This work focuses on the study of the application of a machine learning model known as LGBM (Light Gradient Boosting Machine) in classifying the best moments to buy a Brazilian stock market asset, serving as a trading system. As predictive variables, technical market indicators calculated from the historical price series are used. To define the best time to buy the asset, the three-barrier method is applied. Furthermore, this work presents a practical approach with the inclusion of Take Profit and Stop Loss limits.