Short-term Stock Market Price Trend Prediction Using a Customized Deep Learning System

In big data era, deep learning solution for predicting stock market price trend becomes popular. We collected two years of Chinese stock market data according to the financial domain, proposed a fine-tuned stock market price trend prediction system with developing a web application as the use-case, meanwhile, conducted a comprehensive evaluation on most frequently used machine learning models and concludes that our proposed solution outperforms leading models. The system achieves an overall trend predicting accuracy of 93%, also achieves significant high scores in other machine learning metrics in the meantime. Thus, this work provides a solid foundation for further price prediction by classifying the price trend accurately. With the detail-designed evaluation on prediction term-lengths, feature engineering and data preprocessing methods, this work also contributes to the stock analysis research community in both financial and technical domain.