Use of Machine Learning Methods for the Analysis and Forecasting of Financial Time Series

This paper represents a research of machine learning methods of predicting and analysis of financial time series. The paper contains three chapters. The first part includes a description of practices in prediction and analysis of stock markets. The second chapter contains a description of machine learning methods of predicting and analysis of financial time series. The third chapter of the paper includes practical implementation of model that is able to predict financial time series. Theoretical and practical aspects of machine learning methods of predicting and analysis of financial time series were researched. Anaconda distribution and Python programming language have been used to complete the practical part of the research. The paper contains evaluation and description of common stages of model implementation. A comparison of efficiency of various machine learning methods was made using empirical evidence. The practical result of the research are ARIMA and LSTM models, which are able to achieve a high forecast accuracy. The results of the research could be interpreted as positive, as high efficiency in predicting of financial time series has been achieved with the practical implementation, proposed in the paper. The results of the research could be interpreted as the evidence of the efficiency of machine learning methods in forming stock market strategies.