Skip to content

Repository files navigation

Deep-Reinforcement-Learning-in-Stock-Trend-Prediction

This is the final project for the Reinforcement Learning course at NTNU. Inspired by Kelly et al. (2023) in their work, '(Re-)Imag(in)ing Price Trends,' I have built a deep Q-learning agent with a CNN network to analyze stock candlestick charts. Although the agent cannot significantly outperform the market in the testing period using only OHLC (Open, High, Low, Close) and volume data for the candlestick charts, I believe that incorporating additional data, such as the ratio of retail investors and the ratio of margin sales, could enable the agent to eventually uncover some unexplainable alpha.

About

This is the final project for the Reinforcement Learning course at NTNU.

Topics

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages