Desenvolvimento de um sistema preditivo para análise de ativos acionários e gestão de risco

The volatility of the financial market demands robust analytical methods for decision-making and risk management. This work presents and validates a model for forecasting and analyzing financial data, combining time series analysis with advanced Machine Learning algorithms. The main advantage of this predictive model is the elimination of human emotional and cognitive bias. In this sense, the software analyzes a large amount of historical data via CSV files. This capability allows for the identification of complex patterns through statistical analysis. Furthermore, based on the collected data, the applied approach provides a data forecasting model for the following days. The LSTM (Long Short-Term Memory) model was built using the Python language, with HTML, CSS, and Javascript for the user interface. This tool allows the user to simulate capital investments, analyze the historical performance of Artificial Intelligence (AI), and predict the future volatility of the asset. It also allows the user to check the AI's performance metrics against the data, in addition to a graph for better data visualization. The results were promising, as the Mean Absolute Error (MAE) and the Root Mean Square Error (RMSE) for Petrobras shares demonstrated high consistency in price forecasting.