Predict Abnormal Stock Returns with Neural Network Approach: )Evidence from Tehran Stock Exchange(

One way to help investors, companies and others are involved money market and capital, prediction models about the overall prospects of companies and so that investors can make good decisions [12]. Presented theories and models to predict abnormal returns indicate that there is no absolute consensus. One of the methods is very good at forecasting the financial variables such as stock prices, stock returns, stock market crash and the use of neural network approach. The major advantage of this method over other methods can be found in this issue that better data consistency is maintained [54].
In this research to predict abnormal returns stock of two artificial neural network and fuzzy neural network approach was used in this way accurately predict abnormal returns are examined by this tool.
Input variables to predict abnormal returns include earnings forecast, the degree of financial leverage, return on investment, accounting transparency, conservative accounting, the value of the brand and management have been too confident. For this purpose were examined, 452 companies - year screening method for a period of five years (2017- 2012) of the companies listed in the Tehran Stock Exchange. Findings demonstrate the power of predictive artificial neural network, fuzzy neural network to predict abnormal stock returns was more than that with the margin of error is less.