Prediction Of Stock Price Trends Based On Rough Sets

The stock market has an important impact on financial market and social stability. In recent years, investors pay more and more attention to stock market with the rapid development of the stock market. People expect predict future stock price accurately. Influenced by many complex factors, the stock price is difficult to predict. Improving the prediction accuracy of stock price not only can help investors reduce investment risks and increase profitability, but also help the government keep abreast of future changes in the market, so as to formulate effective policies, maintain financial market and social stability. Therefore, the establishment of effective stock price prediction model has important significance.The paper analyzes the disadvantages of the existing prediction methods based on reference a large number of literatures. Taken into consideration the lack of the exiting prediction models and the characteristics of the stock market, we propose to establish a prediction model based on rough set theory.SSE 180 Index, China Merchants Bank and other three samples will be used to describe the procedure of this prediction model. First, the paper select technical indicators as conditional attribute based on a correlation matrix. The decision attribute is the change of the stock price in one day, then create decision table. The stock data in 2013 is employed as training data, the data in January 2014 as sample data to verify the validity of the rules. All attributes were discredited using a semi na?ve scaler algorithm. Once data discrete has occurred, reduction can be carried out. After the reducts were acquired, the decision rules pertaining to each reduct were generated. Finally, test the generated rules to verify the validity. The empirical results show that this model can get a good result and it can provide effective suggestion to investors.