DEEP LEARNING TECHNIQUES APPLIED IN STOCK MARKET FORECASTS (TÉCNICAS DE DEEP LEARNING APLICADAS EM PREVISÕES NO MERCADO DE AÇÕES)

Resumo

https://doi.org/10.5281/zenodo.15670699

The Brazilian stock market has proven to be very profitable in recent years, even in adverse scenarios. For example, in 2021, when many industries faced significant challenges such as economic crises and political instability, affecting a wide range of sectors, the market continued to attract investors, with some companies achieving impressive results and surpassing expectations. This growth reflects the importance of the stock market as one of the pillars of the Brazilian economy, being essential for the development of businesses of all sizes and sectors, from the largest corporations to small businesses. In times of crisis, the stock market offers opportunities for both investors and companies seeking to capitalize on their operations and expand their activities. Given the relevance of the stock market to the Brazilian economy, we aimed to analyze various factors that influence stock performance, such as economic policies and fluctuations in interest rates. To forecast stock prices for different companies, we collected historical stock price data along with economic indicators.

This data was then processed and analyzed using time series models, classification and regression techniques. The time series models facilitated the identification of patterns over time, while the classification and regression techniques helped detect the key factors that influence price variations. With this combination of methods, it was possible to conduct a thorough analysis; now we present the results obtained from the application of each model.