Description

Various data sources are available in the era of Big Data to gain an improved market understanding. As one example, price prediction based on news ticker contain a broad range of valuable information. Such an automatic analysis will support traders to maximize profits. Only a few approaches regard textual documents in this context. All of them predict in specific time intervals. The paper presents a general forecasting approach to process news ticker and market data in real time. Price trend classification is executed applying data mining techniques. However, the effect on prices by news tickers is hard to obtain. Irrelevant tickers will decrease the performance. Several approaches are tested to identify relevant articles in an automatic fashion. A prototype is implemented and the functionality is demonstrated in two different case studies. Our approach supports any market where important events have to be considered instantly. _x000D_

Share

COinS
 

Price Trend Forecasting Through Textual Data

Various data sources are available in the era of Big Data to gain an improved market understanding. As one example, price prediction based on news ticker contain a broad range of valuable information. Such an automatic analysis will support traders to maximize profits. Only a few approaches regard textual documents in this context. All of them predict in specific time intervals. The paper presents a general forecasting approach to process news ticker and market data in real time. Price trend classification is executed applying data mining techniques. However, the effect on prices by news tickers is hard to obtain. Irrelevant tickers will decrease the performance. Several approaches are tested to identify relevant articles in an automatic fashion. A prototype is implemented and the functionality is demonstrated in two different case studies. Our approach supports any market where important events have to be considered instantly. _x000D_