Non-Gaussian time series and nonlinear dependence in finance markets

Two key statistical features of finance markets are non-linearity and Non-Gaussianity. I was awarded funding for the project "non-Guassian time series and nonlinear dependence" from Finansmarkedsfondet for 2011-2012, mainly for coordinating research betwe en several international research groups by arranging visiting activity, including a visit by our PhD student Geir Berentsen at the Department of Econometrics and Business Statistics at the Monash University, Melbourne. The activity was spread over 3 sub -projects, namely i) Local Gaussian correlation, ii) Nonlinear cointegration and iii) Integer-valued time series. In the referee report for that project it was pointed out that i) is most relevant for financial markets. In the new project I will focus mu ch more strongly on this, and I am asking for funding of a post doctoral fellow over a two-year period mid-2013 to mid-2015. Topics from i) that will receive attention will be multivariate heavy tail dependence, multivariate extreme events, and multivaria te portfolio analysis and risk. In addition we would like to consider nonlinear cointegration theory using local Gaussian correlation, this combining i) and ii). Finally we would like to explore the relationships to multiple copula theory, in particular t he vine theory. I believe all of this will be relevant for the statistical analysis of financial markets.