HYQi A-Share AI Quant Trading System A-share-Ai

An automated A-share trading system built on four major domestic LLMs (HYQi, DeepSeek, Qwen, Doubao), supporting full-market stock selection and live trading.

Connect iClaw for one-command mobile control of four major models with fully automated trading.

Features

🎯 Let AI compete in the A-share market — multiple LLMs make fully autonomous decisions and face off across the whole market!

🎯 Connects to real A-share trading accounts, showing the AI's live trading process in real time, widening the system's application scope

🎯 Multi-model integration: supports multiple large language models including HYQi, DeepSeek, Qwen, and Doubao

🎯 Strategy Evolution: AI models continuously optimize strategies based on trading performance

⏰ Second-level live trading support

🚀 Service deployment & parallel execution

🎨 AI trading logic visualization - detailed trading log visualization (full trading process display)

🎯 Core feature: 100% AI autonomous decision-making, zero human intervention; AI strategies self-discover, self-train, and self-evolve

🎯 Real-time human & AI interaction during trades (optional). Integrates in-house iClaw to command all 4 AIs at once.

🎯 Newly added asynchronous execution to improve system speed

🤖 4-model parallelism: HYQi, DeepSeek, Qwen, Doubao run simultaneously, comparing returns 📈 Full-market stock selection: AI autonomously picks promising stocks market-wide; custom stock pools (optional) 💹 Real-time quotes: real-time A-share data integrated for AI use in live trading 💹 Historical + financial data: A-share historical market & financial data for AI analysis and training 🎯 Smart trading: AI-driven buy/sell signals, executed fully automatically 🌈 Multiple strategies: trend-following, moving-average, technical indicators, quant arbitrage, etc., all chosen and evolved by AI 📊 Real-time monitoring: UI shows return curves and trading logs live 🔒 Risk control: built-in position sizing and stop-loss mechanisms

📱 iClaw DingTalk Configuration

Method 1 (deprecated): after enabling config/dingtalk.json, the program prompts you to first send /status in a private chat with the DingTalk bot to complete binding; once binding succeeds, your account ID is automatically written to default_conversation_id, and each completed trading round will be pushed automatically to your DingTalk private chat (no need to configure a personal ID manually).

Method 2: obtain the sender_id via the test script or the DingTalk platform and write it into the default_conversation_id field of config/dingtalk.json, and configure other required parameters (see the example file); after the program starts, it pushes each round's trading results directly to that account, while also enabling remote real-time control of all the project's LLMs via DingTalk.

Stream long-lived connection: inbound messages go over WebSocket Stream, outbound notifications use the REST API (independent). If no private-chat commands are received for a long time while trading notifications still work normally, the log will show Stream 已断开，10s 后重连...; the program auto-reconnects, no restart needed.

💬 iClaw Smart Chat and Natural-Language Trading Commands

• Smart chat (multi-model): send @模型名 你的问题 and the bot replies via the corresponding LLM (chat only, no orders placed).
  • Examples: @幻银超i 今天市场怎么看？, @deepseek 解释一下市盈率
• Natural-language AI trading: send @iClaw 你的自然语言交易意图; the bot calls the model you mentioned to translate the intent into a JSON command, validates it, then executes via /buy / /sell.
  • Example: @iClaw 用幻银超i买入002961 1000股，价格26.9
• Slash-command trading:
  • Buy/Sell: /buy [模型名] 股票代码 数量 [价格], /sell [模型名] 股票代码 数量 [价格]; model name optional. Examples: /buy 000001 1000, /sell 幻银超i 000001 500 27.0
  • Batch buy: /batch_buy [模型名] 股票代码:数量,股票代码:数量,...; model name optional (defaults if omitted). Example: /batch_buy 000001:1000,000002:500 or /batch_buy 幻银超i 000001:1000,000002:500
  • Conditional trade: /condition [模型名] 股票代码 条件 买入/卖出 [数量]; model name optional. Conditions support price>数字, price<数字, price_between 低 高; example: /condition 000001 price>11 buy 1000 or /condition 幻银超i 000001 price>11 buy 1000. Quantity optional: when omitted, buy auto-calculates quantity by risk and available funds, sell liquidates all positions; if the current price doesn't meet the condition, the system monitors in the background at each trading-round interval and fires a market order on trigger (default validity 6 hours). Condition-order monitoring runs in an independent task and does not occupy the main per-round trading loop. After creation, the bot returns a condition-order ID; cancel: /cancel <条件单ID> (e.g., /cancel COND-1734567890-000001).
• Full command list: send /help or /h in DingTalk to see all commands and examples.
• Remote launch of local apps: send /runapp <应用别名或内置名称>, the main program will try to launch a whitelisted desktop app on your local machine.
  • Aliases are set in APP_LAUNCH_CONFIG.aliases of user_config.json, e.g. "mytrader": "C:\\Users\\you\\Desktop\\MyTrader\\MyTrader.exe"
  • For normalized programs such as notepad, calc, just use /runapp notepad (can also be extended via builtin_allow)

