News And Events Driven Financial Market Business Forecasting

The fluctuation of financial market is continuous. Besides, the financialstatistics are always organized in the order of time. Thus, to solve theforecasting problem of financial market business, time series analysis isalways the common choice. However, only use financial statistics itself fortime series model analysis cannot generate perfect forecasting result, as themodel only uses a small percentage of all the information about financialmarket. The valuable information of policies, official statements, commentsof experts and agencies and people’s opinions are not being considered,which can deeply influence the forecasting result. These external indicatorscannot be reflected from statistics data, but is always maintained in financialnews and events which is the main part of media information. In this paper,the time series analysis of financial market will be combined with datamining on financial news and events. The external indicators of financialmarket can be obtained by sentiment analysis on news and events corpus. Thequantifiable results of the indicators can be important complement andmodification of time series model to generate better forecasting result. In thispaper, we use micro-blog to be the media information resource of financialmarket business. As the particularity of Chinese micro-blog, the sentimentanalysis can be achieved by rule-based Chinese syntactic parsing algorithm.The attitude of future trend of financial market can be obtained inquantitative form. An NARX time series model based on neutral network iscreated for fitting and forecasting. According to the analysis and forecastingresult generated by the model mentioned in this paper from real stock index data, the conclusion shows that the targeted sentiment analysis algorithm forChinese micro-blog can obtain better quantitative value of external indicators,and combining the external indicators with original time series model canforecast the trend of financial business better.