Coherent Risk Measure Using Feedfoward Neural Networks

Coherent risk measures have recently emerged as alternative measures that overcome the limitation of Value-at-Risk (VaR). In this paper, we propose a new method to estimate coherent risk measure using feedforward neural networks and an evaluation criterion to assess the accuracy of a model. Empirical results are conducted for KOSPI index daily returns from July 1997 to October 2004 and demonstrate that the proposed method is superior to the other existing methods in forecasting the conditional expectation of losses beyond the VaR.