{"repo":"amin-sharifi-github/quant-rl-trading-agent","free":true,"listed":false,"github":"https://github.com/amin-sharifi-github/quant-rl-trading-agent","clone":"git clone https://github.com/amin-sharifi-github/quant-rl-trading-agent.git","description":"End-to-end RL trading framework with PPO agent, self-attention neural network, custom Gym environment, and advanced backtesting.","language":"Python","stars":29,"topics":["ai","algotrading","quantitative-finance","quantitative-research","quantitative-trading","reinforcement-learning","reinforcement-learning-agent","reinforcement-learning-algorithms","trading-systems","attention-mechanism"],"license":"MIT","category":"trading","readme_excerpt":"QuantRL: Deep Reinforcement Learning System for Algorithmic Trading QuantRL is a modular, research-grade reinforcement learning (RL) framework designed to model, train, and evaluate AI-based trading agents in a realistic financial environment. It is built around the Proximal Policy Optimization (PPO) algorithm using a custom self-attention neural network architecture and incorporates feature-rich market simulation, advanced backtesting, and performance evaluation. --- System Architecture Overview The diagram illustrates QuantRL’s core training loop: historical market data is processed by the DataHandler , passed through a VecNormalize wrapper, and fed into a custom TradingEnvironment built on OpenAI Gym. The PPO-based RL Agent , enhanced with a self-attention policy network, interacts with the environment and updates its strategy through repeated episodes. This modular pipeline enables flexible experimentation and rigorous evaluation. Key Features End-to-End Research Pipeline - Historical OHLCV data ingestion from CSV or yfinance - 50+ engineered features, including: - Price-based indicators (returns, momentum, log returns) - Technical signals (RSI, MACD, Bollinger Bands, ATR, ADX) - Market structure metrics (volatility regimes, microstructure signals) - Time-of-day and session-based features (intraday structure) - Normalization options: standard, min-max, and robust scaling - Temporal train/validation/test split to ensure realistic evaluation Custom OpenAI Gym Environment - ","default_branch":null,"files":null,"tree":[],"storefront":"/r/amin-sharifi-github","claimed":false,"request_supported":{"post":"https://gitbuyer.com/r/amin-sharifi-github/quant-rl-trading-agent/request-supported","requests":0},"note":"indexed from public GitHub; nothing is for sale on this page. Clone it from GitHub. Paid listings live at /search."}