Gholamreza Mansourfar; Farzad Ghayour; Shabnam Khaleghparast Athari
Abstract
The purpose of study is to investigate comparative ability of accountinginformation to predict indices volatility of companies listed in Tehran StockExchange using intelligent methods including Support Vector Machine,Artificial Neural Network and classic Logistic Regression model. Sample ofstudy includes ...
Read More
The purpose of study is to investigate comparative ability of accountinginformation to predict indices volatility of companies listed in Tehran StockExchange using intelligent methods including Support Vector Machine,Artificial Neural Network and classic Logistic Regression model. Sample ofstudy includes 91 companies listed in Tehran Stock Exchange that have beenclassified in 9 industrious groups during time period of 2003-3013.Considering 11 corporate financial variables, study results show that despitepredicting ability of around 60% by Support Vector Machine and ArtificialNeural Network, there is significant difference between actual and predictedresults. Classic Logistic Regression model also can explain only 4%industries’ indices volatility using selected 11 corporate financial variables.Finally, although intelligent methods are superior to classic methods,accounting information solely are not well-explainer variables for predictingindustry index volatility and variety of variables such as financial, political,economical are effective in predicting industry index volatility.