🌟 KairosPredict

📈 Predicting Future Stock Market Trends with Machine Learning

KairosPredict is an intelligent stock trend prediction application that analyzes historical patterns using clustering algorithms and machine learning to forecast future market behavior. It provides real-time insights through a user-friendly graphical interface backed by robust data processing and model evaluation.

---

🚀 Key Features

✅ Predict future stock market trends using advanced Machine Learning
✅ Real-time stock data fetched using TradingView API
✅ Interactive and modern Graphical User Interface (GUI) built with CustomTkinter
✅ Supports multiple clustering algorithms (KMeans, DBSCAN, Agglomerative)
✅ Intelligent pattern detection using Perceptually Important Points (PIPs)
✅ Multi-threaded execution for smooth and responsive performance
✅ Customizable themes and chart configurations (Dark/Light mode, colors, MA lines)
✅ Authentication system with OTP-based password recovery
✅ Pattern cluster visualization and future pattern simulation
✅ Future-ready: Designed to scale as a SaaS platform

---

⚠️ Note

If you're facing issues running app.py, you can use the lightweight version:

python app_lite.py

---

🛠 Installation & Setup

1️⃣ Install Python (version ≥ 3.8)

2️⃣ Install dependencies:

pip install -r requirements.txt

3️⃣ Run the application:

python app.py

---

📂 Modules & Their Role

🏷 Module	📌 Usage
📊 NumPy	Efficient numerical computations
🧠 Scikit-learn	ML algorithms: clustering, classification
🖼 CustomTkinter	Beautiful, modern GUI for dashboard
📑 Pandas	Data handling, analysis, and transformations
⚙️ Threading	Enhances app responsiveness and multitasking
📈 TradingView API	Fetches live stock market data
🖼 Pillow (PIL)	For image and icon rendering in GUI

---

🔍 How It Works

1. Pattern Extraction: Uses a PIP algorithm to extract significant trend points from historical data using various distance metrics (Euclidean, Perpendicular, Vertical).

2. Clustering: Clusters the patterns using KMeans, DBSCAN, and Agglomerative clustering. Only unique patterns (based on ID) are stored.

3. Classification: Current trend is analyzed and classified using a Random Forest model to find the most similar historical pattern.

4. Prediction: The matched pattern’s future segment is scaled and aligned with the current trend, providing a visual forecast of the upcoming movement.

5. Voting System: Among different cluster outputs, a voting mechanism selects the most accurate matching pattern for display.

---

📈 Performance

✔️ Achieves over 65% prediction accuracy
📄 Check detailed reports in model_evaluation_report.md

✔️ Scalable architecture designed for future integration with additional models and datasets.

---

🔮 Planned Enhancements

• ✅ Expand ML model variety including AI-driven deep learning methods
• ✅ Add support for more stocks and indices
• ✅ Release mobile and web versions
• ✅ Enable full SaaS deployment with subscription-based access
• ✅ Integration of technical indicators and auto-detection tools

---

🧠 Architecture Diagram

---

📞 Contact & Support

📧 Email: [email protected]
📱 Phone: +91 63804 98136