🤖 AI Quantitative Trading Bot

One-Stop AI Quantitative Trading Platform · From Learning & Simulation to Live Trading

中文版 | 日本語版

_{👇 Follow the WeChat account · Join the Knowledge Planet for more hands-on content and video tutorials}

    

🔥 New Features • 📖 Introduction • 🚀 Quick Start • 📊 Quant Strategies • 🤖 LLMs • ⛏️ Alpha Mining • 💾 Data • 🛠️ Tools • 🎁 Resources

• 🏢 Institutional Investors
• 👨‍💻 Retail Traders (with programming background)
• 🌱 Retail Traders (no programming background)

ai_quant_trade
├── ai_notes ........... Financial quantitative trading knowledge (Markdown / Jupyter Notebook knowledge system)
│   ├── 资源 ........... Continuously collected excellent resources from across the web
│   ├── 实战 ........... Hands-on usage, frameworks, libraries, and pitfalls
│   └── 热点 ........... Financial market hot topics, tech hot topics, paper interpretations
├── docs ............... Usage documentation for this repo
├── egs_aide ........... Trading assistant tools (monitoring tools, etc.)
├── egs_alpha .......... Alpha library & alpha mining
├── egs_data ........... Data acquisition & processing (Wind / open-source tools)
├── egs_fin_nlp ........ Text analysis (sentiment analysis, etc.)
├── egs_llm ............ LLM applications (stock prediction / financial analysis)
├── egs_online_platform  Online research platform strategies (JoinQuant / Uqer)
├── egs_trade .......... Local quantitative stock trading strategies
│   ├── paper_trade .... Live trading simulation (Wind)
│   ├── rl ............. Reinforcement learning for trading
│   ├── ms_qlib ........ Microsoft Qlib framework
│   └── vanilla ........ Traditional rule-based strategies
├── quant_brain ........ Core algorithm library
├── runtime ............ Model deployment and real-world usage
├── tools .............. Auxiliary tools
└── README.md

Each egs_*/ sub-project now carries its own requirements.txt — pick the project you want to run, then pip install -r requirements.txt inside its directory. There is no global dependency file at the repo root.

This repository is not packaged as a Python package yet. Please clone the entire project, then navigate to each egs directory for detailed usage instructions and principle explanations.

# 1. Clone the repo
git clone https://github.com/charliedream1/ai_quant_trade.git

# 2. Pick a project and install its own deps (no root requirements.txt)
cd egs_trade/vanilla/double_ma
pip install -r requirements.txt

# 3. Follow the project's own README to get started

Python baseline: 3.8+ on x86_64. Apple Silicon (M1/M2) users, please follow the conda setup notes in each project's README.

Track record of dependency and toolchain changes that affect end users. See docs/CHANGELOG.md for the full version history.

To stay compatible with the latest AI / RL / LLM ecosystem and to clear the 46 GitHub Dependabot security alerts flagged in Sep 2026, we rolled out a coordinated upgrade in PR #25 and the follow-up PR #43. All upgraded projects were end-to-end verified before merge.

1. gym → gymnasium. Custom environments built on gym 0.21 must:

   • import gymnasium as gym and from gymnasium import spaces;
   • return obs, reward, terminated, truncated, info from step();
   • return obs, info from reset() and accept a seed= kwarg;
   • rename metadata = {'render.modes': [...]} to metadata = {'render_modes': [...]}. See StockTradingEnv0.py for a worked example.

2. Python 3.8 still supported; 3.10+ should now install cleanly thanks to new wheels.

3. Apple Silicon: still prefer conda create -n quant python=3.8 first.

Each affected project ships a run_e2e.py smoke test:

# RL project (a001)
cd egs_trade/rl/a001_proto_sb3 && python run_e2e.py

# RL project (a002 / finRL tutorial)
cd egs_trade/rl/a002_finRL_tutorial/a01_Stock_NeurIPS2018 && python run_e2e.py

# LLM project (a04 — needs GPU ≥16 GB)
cd egs_llm/a01_train/a01_unsloth_stock_forcaster/a04_train && python run_e2e.py

These scripts are the upgrade regression baseline and are re-run after every dependency change to ensure the original examples still execute end-to-end.

• .github/dependabot.yml — weekly scans for GitHub Actions + pip packages; safe-version PRs are opened automatically.
• .github/workflows/codeql.yml — weekly CodeQL security + quality scan on master.
• See Security tab for live alerts.

📁 Code Directory: egs_trade

🎯 Each example comes with a comprehensive tutorial — from principles and usage to code walkthrough.

You can build an independent quantitative trading system locally, covering the following strategy types:

📁 Code Directory: egs_trade/rl

Since the AlphaGo vs. Ke Jie match in 2017, deep reinforcement learning has taken off.

