The Usage of Machine Learning Methods for Fundamental Analysis of Exchange Rates

This project proposal is an attempt at highlighting some issues related to the sphere of finance. The central research question could be framed as follows: what methods of machine learning are most effective in conducting a fundamental analysis of exchange rates. To answer this question, it is proposed to get acquainted with all the existing methods and to test the effectiveness of each model by practical way. The main results of the research have created a model for the analysis and the forecasting exchange rates. It is expected that the main results would be interesting for all participants of the currency market in the Russian Federation. In the course of the practical implementation of this work, a hypothesis was put forward on the relationship between the data of two neighboring trading days on the exchange. The hypothesis was as follows: in order to solve classification or regression problem and study the time series, it is necessary to shift the values of the variable factors by 1 day ahead, and leave the value of the target variable at its position. This hypothesis was confirmed, as it was possible to create models demonstrating the high quality and the explicit dependence between independent variables and the target variable on the training sample. It was shown that the fundamental analysis could not be reduced to the study of the time series or classification problem solving as these models had demonstrated extremely poor quality on the test sample. Simultaneously, it was possible to build a model for the forecasting exchange rates with the high quality using regression analysis. In the course of the study, it was established that the best models for the fundamental analysis of exchange rates could be based on building a decision tree using the gradient boosting algorithm or by dint of constructing a random forest. Keywords: time series, regression analysis, classification, decision trees, gradient boosting, random forest, exchange rate, USD, EUR, RUB.