With R I do extreme value distribution fitting and CVAR , Expected Shortfall modelling under extremes, with Rmetrics, performance Analytics packages, and quantlib, 

If we could predict the behavior of stock prices in the future then we could
become very rich. Although many people have claimed to be able to predict
prices with varying degrees of accuracy, no one has yet made a completely
convincing case. Stock prices have a large element of randomness. This does not mean that we cannot model
stock prices, but it does mean that the modeling must be done in a probabilistic
sense. No doubt the reality of the situation lies somewhere between complete
predictability and perfect randomness, not least because there have been many
cases of market manipulation where large trades have moved stock prices in a
direction that was favorable to the person doing the moving.