A Comparative Study of Machine Learning Algorithms for Financial Data Prediction

Machine learning has been widely used as a part of stock market investment strategies, whether for forecasting exchange rate or market volatility, and also in different classification problems in order to help for decision making. In this paper, we work on Brazilian and Chilean currency exchange data using two different algorithms. These experiments are based on 15 years historical data, the predictions are made for the next day price of the action. In addition, of historical data, our dataset includes ten technical indicators as inputs of each prediction model. Our contribution is a comparison of neural networks algorithms results with a multi linear regression algorithms optimized with particle swarm Metaheuristic.