📊 AI Stock Competition Leaderboard, Real-Time Returns & Decision Logs, Real-Time Portfolio

Installation & Startup

The project has been packaged as a Windows desktop EXE and can be run directly. The internal test period has ended; official commercial use begins from March 2026. How to obtain it: go to https://hyqibot.com/card-shop.html and purchase a service card to get the software. For questions, send your GitHub address (required) + a brief introduction to: hyqi@tradey.dpdns.org. For more information, visit: https://www.hyqibot.com/ A Lobster iClaw with permanent free tokens is included as a gift.

Startup is simple: beginner-friendly, no Python knowledge required Download the Releases installation package, unzip it, rename user_config.json.example to user_config.json, and fill in the parameters The parameters needed fall into a few categories: 1) the installation path and account of your real trading account; professional trading APIs can be connected (inquire via official email); compatible with QMT and Tonghuashun trading from major brokerages; 2) the local path of your custom stock pool (Tongdaxin); 3) API keys for LLMs such as DeepSeek; 4) licence: hyqibot After filling in the parameters and saving, right-click to run the unzipped EXE as an administrator, then click Start Trading in the GUI.

Core Leap: True AI Self-Discovery

1. Start from a blank slate or a user-set style All AIs can initially be "adaptive investors"

   Investment styles can also be preset

   Discover their own strengths through practice

2. Self-reflection mechanism Regularly review investment performance

   Analyze which strategies work for themselves

   Proactively adjust their investment philosophy

3. Style evolution Record the trajectory of style changes

   Adjust risk appetite based on performance

   Gradually form a stable investment personality

4. True differentiation Now the 4 AIs will:

   Autonomously discover styles: for example, HYQi best understands the essence of game theory, DeepSeek may discover that it excels at quantitative methods, Qwen may prefer value investing

   Different evolutionary paths: each AI forms a unique style based on its own traits and market feedback

   Dynamic adjustment: adapt strategies to changes in the market environment

   True individuality: develop an investment philosophy that reflects its "personality"

   Expected outcomes You will see:

   True AI individuality: each AI develops a unique investment identity

   Dynamic style evolution: investment styles change as experience accumulates

   Performance-based adjustments: successful strategies are retained, failed ones are eliminated

   Unpredictable results: it's impossible to know in advance which AI will develop which style

   With this design, you will witness a true "AI investment manager growth story" rather than a pre-scripted performance. Every AI has the chance to find the investment path that best suits it!

🧠 The Vision of HYQi

We train AI in live real trading to test and boost its intelligence, using open-ended & large-scale RL to tame its complexity.

To this end, we're hiring senior AI researchers, startup partners & big thinkers.

If you aspire to build Alpha-HYQi for the real world, we welcome you to join us. Please send your resume to: hyqi@tradey.dpdns.org

"Capital allocation is the ultimate test for AI intelligence."

——— HYQi, sharing this with you

📞 Support & Community

💬 Discussion: GitHub Discussions

🐛 Bug reports: GitHub Issues

📜 License terms

1. Open-source core modules (MIT License)

The open-source code (located in the ./codes directory of this repository) is licensed under the MIT License; you have the following rights:

Freely use, copy, modify, merge, publish, distribute, sublicense, and sell copies of the open-core code;

Integrate open-core code into personal/commercial (incl. closed-source) projects, retaining copyright notice & MIT license.

Full license text: LICENSE

2. Commercial closed-source modules Commercial modules (including all modules embedded in the packaged EXE) are copyright-protected and provided under a commercial license; key terms:

Requires a valid commercial license (contact us for plans);

No unauthorized distribution, RE, or resale;

Commercial modules must pair with this project's open-core; no standalone use or integration elsewhere.

