The Application of Big Data Screening and Machine Learning in Developing Investment Strategies for the Stock Market

Substantial investors often rely on valuation and financial indices to forecast stock prices, but no method guarantees profits, as even financially strong companies can see their stock prices decline. In this case, more research should be done to look for some specific characteristics of the rising of stock price and verify of what degree can these characteristics help forecasting the changes of stock price. This study will shift out the relatively valuable data with the help of a system function which represent the difficulty to predict a result for polytomized variable. For continuous variable, using reduced classification method or calculating the variance of data to evaluate its value. Moreover, this study will explain how to evaluate machine learning models with cost matrix and how to improve the accuracy of machine learning models with bagging method and Adaboost method. With the theories introduced in this study, investors can shift out more valuable message from applying the difficulty of prediction function or calculating variance of continuous variable data and update their invest strategies more efficiently and earn more profit with the advanced machine learning models.