An advanced AI-powered stock trading system featuring 250+ intelligent agents (200 retail + 50 institutional) powered by GPT-5 and Claude 4.5 Sonnet, plus a 12-expert consensus panel conducting multi-round deliberations to achieve 30%+ annual returns.
- 🤖 Dual AI Engine: GPT-5 + Claude 4.5 Sonnet working together
- 👥 250+ Independent Agents: Each makes autonomous decisions
- 200 Retail Agents (5 types: momentum chasers, panic sellers, herd followers, value seekers, technical traders)
- 50 Institutional Agents (5 types: quantitative, statistical arbitrage, machine learning, high frequency, market makers)
- 🎓 12 Elite Experts: Multi-round consensus deliberation
- 📊 Focused Assets: Index ETFs, Gold, Precious Metals, Blue Chips
- 🏛️ Institutional Strategy: Follow smart money, counter retail sentiment
- 🎯 Target: 30%+ annual returns with controlled risk
pip install numpy pandas anthropic openai yfinance# GPT-5 (OpenAI)
export OPENAI_API_KEY="sk-xxxxx"
# Claude 4.5 Sonnet (Anthropic)
export ANTHROPIC_API_KEY="sk-ant-xxxxx"# Full system with agents + experts
python ultimate_trading_system.py
# Expert-only version
python expert_consensus_system.py
# Elite version (institutional-grade)
python elite_trading_system.py| File | Description | Target | Best For |
|---|---|---|---|
ultimate_trading_system.py |
250+ Agents + 12 Experts | 30%+ | Maximum intelligence |
expert_consensus_system.py |
12 Experts only | 30%+ | Focused decision-making |
elite_trading_system.py |
Full position, multi-factor | 15-30% | Institutional-grade |
ai_trading_complete_system.py |
Original complete system | 15-25% | Learning & research |
forex_trading_example.py |
Forex with MT5 | High risk | Forex traders |
Market Data (yfinance)
↓
┌─────────────────────────────────────────────┐
│ 250+ Independent Agents │
│ ┌──────────────┐ ┌──────────────────┐ │
│ │ 200 Retail │ │ 50 Institutional │ │
│ │ - Momentum │ │ - Quant │ │
│ │ - Panic │ │ - Stat Arb │ │
│ │ - Herd │ │ - ML │ │
│ │ - Value │ │ - HFT │ │
│ │ - Technical │ │ - Market Maker │ │
│ └──────────────┘ └──────────────────┘ │
│ ↓ ↓ │
│ Each agent makes independent decision │
│ Powered by GPT-5 or Claude 4.5 │
└─────────────────────────────────────────────┘
↓
┌─────────────────────┐
│ 12 Expert Panel │
│ 6-Round Discussion │
└─────────────────────┘
↓
Final Consensus Decision
↓
Trade Execution (100% position)
"Follow Institutions, Counter Retail"
- When retail panics (extreme fear) + institutions buy → BUY
- When retail euphoric (extreme greed) + institutions sell → SELL
- Track institutional flow via market microstructure analysis
- Exploit retail behavioral biases
Index ETFs: SPY, QQQ, DIA, IWM Sector ETFs: XLF, XLK, XLE Gold & Precious Metals: GLD, GDX, GDXJ, SLV Blue Chips: AAPL, MSFT, NVDA, TSLA, AMZN
- Stop Loss: 4%
- Take Profit: 12%
- Trailing Stop: 3%
- Max Drawdown: -15%
- Target Sharpe: > 2.0
-
English:
EXPERT_CONSENSUS_GUIDE.md- Expert system guideFINAL_SUMMARY.md- Project summaryFOREX_TRADING_GUIDE.md- Forex trading guide
-
Chinese:
快速开始.md- Quick start guide一键回测使用指南.md- Backtest guide精英系统使用指南.md- Elite system guide系统版本对比.md- Version comparison
| Metric | Target | Description |
|---|---|---|
| Annual Return | 30%+ | Aggressive returns |
| Sharpe Ratio | > 2.0 | Risk-adjusted performance |
| Max Drawdown | < -15% | Capital preservation |
| Win Rate | > 65% | Trading accuracy |
- AI Models: GPT-5, Claude 4.5 Sonnet
- Data: yfinance (real US stock data)
- Analytics: NumPy, Pandas
- Factors: Multi-factor quantitative models
- Risk: VaR, CVaR, Kelly Criterion
MIT License - See LICENSE file
This system is for educational and research purposes only. Trading carries significant risk. Past performance does not guarantee future results. Never trade with money you cannot afford to lose.
