skip to content

Backtesting and Elicitability Of Risk measures

Tuesday 9th December 2014 - 14:00 to 15:00
INI Seminar Room 1
With the suggested change to the use of expected shortfall for market risk measurement in the trading book of a bank, there has been renewed interest in the problem of backtesting expected shortfall. At the same time it has been suggested that there is a fundamental problem with backtesting expected shortfall due to its "non-elicitability" (Gneiting). Elicitable risk measures are functionals of distributions that minimise expected scores calculated using so-called consistent scoring functions; examples are Value-at-Risk and the expectile risk measure. We examine different ways in which scoring functions may be used in practical backtesting. While scoring functions offer new tools for comparing the effectiveness of banks' risk models, we concur with Acerbi and Szekely that they in no way undermine the potential use of expected shortfall as the main risk measure for the trading book.
The video for this talk should appear here if JavaScript is enabled.
If it doesn't, something may have gone wrong with our embedded player.
We'll get it fixed as soon as possible.
University of Cambridge Research Councils UK
    Clay Mathematics Institute London Mathematical Society NM Rothschild and Sons