FINANCIAL ENGINEERING AND SECURITIES MANAGEMENT (PORTFOLIO MANAGEMENT)
Year:2012 | Volume:3 | Issue:11|Papers Count:7

In this research we have carried out a comparative forecasting of Tehran stock price index (TEPIX) based on the artificial Neural Network (MLP, RBFN, GRNN) and the Box-Jenkins model. The forecasts results were compared using RMSE, MAPE, MAE and R2 criterion. The steps involved were building ANN and ARIMA models first and then using this models to forecast the Tehran stock price index. Total of days were 1142 days that we used 80% of them for instructing of models and 20% of them for testing. For making of ANN models we used MA 1LAB and for making of Box-Jenkins models we used SPSS and EVIEWS. The results showed that all of ANN models performed better than Box-Jenkins models. in the other hand, RBFN models performed better than the other ANN models.GRNN models performed the worst among ANN models.

Surely one of the most important topics for financial economists is to identify the relationship between profitability and corporate financing decisions or capital structure. Optimal capital structures will maximum value of the company and should minimum cost of capital. Financial economics are always in challenge to achieve optimal capital structure and maximum profitability. This research analyzes the relationship between capital structure and profitability of listed companies in Tehran Stock Exchange (TSE) based on Generalized Method of Moment (GMM) panel approach. In this research we use financial statements (balance sheet and profit and loss account) of selected companies that listed in TSE during the five year period (2006 to 2011) as balanced panel data set.In order to test the hypothesis, first we use pooled regression to determinant the relationship between capital structure and profitability. Then we use panel data set to achieve more efficient estimation. After employing chow and hasman test, we selected fixed effect panel data model.Our results from linear mood suggest that the relationship between leverage and market structure isnegativedue to the complex interaction of market conditions and Problems based on debt financing (in terms of rate of intake facilities and profitability of projects) in the companies. Our evidence shows that Tehran listed companies are generally more subjected to Pecking order theory that means profitable companies tend to use internal financing than debt financing.Finally, we employed GMM method to reach on more efficient result and accounts for unobservable firm-specific characteristics and endogeneity problems. The results show a saucer-shaped (U shape) relationship between capital structure and profitability and also combined interpretation of theories in three stages (low, average and high profitability) is more efficient

This research has been conducted in order to introduce a proper pattern for Iranian heavy crude oil prices. The applied data in this research has been dated on third week of April 2002 to fourth week of July 2011 and is composed of 485 observations, which are employed to severance within the sample forecasts and out of the sample forecasts, in order to estimate model coefficients, almost 90% of observations (about 50 observation) has been used and remained, observations. has employed in out of sample predictions. In this research templates have been used as: An Artificial Neural Network model based on genetic algorithms and also a Linear regression model. The findings show that artificial neural network model in out of sample forecasts, according to Forecast error mean square error (MSE) criteria and also Mean squared error criterion (RMSE), have a much better performance comparing to Linear regression model (ARIMA).

Many studies have examined the occurrence of earnings management in various contexts. The most studies have assumed that earnings are managed through accounting accruals. Thus, a variety of accrual based earnings management detection models have been suggested. The ability of these models to detect earnings management has, however, been questioned in many research. An explanation to the poor performance of the existing models is that the most models use a linear approach for modeling the accruals process even though the accrual process is non-linear accordance to several studies. An alternative to deal with the non-linearity is to use various types of neural networks. The purpose of this study is to assess whether selected mathematical models can be used in detecting earnings management and also whether neural network-based models outperform linear-based model in detecting earnings management. The study comprises neural network models based on a Multilayer perceptron (MLP) and a Radian Basis Function (RBF). The results show that the final selection between these two models is under question and it depends to the ability of modeling and the type of the selected topology.

Brand is one of the most important elements in marketing and success of a company, an image that introduces the company implying what it is or wants to be in the mind of observers, customers and consumers. In fact, brand is a symbol that distinguishes a company from its competitors meanwhile introducing its offered products and services.This research is based on the hypothesis that "The brand valuation models have the ability to evaluate the brand value of banks under study." In this research, the brand valuation models have been studied and adjusted to banking industry specifics, and finally Intangible Business Model has been selected as the best matched model using TOPSIS method. Practically applying Intangible Business Model, the brand valuation of Mellat, Pasargad and Persian banks have been carried on as the case studies. Evaluated and ranked according to the model results, Mellat, Persian and Pasargad stand respectively. These brands, afterwards, were evaluated using the second matched model called Financial World in order to compare and validate the initial results, The new results were almost confirmative, the small divergences of which can be explainable if differences in model specifications are considered. At the end, the resulted brand values derived from both models were compared to the total asset value of the each bank. The outcomes support this research hypothesis.

One of the most important issues in financial markets is the stock return prediction. Various models was introduced for this purpose. This paper compares two most applicable stock return prediction models, Fama and French three factor model and general regression neural network (GRNN). Two hypotheses have been designed for this purpose.The first one compares the models accuracy in stock return prediction among the selected companies, and the other one do this job among 6 portfolios which constructed in order of companies size and book to market ratio.The study uses Tehran Stock Exchange companies' data from 1378 to 1388.After the calculation of models MSE, the equality of average of MSEs for two models tested by t-paired sample statistical test at confidence level 9 5 %.The results show that there is a meaningful difference between models prediction accuracy in the rate of return prediction. This difference means that the GRNN model out performs the Fama and French model at in dividual stock and portfolios.

In the seventy decade the Black Scholes model had a critical role in the pricing of financial derivatives. But later, due to its major weakness, other various models were presented. The class of Levy processes is one of the most common models used for pricing financial assets. The one of this processes is Generalized hyperbolic process that is based on the generalized hyperbolic distribution.First, In this paper we introduce this distribution and then fit data related to the stock price in Iran market. (Iranian asset returns)