🛡️ AI-Powered ETF Core-Satellite Strategy Backtester

AI 智能投資回測系統：核心定投 + 衛星動能

🚀 Demo / 線上試用

Try the app instantly with AI Analysis: 立即體驗包含 AI 分析功能的完整回測：

👉 Launch App (開啟應用程式)

(Note: You will need your own Google Gemini API Key to enable the AI features. Getting a key is free!) (注意：您需要輸入自己的 Google Gemini API Key 才能啟用 AI 對話功能，申請 Key 是免費的！)

---

📖 Introduction / 專案簡介

This requires a modern approach to backtesting. Not just numbers, but Insights. 這不僅僅是一個回測工具，而是一個擁有 AI 副駕駛 的投資分析系統。

This project integrates Quantitative Finance with Generative AI (LLM). It tests a hybrid strategy combining "Core Holding" (Beta) and "Satellite Trading" (Alpha), while an AI Agent analyzes the results, diagnoses risks, and provides optimization suggestions in real-time.

本專案結合了 量化金融 與 生成式 AI。除了測試「核心持股 + 衛星交易」的混合策略外，更內建了一位 AI 投資顧問，能即時解讀回測報表、診斷風險，甚至回答您關於「如何提高報酬」或「分析空頭市場表現」的提問。

---

✨ New Features / 全新功能

🤖 AI Investment Co-pilot (AI 投資副駕駛)

• Automated Diagnosis: Automatically generates a performance report (Risk, Profitability, Suggestions) after backtesting.
• Interactive Q&A: Chat with the AI about your strategy! Ask questions like "Why did the drawdown happen in 2022?"
• Smart Suggestions: Quick-action buttons to evaluate risk or optimize parameters.
• 自動化診斷：回測結束後，AI 自動生成風險評估與參數優化建議。
• 互動式問答：對數據有疑問？直接在對話框詢問 AI，就像身邊有一位專業分析師。
• 智慧建議：提供快捷按鈕，一鍵分析風險或獲取優化方向。

📊 Advanced Visualization (進階視覺化)

• Interactive Equity Curves: Compare your strategy vs. Benchmark (Buy & Hold).
• Asset Allocation Stacked Chart: Visualize how funds flow between Core (Safe) and Satellite (Aggressive) assets.
• 互動式權益曲線：動態比較策略與大盤的績效差異。
• 資產堆疊圖：清晰呈現核心與衛星資產的資金流向與倉位變化。

---

💡 Strategy Logic / 策略邏輯

The strategy divides the portfolio into two parts: 本策略將投資組合分為兩個部分：

1. 🛡️ Core Position (核心資產 - ex: VOO)

• Goal: Long-term stability.
• Method: Monthly Dollar-Cost Averaging (DCA).
• 目標：長期穩健增長。
• 方法：每月定期定額投入。

2. 🚀 Satellite Position (增強資產 - ex: QQQ)

• Goal: Capture Alpha during market dips.
• Method: Simulates aggressive buying (or leveraged ETFs/Options) when price drops below a threshold.
• Exit Rule: Staged profit-taking (0-9 months). Force sell after 9 months.
• 目標：市場大跌時進場撿便宜。
• 方法：跌幅觸發買入，可模擬槓桿操作。
• 出場：階梯式止盈，超過 9 個月強制平倉。

3. 🛡️ Risk Management (風控機制)

• 200 MA Filter: Stops buying when the price is below the 200-day Moving Average.
• 200 MA 濾網：跌破年線時停止買入，避開空頭市場。

---

📸 Screenshots / 介面預覽

策略參數設定與回測結果

資產配置資金流向視覺化

AI諮詢分析生成

---

💻 How to Run Locally / 如何在本地執行

1. Clone the repository

   git clone [https://github.com/bullhsu/etf-backtest-app.git](https://github.com/bullhsu/etf-backtest-app.git)
   cd etf-backtest-app

2. Install requirements

   pip install -r requirements.txt

3. Get a Free Gemini API Key

   • Go to Google AI Studio and create a new API key.
   • 前往 Google AI Studio 免費申請 API Key。

4. Run the app

   streamlit run app.py

---

🛠️ Tech Stack / 使用技術

• Frontend: Streamlit
• AI Engine: Google Gemini API (via google-generativeai)
• Data Source: yfinance
• Analysis: Pandas, NumPy
• Visualization: Plotly

---

⚠️ Disclaimer / 免責聲明

This tool is for educational and research purposes only. It does not constitute financial advice. The AI analysis is generated by a Language Model and may contain errors. 本工具僅供教育與研究用途，不構成任何投資建議。AI 分析結果由語言模型生成，可能存在誤差，投資請自行評估風險。