The stock exchange is extremely dynamic, making operations in stock market a complex task with high intrinsic risk. In contrast, the financial return can be far superior to other investment modalities. Attempts to anticipate asset pricing movements in such environment can significantly reduce the risk of operation, although there is not a consensus on the possibility of making this type of prediction. Based on the premises of the technical analysis, that market assimilates and responds to political and emotional factors, and price movement follows certain trends and recurrent patterns, this research project proposes stock price movement trend predictions in the Brazilian financial market using Machine Learning, in particular, Support Vector Machine (SVM) and Long Short Term Memory (LSTM) neural networks, in order to optimize the decision-making process. For algorithm training, finance time series will be applied, containing the price history of the selected assets, as well as the technical indicators that will be generated from the time series. Results will be analyzed through metrics that compare the real values and those obtained by the algorithms, allowing to check the most efficient method for solving the problem. The application of the aforementioned techniques will make use of the Python programming language, one of the main languages currently adopted in Data Science and Machine Learning.
News published in Agência FAPESP Newsletter about the scholarship: