Risk attribution and semi-heavy tailed distributions

In this thesis we discuss the problem of risk attribution in a multifactor context using nonparametric approaches but we also introduce a new distribution for modeling returns. The risk measures considered are homogeneous since we exploit the Euler rule. Particular attention is given to the problem of attributing risk to user defined factors since the existing literature is limited when compared to other research arguments but of practical relevance. We point out the problems encountered during the analysis and present some methodologies that can be useful in practice. Each chapter combines both theoretical and practical issues.