02. Instructions

Project Instructions

• Setup and Data Preparation: You will begin by setting up your development environment and preparing historical market data for training your agent.
• Environment Design: Next you will create a simulated trading environment where your agent can learn and test its strategies.
• Agent Implementation: You will then encode the DQN model, defining the neural network architecture and training loop.
• Training the Agent: Once all components are complete, you get to train your agent using historical data. For this project, we will be using a small dataset and a low number of training rounds, in order to save time and focus on the process of building our agent and training loop.
• Evaluation and Testing: Finally, you will evaluate your agent's performance through a single validation test.

As you progress, you will find additional instructions and sections to complete inside the Project Workspace.

Putting It All Together

In this project, you'll apply the concepts and techniques you've learned in previous modular assignments. Each module provided you with a critical building block—whether it was understanding RL fundamentals, designing reward functions, or implementing DQN algorithms. Now, you will integrate these elements to develop a fully functional trading agent. This will cement your understanding of each component and illustrate how they interact within a complete RL system. By seeing how everything plays together, you’ll gain a deeper appreciation of the intricacies involved in creating an effective RL trading agent.

Keep in mind that this project is meant to teach you the building process - the goal of this project is not to create a profitable trading agent. Once you complete this project, I encourage you to review your learnings regarding how to improve your results, such as optimizing reward functions and hyperparameters in training, training for longer or on more data, and utilizing backtests with risk reducers such as stop loss and take profit in order to optimize your learned strategy. Use these learned concepts along with the completed project and your own intuitions in order to make your own, personalized, profitable trading agent. Let's get started on building your very own RL trading agent!