Compared with ML and DL, RL is goal-oriented (uses interaction as the objective), while many other methods consider isolated sub-problems (such as "stock price prediction", "market prediction", "trading decision", etc.) and cannot directly obtain interactive actions. RL is directly oriented to "completing the commander's task" and can produce a sequence of actions.

Strategy List:

Backtest Results:

Although traditional strategies may seem outdated, they are more operable and still have some practical value. DL and ML often need to be combined with rules to work effectively.

1. Double Moving Average Strategy + Simple Handwritten Backtest Framework

   • Detailed Tutorial
   • Includes strategy code + self-built pure handwritten backtest framework
   • Includes nice plotting indicating buy/sell points
   • 🎯 Goal: Through this example, understand how to build a complete quantitative trading framework

2. Portfolio Management 7-Lesson Course

📁 Code Directory: egs_trade

1. Wind Local Live Trading Simulation: Double MA Strategy
   • Live trading simulation implemented with Wind software
   • Wind is often the data source of choice for major financial institutions; due to its high price, it's more suitable for institutions
   • 🏢 Target Users: Institutions

📁 Code Directory: egs_aide

1. Stock Monitoring with Excel V1

   • 👀 Hard to be spotted while monitoring stocks
   • 📋 Customizable stock watchlist
   • ⚡ Use Excel for fast calculation and data processing

2. "Stealth Stock Trading" Tool V2: Modular Monitoring System 🆕

   

   • 🏗️ Architecture Refactor: Upgraded from a single file to a modular excel_monitor package with Sheet Handler pattern; each Sheet refresh is isolated
   • 🔔 Alert Monitoring: Customizable upper/lower price/change limits; triggered rows turn red + popup reminders
   • 📈 K-Line Charts: Draw K-line charts with a button click in Excel (mplfinance candlestick + MA lines), no need to switch software
   • 💰 Capital Sentiment: New Sheet aggregating northbound capital + Weibo sentiment + news sentiment + Guba hot posts
   • 🔍 Stock Pool Selection: Built-in all A-shares; fuzzy search by code/name/pinyin initial + dropdown selection, no need to look up codes
   • 🔄 Multi-Source Fallback: qstock main source + akshare/Eastmoney/Tencent/NetEase/efinance backups; auto-switch when one fails
   • ⚙️ Hot Config Reload: YAML + Excel "Config" Sheet; watchlist/refresh interval can be hot-applied in Excel without restart
   • 🧪 Out-of-the-Box: One command auto-generates the Excel template (7 Sheets), no need to prepare stock lists in advance
   • ✅ Unit Tests: pytest covers core logic (156 items), stable for long-running sessions

3. Streamlit Real-Time Market Monitor

   • 🌐 Web-based real-time market dashboard

📁 Code Directory: egs_alpha

📁 Code Directory: egs_data

• Detailed usage of various common data sources
• Unified data source interface

📁 Code Directory: egs_fin_nlp

📁 Code Directory: egs_llm

📁 Directory: a_全网优秀资源

⭐ The Highlight Section of This Repo: Curated, organized, and reviewed top-quality quantitative resources from across the entire web, all in one place!

This is the most essential "resource treasure trove" of this repo — we spent tremendous effort curating, organizing and reviewing tens of thousands of materials from across the web, classified by the full quantitative trading workflow so that you can quickly find the tools and materials you need, avoiding detours.

Differences from Other Sections of This Repo:

• 🔍 Best of the Best: Selected from the vast sea of resources across the web to avoid repeated pitfalls
• 📂 Clear Categorization: Classified by the full quant workflow (data → strategy → backtest → trading), easy to find what you need
• 📝 With Reviews: Not just links, but also pros/cons analysis and getting-started guides
• 🔄 Continuously Updated: Keeping up with technology development, continuously adding new resources

• 🤖 LLM Applications in Finance: Covers the latest research and practice including FinGPT, FinMem, Self-Reflective, Stock-chain, TradingAgents, FinRobot, etc.
• 🛠️ Skill Pack Collection (60+): Includes Chan Theory, technical analysis, quant statistics, fundamental analysis, crypto, macro analysis, etc.
• 📊 Backtest Framework Multi-dimensional Comparison: Hands-on comparison of Backtrader, Zipline, RQAlpha, PyAlgoTrade, QuantDigger, etc.
• 🔬 Research Report Reproduction: Selected high-quality broker research reports with reproduction code
• 💹 Complete Trading Strategies: From traditional double MA to RL and GNN, covering all types of strategy resources

💡 Usage Tip: Browse the a_全网优秀资源 directory as needed; if you're interested in a project, click to view the detailed introduction and review.

For easier maintenance, the original ai_wiki directory content (system operations, programming basics, AI basics, AI practice, etc.) has been independently synchronized to the repo AI LLM Pitfall Guide.

