FinGPT GUI

🚀 Features

1. Multi-Model Prediction Engine

• LSTM (Long Short-Term Memory): Captures sequential dependencies in daily price movements.
• Transfomer (TFT): A lightweight Temporal Fusion Transformer for attention-based time series forecasting.
• Hybrid "Week Hunter" Strategy: Combines models to identify stocks with >50% weekly upside potential.

2. Interactive Streamlit Dashboard

• Single Stock Deep Dive: Real-time charts, RSI/MACD indicators, and AI prediction overlays.
• Market Brain: Mass training capability to let the model learn generic market physics from multiple tickers.
• Opportunity Scanner: Auto-scans the S&P 500/NASDAQ for the best Conviction setups.
• 50% Week Hunter: Specialized high-volatility scanner for aggressive growth opportunities.

3. Robust Backend

• Dockerized Environment: One-click deployment with docker-compose.
• GPU Acceleration: CUDA-enabled PyTorch for fast training and inference.
• Polygon.io Integration: High-fidelity market data fetching.
• Sentiment Analysis: VADER-based news sentiment scoring.

1. Single Stock Analysis

• Enter a ticker (e.g., NVDA).
• Select a historical period (1y, 2y).
• View the AI's Next Day Target and Signal (BUY/SELL/HOLD).

2. Switching Models

• Use the Sidebar Selector to switch between LSTM (Default) and TFT (Transformer).
• Compare predictions to confirm trends.

3. The "50% Week Hunter"

• Go to the High Alpha Hunter tab.
• Click HUNT.
• The AI will scan high-volatility tickers and rank them by Conviction Score (Volatility + AI Confidence + Sentiment).

🏗️ Project Structure

FinGPT/
├── FinGPTGUI/
│ ├── app/
│ │ ├── gui_dashboard.py # Streamlit Frontend
│ │ ├── fin_algo.py # AI Backend (LSTM, TFT, Data)
│ │ ├── requirements.txt # Python Dependencies
│ │ └── Dockerfile # GUI Container Config
│ └── ...
├── docker-compose.yml # Service Orchestration
└── README.md # This file

⚠️ Disclaimer

This software is for educational and research purposes only. It is not financial advice. Trading stocks involves risk.