Skip to content

SwaksharDeb/Investment-decession-making-with-reinforcement-learning

Repository files navigation

Investment Decision Making with Deep Reinforcement Learning

This is a repository for an automated trading bot for the stock market using reinforcement learning algorithms. Specifically, here we implemented Deep Q Learning, Double Deep Q Learning algorithms with Prioritized Experience Replay.

How to Run

DDQN_PER.py is the code for the double deep Q learning with the prioritized experience replay algorithm. DQN_PER.py is the implementation for deep Q learning with prioritized experienced replay. dqn_agent.py is the deep Q learning with a vanilla experience replay algorithm. model.py is the architecture of the neural network that has been utilized in this project.

To train the stock trader bot using Deep Q Learning with Prioritized Experience Replay run this command in the terminal python stock_trader_PER_trend.py . To train the stock trader bot using Double Deep Q Learning with Prioritized Experience Replay run this command in the terminal stock_trader_trend_DDQN_PER.py. To train the stock trader bot using Deep Q Learning with Vanilla Experience Replay run this command in the terminal stock_trader_with_trend.py.

Results

The results are shown using deep double Q learning with prioritized experience replay.

Result in the training set. The training set is the 2018 Walmart stock market.

Result in the test set. The test set is the 2019 Walmart stock market. DDQN_PER with market factors training set.png

Releases

No releases published

Packages

No packages published

Languages