Wavelet-Aided Stock Forecasting Model based on Ensembled Machine Learning

Abstract

The stock market is a barometer of a country's economic situation. The research on the stock market is always highly valued, and the prediction of short-term stock price trends is the focus of investors. The stock price data not only has time-domain correlation, but also has certain independence due to the influence of the market environment. In this study, we focus on predicting stock price movements through machine learning, which is a challenging task because there is a significant amount of noise and uncertainty in the information related to stock prices. Therefore, this paper utilizes wavelet transform and multi-step smoothing to denoise the data, obtain the multi-dimensional stock price feature vectors. Subsequently, we apply the LightGBM classification algorithm to predict the price trend in ten days. Experimental results show that the method proposed in this paper has noticeable advantages in the task of short-term stock price trend prediction.