Systematic trading strategies powered by transformer-based reinforcement learning.
Research-driven. Backtested. Open-source.
Cutting-edge reinforcement learning infrastructure for algorithmic trading research.
Open source. Transparent. Research-first.
Publicly available research code with transparent methodology and real trading results
Cutting-edge RL implementations with modular architecture for rapid experimentation
Active development pipeline focusing on advanced trading strategies and market modeling
Open to research partnerships, technical discussions, and collaboration opportunities in quantitative finance
While the world chases LLMs✨ and AI art🎨
we're reviving reinforcement learning for quantitative finance
"It started with frustration - watching RL get abandoned for shiny new toys like LLMs and Generative AI while Reinforcement Learning in trading remained a niche. I wanted to build a system that could actually learn from the market, not just regurgitate past data."
"30,000 raw features later, I learned the hard way that most were noise. The current pipeline uses aggressive PCA dimensionality reduction to distill market structure into something the transformer can actually learn from."
"Now the real work begins - making this actually trade live without blowing up accounts. Model v2.0 is in development with improved risk management and live integration consuming all my debugging patience."
💭 Current Focus:
"Working on hyperparameter optimization and transformer architecture improvements. Target: 900+ episode reward by end of Q4."
"The markets are never wrong, only opinions are."
This is a research-first project. If you're tired of AI hype and want to work on real trading systems, join the fight.
Direct communication for quantitative finance research and collaboration
Real-time performance data from our PPO transformer training sessions
Model achieved 860 average episode reward after 325.40 minutes of training
Live repository metrics and development insights from the quantitative trading research project
Quantitative reinforcement learning research
Interested in quantitative finance research, algorithmic trading strategies, or collaboration opportunities?
Direct communication for research discussions, collaboration proposals, or technical inquiries.
seanman519@gmail.comContribute to the project, report issues, or explore the codebase. All research is open and transparent.
View Repository