Identifying and ranking predictors of stock bubble: Application of Logistic regression and artificial neural network

The aim of this study is to identify and ranking the factors predicting stock price bubble in the Tehran Stock Exchange. For this purpose¡ at first through skewness¡ Kurtosis and runs tests on price of 158 stock symbol bubble status during the period from 1389 to 1392 were identified. According to research literature¡ affecting factor including information transparency¡ leverage¡ liquidity¡ ratio of book value to market value¡ p/e¡ liquidity¡ institutional ownership and firm size were used. Then¡ using logistic regression effect of this variable on price bubble was confirmed. Results show that the increase in transparency variables¡ B/M¡ liquidity¡ institutional ownership and firm size reduces the probability of forming bubble in share prices. After training artificial neural network¡ using the sample data the network were optimized by out of the sample data. Finally¡ using sensitivity analysis through neural network¡ these variables based on the ability to predict the share price bubble were ranked.