Forecasting markets, including stocks, has been attractive to researchers and investors due to the high volume of transactions and liquidity. The ability to predict the price enables us to achieve higher returns by reducing risk and avoiding financial losses. News plays
More
Forecasting markets, including stocks, has been attractive to researchers and investors due to the high volume of transactions and liquidity. The ability to predict the price enables us to achieve higher returns by reducing risk and avoiding financial losses. News plays an important role in the process of assessing current stock prices. The development of data mining methods, computational intelligence and machine learning algorithms have led to the creation of new models in prediction. The purpose of this study is to store news agencies' news and use text mining methods and support vector machine algorithm to predict the next day's stock price. For this purpose, the news published in 17 news agencies has been stored and categorized using a thematic language in Phoenician. Then, using text mining methods, support vector machine algorithm and different kernels, the stock price forecast of the chemical products group in the stock exchange is predicted. In this study, 300,000 news items in political and economic categories and stock prices of 25 selected companies in the period from November to March 1997 in 122 trading days have been used. The results show that with the support vector machine model with linear kernel, prices can be predicted by an average of 83%. Using nonlinear kernels and the quadratic equation of the support vector machine, the prediction accuracy increases by an average of 85% and other kernels show poorer results.
ارسال
Manuscript profile