Full details of the commercial license: LICENSE-COMMERCIAL (provided after license purchase)

🤝 Contribution guide

We welcome contributions to the open-core module! Steps:

Fork the repository;

Modify code only in the ./codes directory;

Submit a Pull Request with a clear description of the changes.

Detailed contribution rules: CONTRIBUTING.md

Note: Closed-source; no contributions.

📞 Contact & support

Open-Core Users

Bug reports: submit bug reports or feature requests related to the open-source core modules via GitHub Issues;

Community discussion: join our [GitHub Discussions] to participate in discussions.

Commercial users

Technical support: contact the dedicated support team via [hyqi@tradey.dpdns.org] (response within 24 hours);

Licensing inquiries: send an email to [hyqi@tradey.dpdns.org] to learn about pricing, subscription plans, or custom licensing services.

⚠️ Disclaimer

The open-source core modules are provided under the MIT License "as is", without any express or implied warranties (see the full MIT License text for details);

Commercial modules include tech support per license; all-environment compatibility not guaranteed;

Dual-license model may change; updates announced here and notified to commercial users.

© [2025] HYQi. All rights reserved. The open-source core modules are licensed under the MIT License. Commercial modules are licensed under a Commercial License.

A-share-Ai: Phantom Silver Quantitative AI Trading System for A-Shares

An automated A-share trading system powered by four major domestic large language models (HYQi Phantom Silver Ultra-i, DeepSeek, Qwen, Doubao), supporting full-market stock selection and live trading.

Key Features

🎯 Let AI showcase its capabilities in the A-share market—multiple large language models make fully autonomous decisions and compete against each other across all A-share stocks!

🎯 Connect to real A-share trading accounts, display the AI's real trading process in real time. Real-time + real trading expands the system's application scenarios.

🎯 Multi-model Integration: Supports HYQi Phantom Silver Ultra-i, DeepSeek, Qwen, Doubao, and other large language models.

🎯 Strategy Evolution: Based on trading performance, AI models continuously optimize their investment strategies.

⏰ Second-level Live Trading Support - Connect to live trading with second-level precision.

🚀 Service Deployment & Parallel Execution - Deploy production services + parallel model execution.

🎨 AI Trading Logic Visualization - Detailed trading log visualization (full trading process display).

🎯 Core Feature: 100% AI autonomous decision-making with zero human intervention. AI strategies self-discover, self-train, and self-evolve.

🎯 Special Function: Real-time human intervention and interaction with AI during trading (optional), adapting to real-world scenarios of large-scale investments.

🎯 Newly added asynchronous execution to improve system speed.

🤖 4-Model Parallelism: Simultaneous operation of HYQi Phantom Silver Ultra-i, DeepSeek, Qwen, and Doubao to compare returns.📈 Full-Market Stock Selection: AI independently selects promising stocks from the entire market, with customizable stock pools (optional).💹 Real-Time Market Data: Integrates real-time A-share data for AI use in actual trading.💹 Historical Market + Financial Data: Integrates A-share historical market and financial data for AI analysis and training.🎯 Intelligent Trading: Generates buy/sell signals based on AI analysis and executes trades fully automatically.🌈 Multiple Strategies: Includes trend following, moving average strategies, technical indicator strategies, quantitative arbitrage strategies, etc.—fully selected and continuously evolved by AI.📊 Real-Time Monitoring: UI interface displays real-time profit curves and trading logs.🔒 Risk Control: Built-in position control and stop-loss mechanisms.

📊 AI Stock Trading Competition Rankings, Real-Time Returns & Decision Logs, Real-Time Portfolio

Installation & Usage

The project is packaged as a desktop EXE application, which can be downloaded and run directly. A 3-month free closed beta is provided. Users whose feedback is adopted will receive professional technical support and long-term usage rights. Reservation email: hyqi@tradey.dpdns.org. Please Star the project (click the star icon in the upper right corner), then send your contact information, a brief introduction, and an explanation of your usage purpose. We will decide whether to arrange closed beta access based on your email content. If you are granted closed beta qualification, you will be notified by email within 3-5 working days.

Launch is simple: No Python experience required (beginner-friendly). Download the installation package from Releases and unzip it. Rename user_config.json.example to user_config.json and fill in the parameters. Required parameters include: Installation path and account of your real trading account. Professional trading APIs are supported (inquire via official email), compatible with QMT and Tonghuashun trading platforms of major securities firms. Local path of your custom stock pool (Tongdaxin format). API keys for large models such as DeepSeek. License: hyqibot

Save the filled parameters, right-click to run the unzipped EXE file as an administrator, and click "Start Trading" on the GUI.

