Stock Price Prediction Using Machine Learning Techniques

Large variations in stock price and unstable fluctuations will not only bring huge losses but also might influence the decisions made by investors. Machine learning in financial areas has been widely used in the past few years, and they are especially superior in forecasting tasks. This article will talk about how machine learning techniques could be used to predict stock prices and some possible ways of improvements. The main method used in this article is the LSTM structure, and the building blocks are constructed into a full Recurrent Neural Network to predict the price for several famous technology companies from NASDAQ. According to the results in this article, a forecasting algorithm based on the LSTM model can minor the error down to a few dollars. It is overall not accurate enough and cannot be used to draw a financial conclusion about investment, but still possible to explain some personal confusion and give solutions to individual difficulties. In addition, this study can be further upgraded in terms of structure and logic. This article hopes to provide a broader research perspective for the application of machine learning in the financial field.