markov switching model

Value at Risk model based on a switching regime approach was used in this study to optimize portfolios consisting of industry index (petroleum products, investment, chemical products, and metal products). For this purpose, the VaR of returns on index should first be extracted through parametric models of the (GARCH) family in each of the above industries by using regime transitions. After the risk of return on index is obtained for each industry, the optimal portfolio is created in the next step based on VaR minimization, and the optimal value of each industry is determined in the portfolio. According to the results, (MRS-FIEGARCH) model had no superiority in VaR estimation over the other parametric models of the GARCH family. In fact (MS-EGARCH-t) was introduced as the optimal model. Among the designated industries, returns on indices followed regime transitions only in chemical products and investment by showing asymmetric reactions to external shocks. Moreover, the optimal weights were on the rise in the industries where VaR decreased over time, whereas the optimal weight of the portfolio decreased in the industries where VaR increased over time. The higher share of an optimal portfolio belonged to the industries where stock returns had lower rates of VaR. The risk-return-ratio was employed to show that the optimal portfolio with a risk rate was measured by considering the switching regime was superior over the optimal portfolio with a risk rate extracted without considering the switching effects. To create an optimal portfolio, it is then recommended to make investments in the industries characterized by higher stability in prices and lower fluctuations in stock returns in the long run. This approach can be employed to obtain the best results from optimal portfolio preparation in the worst-case scenario of the market fluctuations.

The stock exchange is considered to be an important establishment to finance long term projects, on one hand, and to collect savings and finance of private section. The stock exchange can be a safe and secure place to invest surplus funds to purchase corporate stocks. As recession and prosperity in this market can have a great role in stockholders` decision-making, it becomes vital to predict these cycles. In this paper, using model MSMH(4)AR(2), we extract the financial cycles of the market. Then, using the ant colony algorithm, we determine the most significant predictors and predict the market financial cycles using neural networks. The results show that the PNN model performs better in predicting the future market with respect to the criteria of mean squared error, the root mean squared error, the model accuracy and kappa coefficient.