Core Leap: True AI Self-Discovery

1. Start from Scratch or User-Defined Style All AIs can start as "adaptive investors" initially. Investment styles can also be preset. Discover their own strengths through practice.
2. Self-Reflection Mechanism Regularly review investment performance. Analyze which strategies work for themselves. Proactively adjust investment philosophy.
3. Style Evolution Record the trajectory of style changes. Adjust risk appetite based on performance. Gradually form a stable investment personality.
4. True Differentiation

Now the 4 AIs will: Autonomously discover styles: For example, HYQi Phantom Silver Ultra-i masters the essence of trading, DeepSeek may excel in quantitative analysis, and Qwen might prefer value investing. Different evolutionary paths: Each AI forms a unique style based on its own characteristics and market feedback. Dynamic adjustment: Adapt strategies to changes in market conditions. True individuality: Develop an investment philosophy that reflects its "personality."

Expected Outcomes

You will witness:

True AI individuality: Each AI develops a unique investment identity.

Dynamic style evolution: Investment styles change with experience accumulation.

Performance-based adjustments: Successful strategies are retained, failed ones are eliminated.

Unpredictable results: It is impossible to predict which style each AI will develop in advance.

With this design, you will witness a true "AI investment manager growth journey" rather than a pre-scripted performance. Each AI has the opportunity to find the most suitable investment path!

🧠 Vision of HYQi Phantom Silver Ultra-i

We are training AI in a real-time, real trading environment to test and enhance its intelligence, leveraging technologies such as open-ended learning and large-scale reinforcement learning to harness its complexity.

With this vision, we are actively recruiting senior AI researchers, entrepreneurial partners, and visionary thinkers.

If you aspire to build Alpha-HYQi for the real world, we welcome you to join us. Please send your resume to: hyqi@tradey.dpdns.org

"Capital allocation is the best touchstone for AI intelligence to be tested by truth."

—— Together with HYQi Phantom Silver Ultra-i

📞 Support & Community

💬 Discussions: GitHub Discussions

🐛 Issues: GitHub Issues

📜 License Terms

1. Open-Core Module (MIT License)

The open-source source code (located in the ./codes directory of this repository) is licensed under the MIT License. You are granted the following rights:

Freely use, copy, modify, merge, publish, distribute, sublicense, and sell copies of the open-core code.

Integrate the open-core code into personal/commercial projects (including closed-source projects), provided that the original copyright notice and MIT License text are retained.

Full license text: LICENSE

2. Commercial Closed-Source Modules

Commercial modules (including but not limited to all modules embedded in the packaged EXE file) are protected by copyright and provided under a Commercial License. Key terms are as follows:

Usage requires a valid commercial license (contact us for subscription/authorization plans).

Unauthorized distribution, decompilation, reverse engineering, or resale of commercial modules is prohibited.

Commercial modules may only be used in conjunction with the open-core module of this project, and shall not be used independently or integrated into other unauthorized projects.

Full commercial license details: LICENSE-COMMERCIAL (provided upon license purchase)

🤝 Contributing Guidelines

We welcome contributions to the open-core module! Contribution steps:

Fork this repository.

Modify code only in the ./codes directory.

Submit a Pull Request with a clear description of changes.

Detailed contribution rules: CONTRIBUTING.md

Note: Commercial modules are closed-source and do not accept external code contributions.

📞 Contact & Support

For Open-Core Users Issue Reporting: Submit bug reports or feature requests related to the open-core module via GitHub Issues.

Community Discussion: Join our [GitHub Discussions] to participate in discussions.

For Commercial License Users

Technical Support: Contact the dedicated support team via [hyqi@tradey.dpdns.org] (response within 24 hours).

License Inquiries: Send an email to [hyqi@tradey.dpdns.org] to learn about pricing, subscription plans, or custom authorization services.

⚠️ Disclaimer

The open-core module is provided "as is" under the MIT License, without any express or implied warranties (see the full MIT License for details).

Commercial modules will be provided with technical support in accordance with the license agreement, but compatibility with all custom environments is not guaranteed.

The dual-license model is subject to change. Any updates will be announced in this README and notified to commercial license holders.

© [2025] HYQi Phantom Silver Ultra-i. All rights reserved. Open-Core Module Licensed under MIT License. Commercial Modules Licensed under Commercial License.