Proceedings of 10th Asia - Pacific Business and Humanities Conference 22 - 23 February 2016, Hotel Istana, Kuala Lumpur, Malaysia ISBN: 978-1-925488-00-5 Comprehensive Comparison of Market Risk Models Tomáš Tichý, Aleš Kresta and Mehdi Toloo Abstract Value at Risk (VaR) models are important part of financial risk management and thus concern especially banks and insurance companies. While a wrong amount of capital would have negative impact on entity performance and might even lead to bankruptcy, application of an improper model could have the same implications. It is therefore natural that there have been many studies focused on market risk model performance evaluation, see eg. Berkowitz et al. (2011) or Berkowitz and O’Brien (2002). In our previous research (Kresta and Tichy, 2012), we have tested various type of models using standard approach of backtesting for risk measure at a single probability level. By contrast here, we adopt an alternative approach based on Data Envelopment Analysis to examine the efficiency of available models comprehensively for a set of probability levels. The results shows some interesting findings and relations among particular models, including recommendation about efficient estimation windows for each model. Key words: Model quality, data envelopment analysis, market risk, Value at Risk, historical simulation, NIG Track: Finance _________________________ Department of Finance, Faculty of Economics, Technical University Ostrava, Sokolská 33, 701 21 Ostrava, Czech Republic. E-mail: ales.kresta@vsb.cz.