先进的AI驱动股票交易系统,配备250+智能Agent(200散户 + 50机构),由GPT-5和Claude 4.5 Sonnet双引擎驱动,加上12位顶级专家进行多轮讨论,目标实现年化30%+收益。
- 🤖 双AI引擎: GPT-5 + Claude 4.5 Sonnet协同工作
- 👥 250+独立Agent: 每个都自主决策
- 200个散户Agent(5类:追涨杀跌、恐慌型、跟风型、价值型、技术型)
- 50个机构Agent(5类:量化、统计套利、机器学习、高频、做市商)
- 🎓 12位精英专家: 多轮共识讨论
- 📊 专注资产: 指数ETF、黄金、贵金属、蓝筹股
- 🏛️ 机构策略: 顺机构、反散户
- 🎯 目标: 控制风险下实现30%+年化收益
pip install numpy pandas anthropic openai yfinance# GPT-5 (OpenAI)
export OPENAI_API_KEY="sk-xxxxx"
# Claude 4.5 Sonnet (Anthropic)
export ANTHROPIC_API_KEY="sk-ant-xxxxx"# 完整系统(agents + 专家)
python ultimate_trading_system.py
# 纯专家版本
python expert_consensus_system.py
# 精英版(机构级别)
python elite_trading_system.py| 文件 | 说明 | 目标收益 | 适合人群 |
|---|---|---|---|
ultimate_trading_system.py |
250+ Agent + 12专家 | 30%+ | 追求最大智能 |
expert_consensus_system.py |
纯12专家系统 | 30%+ | 专注决策质量 |
elite_trading_system.py |
全仓轮动、多因子 | 15-30% | 机构级别 |
ai_trading_complete_system.py |
原始完整系统 | 15-25% | 学习研究 |
forex_trading_example.py |
外汇MT5交易 | 高风险 | 外汇交易者 |
市场数据 (yfinance)
↓
┌─────────────────────────────────────────────┐
│ 250+独立Agent │
│ ┌──────────────┐ ┌──────────────────┐ │
│ │ 200散户Agent │ │ 50机构Agent │ │
│ │ - 追涨杀跌 │ │ - 量化 │ │
│ │ - 恐慌型 │ │ - 统计套利 │ │
│ │ - 跟风型 │ │ - 机器学习 │ │
│ │ - 价值型 │ │ - 高频交易 │ │
│ │ - 技术型 │ │ - 做市商 │ │
│ └──────────────┘ └──────────────────┘ │
│ ↓ ↓ │
│ 每个Agent独立决策 │
│ 由GPT-5或Claude 4.5驱动 │
└─────────────────────────────────────────────┘
↓
┌─────────────────────┐
│ 12专家委员会 │
│ 6轮深度讨论 │
└─────────────────────┘
↓
最终共识决策
↓
交易执行(全仓)
"顺机构,反散户"
- 散户恐慌(极度恐惧)+ 机构买入 → 买入
- 散户狂热(极度贪婪)+ 机构卖出 → 卖出
- 通过市场微观结构分析追踪机构流向
- 利用散户行为偏差
指数ETF: SPY, QQQ, DIA, IWM 行业ETF: XLF, XLK, XLE 黄金及贵金属: GLD, GDX, GDXJ, SLV 蓝筹股: AAPL, MSFT, NVDA, TSLA, AMZN
- 止损: 4%
- 止盈: 12%
- 移动止损: 3%
- 最大回撤: -15%
- 目标夏普比率: > 2.0
-
英文文档:
EXPERT_CONSENSUS_GUIDE.md- 专家系统指南FINAL_SUMMARY.md- 项目总结FOREX_TRADING_GUIDE.md- 外汇交易指南
-
中文文档:
快速开始.md- 快速入门一键回测使用指南.md- 回测使用指南精英系统使用指南.md- 精英系统指南系统版本对比.md- 版本对比
| 指标 | 目标 | 说明 |
|---|---|---|
| 年化收益 | 30%+ | 激进收益 |
| 夏普比率 | > 2.0 | 风险调整后收益 |
| 最大回撤 | < -15% | 资本保护 |
| 胜率 | > 65% | 交易准确率 |
- AI模型: GPT-5, Claude 4.5 Sonnet
- 数据: yfinance(真实美股数据)
- 分析: NumPy, Pandas
- 因子: 多因子量化模型
- 风险: VaR, CVaR, Kelly公式
MIT License - 查看LICENSE文件
本系统仅供教育和研究用途。交易存在重大风险。历史业绩不代表未来表现。切勿用无法承受损失的资金进行交易。
Contributions are welcome! Please feel free to submit a Pull Request.
欢迎贡献!请随时提交Pull Request。
For questions or support, please open an issue on GitHub.
如有问题或需要支持,请在GitHub上开issue。
Made with ❤️ by AI Trading Team
由AI交易团队用❤️制作