Reinforcement Learning for Algorithmic Trading

Testing applications of RL to algo trading. Thus far, work includes:

• Simulating stock prices as sinusoids with added noise having a specified distribution
• Fetching historical stock data using the yfinance API
• Implementing Q-learning algorithms using TD(0) update rules
  • Single stock: Q-table
  • Multiple stocks: Approximate Q-function with artificial neural network containing single hidden layer
• Applying Supervised Learning to enrich/augment/improve the agent's environment for decision making
  • Pull data from FRED API to construct macroeconomic environment
  • Applying SVMs with various kernels to predict improvement in economic conditions for trading