Comparison of Performance of Traditional Value at Risk Models with Switching Model in Tehran Stock Exchange

Abstract

The problem of portfolio optimization has made many advances since Markowitz proposed an average-variance-based optimization. It can be said that the most important achievement of the Markowitz model was the introduction of variance as a risk indicator and indeed, the introduction of a quantitative benchmark into it. This research is a model for predicting value at risk. This model extends the previous methods to provide a prediction model for switching to increase the effectiveness of predictions. The switching model is explicitly designed to solve the problem with risk managers who do not trust a particular Value-At-Risk model and allows the model to calculate the value at risk in different times and conditions. In this study, predictive methods such as EWMA, historical simulation, Monte Carlo and constant variance model will be discussed. This approach is explicitly designed to predict the predictive problems of managers who do not estimate their estimates for a specific VaR model, and allows the estimated model to change over time. This approach assumes that investors at any point of time use only the historical information available to select a model, and that the choice of model is based on a pre-determined selection criterion, and then the choice of model used to predict value at a later date. The results of the research indicate that the switching model is highly desirable compared to other models over time.<br /> <br />