Reinforcement machine learning system discovers hidden market relationships and constantly learns how to provide early cues for price direction. The neural network has been initially trained on intermarket correlation data from stocks, bonds, commodities, and currencies, as well as how independent global events impact everyday trading decisions. For each and every trading instrument, the reinforcement learning network identifies and analyzes several hundred of its correlated data in real time. The performance of this model is further enhanced using feature engineering. Feature engineering is the process of using domain-specific knowledge to create additional input data that improves the machine learning model. By adding insightful technical indicators to the data set, as well as the output from the StatsModels SARIMAX prediction model should provide a nice balance of useful observations for the machine learning model to learn and act in this complex environment.