Abstract:

With the rapid development of the economy, uncertainty of financial activities has been increasing, so financial price prediction attracts great attentions from the academia and the financial industry. How to obtain predictive judgment and speculation and how to grasp the prospective development tendency and regularity of the financial world becomes urgent. Recently, with the help of rapid internet development, solely relying on the data mining technology to obtain massive financial information on the price history cannot reflect the impact of multiple factors in the financial market. We therefore propose a financial price trend prediction method based on financial news mining, which can digthe emotional tendency messages from the financial news information. Combined with the financial historical price data, we use a combination model of multiple linear regression and differential self-regression to predict future financial price trend. Real data show that the proposal can obtain more accurate prediction results.

Key words: financial news, financial market predicting, multiple linear regression, ARIMA, sentimental information