It records a large number of problems and solutions encountered in actual development, and tracks cutting-edge technology developments in real-time. Welcome to follow and Star ⭐

✨ AI LLM Pitfall Guide

• Github: https://github.com/charliedream1/ai_wiki
• Gitee (Domestic Mirror): https://gitee.com/charlie1/ai_wiki.git
• Introduction: Share various practical cases, track cutting-edge technology developments, covering the full stack of AI knowledge — including LLMs, programming techniques, machine learning, deep learning, reinforcement learning, GNN, speech recognition, NLP, image recognition, etc.

📁 Code Directory: egs_online_platform

Domestic quant platforms such as JoinQuant, Uqer, MiRuo, GuoRen, and BigQuant can be tried by interested readers.

Research platforms are cloud platforms tailored for quant enthusiasts (quants), providing free stock data acquisition, accurate backtesting, high-speed live trading interfaces, easy-to-use API documentation, and a strategy library from easy to difficult, making it convenient to quickly implement and validate strategies.

⚠️ Note: The following strategies are only valid for the backtest periods described and have not been carefully tuned or verified for full-period performance. No strategy can guarantee effectiveness across all periods, so please be cautious when using them in live trading.

🔗 JoinQuant Platform · Welcome to follow me: 量客攻城狮

• For detailed strategy introductions and source code, please click the corresponding strategy link
• JoinQuant Usage Introduction: egs_online_platform/聚宽_JoinQuant
• This part of the code can only run on JoinQuant Platform

Stock Quant Strategies:

Stock Analysis Research:

• Hands-on Tutorial on "ML - Dynamic Multi-Factor Stock Selection" (with nanny-level tutorial)
• Dragon-Tiger List Data Filtering
• Concept Sector Data Acquisition and Stock Selection
• Detailed Explanation: Stock Data Acquisition & Graphical Analysis (with detailed code)

(Our article with 26,000+ reads on Zhihu) The Ultimate Collection of AI Stock Quantitative Trading Tools and Open-Source Projects

We have re-categorized and reviewed all tools, collected in the ai_notes folder for easy lookup.

🎯 In Development:

• Continuously reviewing all tools to facilitate selection
• Continuously documenting the pros and cons of each tool to form comparison tables for easy selection
• Continuously documenting usage: We don't make comprehensive tutorials, only list the most commonly used and practical features, so you can get started quickly

This code repository adheres to the principle of parallel paid and free offerings.

Register on the official Knowledge Planet website for guaranteed user rights. Planet Content Introduction

👇 Scan the QR code below or click the link to enter the Planet and view more detailed introductions 🎏

Planet Video Introduction:

• Planet Usage Guide: https://mp.weixin.qq.com/s/SGc49e0xf24q5aUbf3rO0g?token=2028063978&lang=zh_CN
• Learning Path & Group Resource Usage: https://mp.weixin.qq.com/s/3-U048mc0riVsdETrKr77g

Planet Join Links:

• AI智投星球: AI quantitative trading crash course, cutting-edge tech, real-world cases, resource library
• AI速成营: In-depth supplements on programming, LLMs, AI basics, principles and finance practice & job-seeking cases, complementary to AI智投星球

Planet Introduction:

• Planet Content Introduction
• New User Guide

👇 Scan the QR code for a more detailed introduction to the "Planet" (there are funny comics inside)!

🎯 This repo will continue to be updated, but some code will be privately maintained and only visible on the Planet. The corresponding features will be noted in the repo.

WeChat Official Account

🔥 Real-time updates on the latest news 🎁 Follow and like any article, DM the admin to receive a free exquisite quant material package!

• Zhihu: 576 Followers
• JoinQuant: 599 Followers

Code Repository (Forever Free)

✨ AI Quantitative Trading Bot

• Github: https://github.com/charliedream1/ai_quant_trade
• Gitee (Domestic Mirror): https://gitee.com/charlie1/ai_quant_trade.git

Companion Repo

✨ AI驯龙笔记 (AI Taming Notes)

• Github: https://github.com/charliedream1/ai_wiki
• Gitee (Domestic Mirror): https://gitee.com/charlie1/ai_wiki.git
• Introduction: Share various practical cases, track cutting-edge technology developments, covering the full stack of AI knowledge — including LLMs, programming techniques, machine learning, deep learning, reinforcement learning, GNN, speech recognition, NLP, image recognition, etc.

Your support is the driving force for me to move forward. Even "0.1 RMB" makes me happy. Thank you for your support (^o^)/

    

Feel free to start a discussion in Github Discussions.

• Feel free to submit issues at Github Issues
• Join the Knowledge Planet for more technical support
  • AI智投星球: Focused on AI quantitative trading knowledge sharing
  • LLM Pitfall Guide: Focused on programming, LLMs, and AI application empowerment

Please see the documentation → FAQ

@misc{ai_quant_trade,
  author={Yi Li},
  title={ai_quant_trade},
  year={2022},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/charliedream1/ai_quant_trade}},
}