Stock Trend Prediction Using Candlestick Charting and Ensemble Machine Learning Techniques With a Novelty Feature Engineering Scheme

Stock market forecasting is a knotty challenging task due to the highly noisy, nonparametric, complex and chaotic nature of stock price time series. With simple eight-trigram feature engineering scheme inter-day candlestick patterns, we construct novel ensemble machine learning framework for daily pattern prediction, combining traditional charting with latest artificial intelligence methods. Several techniques, including deep methods, are applied data predict direction closing price. This can give suitable prediction method each based on trained results. The investment strategy constructed according techniques. Empirical results from 2000 2017 China's confirm that our has effective predictive power, accuracy more than 60% some trend patterns. Various measures such as big data, standardization, elimination abnormal effectively solve noise. An excels in both individual portfolio performance theoretically. However, transaction costs have significant impact investment. Additional technical indicators improve forecast varying degrees. Technical indicators, especially momentum most cases.