Reproduce “Synergistic Formulaic Alpha Generation for Quantitative Trading based on Reinforcement Learning” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Hong-Gi Shin, Sukhyun Jeong, Eui-Yeon Kim, Sungho Hong, Young-Jin Cho, Yong-Hoon Choi
Researchers at Kwangwoon University developed a reinforcement learning framework that enhances the automated discovery of formulaic alpha factors for quantitative trading. Their approach expands the search space for alpha factors and initializes the RL agent with pre-generated synergistic alpha sets, leading to an Information Coefficient of 0.085 and consistently higher cumulative returns in backtesting compared to existing methods.
Key insights
- Expanding the search space for formulaic alpha factors leads to the discovery of more effective individual alphas and better overall combinations, reflected in higher Information Coefficients.
- Initializing a reinforcement learning agent with an already synergistic set of pre-generated alpha factors significantly boosts its ability to efficiently discover even more potent and complementary alpha factor combinations.
- Generic "off-the-shelf" alpha factors, such as those from Alpha 101, often do not provide synergistic benefits when used for initialization and may be discarded by the combination model due to poor individual performance.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.