Predicting stock market indices using fusion of machine learning techniques

Keywords: Python, IMOEX, HSI, S&P500, machine learning, price forecasting for stock market instruments, technical analysis indicators. This paper describes machine learning methods used to predict the closing price value of stock market indices such as IMOEX, HSI, and S&P500. The study consists of 3 parts. The first part contains theoretical concepts for predicting financial instruments using machine learning. The second Chapter examines previous research in this area around the world. In the third part, machine learning algorithms are implemented for predicting indexes that use technical analysis indicators as input data. In the practical part, we compare the one-step approach with the two-step approach of machine learning. The one-step approach compares the quality of SVR, MLP, and SVR-MLP-LR models. In the two-stage algorithm, technical analysis indicators are predicted using SVR at the first stage, and prices are forecast using SVR, MLP, and SVR-MLP-LR models at the second stage. Then we compare the forecast quality of one-stage approach models with two-stage approach models based on the metrics MSE, rRMSE, MAE, MAPE, and MASE.