Exercise

CVaR and risk exposure

Recall that CVaR is the expected value of loss given a minimum loss threshold. So CVaR is already in the form of a risk exposure--it is the sum (or integral) of the probability of loss in the distribution tail multiplied, by the loss amount.

To derive the 99% CVaR you'll first fit a T distribution to available crisis_losses portfolio data from 2008 - 2009, using the t.fit() method. This returns the T distribution parameters p used to find the VaR with the .ppf() method.

Next you'll compute the 99% VaR, since it's used to find the CVaR.

Finally you'll compute the 99% CVaR measure using the t.expect() method, which is the same method you used to compute CVaR for the Normal distribution in an earlier exercise.

The t distribution from scipy.stats is also available.

Instructions

100 XP
  • Find the distribution parameters p using the .fit() method applied to crisis_losses.
  • Compute VaR_99 using the fitted parameters p and the percent point function of t.
  • Compute CVaR_99 using the t.expect() method and the fitted parameters p, and display the result.