Financial market prediction is one of the most attractive research areas in financial data mining field because of its available data and potential profits. Financial market prediction has been paid much attention by many researchers and practitioners. Huge numbers of forecasting methods including judgmental methods and statistical methods have been proposed. In previous studies, judgmental methods mainly focus on expert judgments, which are waste of resources in some extent and may lead to lower accuracy, while statistical methods in terms of statistical models can achieve better performance, but when the unpredictable events appear, these models are ineffective and useless sometimes. In this paper, by incorporating online user’s opinions, two novel paradigms are proposed for financial market prediction, which may overcome the aforementioned shortcoming. In the first paradigm, the opinions of several selective online users are extracted, and then their opinions are integrated to forecast financial market. In the second one, a data mining model is constructed by combining user’s opinions and financial time-series data, and then the model is used for financial market prediction. Moreover, these paradigms are validated and compared using real financial market data. The empirical results show that our proposed paradigms are useful and feasible for financial market prediction, and furthermore, the combined model outperforms the traditional judgmental model. These findings imply that the proposed method is a promising alternative for financial market prediction.
Xu, Wei; Sun, Daoyuan; Meng, Zihe; and Zuo, Meiyun, "Financial Market Prediction Based On Online Opinion Ensemble" (2012). PACIS 2012 Proceedings. 137.