluated step by step.
Zielgruppe/Target groups: Banken, Versicherungen, Portfolioanalysten, Risikoanalysten, Research Abteilungen in Finanzinstituten, Verteilungs-Forecaster
The issue addressed concerns the observed non-normal behavior of asset returns. Recent market turmoil and related losses have increased awareness of stronger-than-expected price movements. Implications of this empirical finding are diverse and lead to impacts ranging from asset allocation and asset pricing to risk management, since to date the theory commonly assumes normal return distributions. The shift of paradigm from assuming normally distributed returns to recognizing non-normal return behavior gives rise to the development of new approaches that allow incorporation of the shifts' effects and exploit related insights.
Two new frameworks are derived from the related fields of signal processing and data mining and their interdisciplinary applications to the financial matter are developed. The concept of Gaussian mixture distributions as approximation for non-normal return densities is the fundamental concept underlying this work.
The main tasks performed are:
- The estimation of the entire non-normal return probability distribution for portfolios and single assets.
- The evaluation of the density estimates.
- Consideration and application of the superior densities for risk measure estimation and evaluation.
The concepts are first derived intuitively and theoretically. Using simple examples and real world data the methodologies are then implemented and evaluated step by step.
Zielgruppe/Target groups: Banken, Versicherungen, Portfolioanalysten, Risikoanalysten, Research Abteilungen in Finanzinstituten, Verteilungs-Forecaster