Aplicação do backtesting para avaliar qual metodologia do value at risk - simulação histórica ou de variâncias-covariâncias, explica melhor o Ibovespa no período de 01/2009 a 12/2010

Abstract

The essential purpose of the present monographic work is to analyze the applicability of backtesting, in order to evaluate and appoint what methodology for Value At Risk better explains the Ibovespa variation from January 2009 through December 2010. To that effect, two methodologies of calculation were considered: the Historic Simulation, based on the histogram of data distribution and Variance-Covariance (parametric), based on volatility models, in which we highlight the Standard Deviation, Moving Averages and EWMA. We concluded that neither of both methods fit to explain Ibovespa variation in the focused period, since the backtesting was superior to the limit of 5% (five per cent), as appointed in this study. Latter, it is important to sustain that the present work has a practical importance for professionals who act at the variable income market, inasmuch it helps them to take decisions concerning the risks of such sector.