This project launched January 20th, 2017, with a webcast by MSCI researcher Carlo Acerbi.
Expected shortfall is a measure of portfolio risk. In short, it is the mean loss, given that the loss exceeds the loss at some quantile level \(q\). It is of interest because it addresses several of the shortcomings of VaR. It has been slower to adopt, though, partly because of assertions that it cannot be backtested.
However, recent research has addressed the issue of backtesting:
In this project, student researchers will implement expected shortfall backtesting. Anticipate that we will prototype in Python+pandas+numpy, then translate to C++. Details to follow....