Recommending Buy/Sell in Brazilian Stock Market through Recurrent Neural Networks

Resumen

This work aims to evaluate the accuracy of Long Short-Term Memory Neural Networks to recommend Buy/Sell signals of some Brazilian Stock Market Blue Chips. The population of this study was composed by top 5 volume stocks, which represented nearly 40% of the total volume of Brazilian Stock Market in 2019. It was analyzed the following features: volume traded, closing and opening price, maximum and minimum price, and last five-day closing prices. Models created can forecast the next day's opening or closing price. Obtained results show that forecasting
and real values have a coefficient of determination (R2) from 0.91 to 0.99, depending on the stock.