Reproduce “NEAT Algorithm-based Stock Trading Strategy with Multiple Technical Indicators Resonance” 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
Li-Chun Huang
A trading strategy based on the NeuroEvolution of Augmenting Topologies (NEAT) algorithm, integrating multiple technical indicators, achieved lower risk and greater stability in stock trading compared to a Buy & Hold strategy. The evolved model demonstrated a 20% lower standard deviation of return and an exposure time of 86.98%, while maintaining an average return of 18.76% on S&P 500 constituents.
Key insights
- NEAT can effectively evolve complex, adaptive neural network architectures for dynamic financial tasks, extending its utility beyond traditional domains.
- Carefully crafted multi-objective fitness functions are crucial for guiding evolutionary algorithms to balance conflicting financial goals, such as maximizing returns while minimizing risk and encouraging specific trading behaviors (e.g., swing trading).
- Progressive training with varying data durations can enhance efficiency and focus the model on different strategic objectives (e.g., trend identification, risk management) across different training stages.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.