Determining the right tail dependence model using R

An important problem in actuarial science and mathematical finance, as well as many other quantitative fields, is to model the statistical behavior of “extreme values” of random quantities. Consider, for example, the losses incurred in successive weeks by a certain portfolio of stocks. What can we say about the maximum weekly loss that will be incurred by a given stock in the portfolio, over a fixed time horizon? And, looking at the portfolio as a whole, what can we say about the dependence between the maximum weekly losses in the portfolio? If one stock yields an exceptionally high maximum loss over the coming year, say, should we expect the other stocks in the portfolio to co-move in a similar fashion? Or is it safe to treat the extreme losses by different stocks as independent occurrences? Clearly, questions like these are very relevant from a risk management point of view.