A comprehensive platform for forecasting Forex prices, detecting arbitrage opportunities, optimizing currency portfolios, and visualizing key financial metrics — all powered by deep learning and modern web technologies.
This project combines advanced machine learning models with an interactive frontend to deliver accurate Forex price predictions and actionable insights. It supports multi-currency analysis, portfolio optimization, and arbitrage detection.
- 📈 AI-Driven Predictions: LSTM neural networks trained on 10+ years of historical Forex data
- 🌍 Multi-Currency Support: Predict and analyze EUR/USD, GBP/USD, USD/CHF, USD/JPY
- 💸 Arbitrage Detection: Spot pricing discrepancies between data sources
- 📊 Portfolio Optimization: Mean-variance optimization with Sharpe ratio maximization
- ⚖️ Risk & Return Analysis: Sharpe ratio, Sortino ratio, max drawdown, and more
- 🧠 Interactive Visualizations: Clean, responsive data dashboards
- 💻 Modern UI: Built with React, Tailwind, and
shadcn/uicomponents
- Python 3.8+
- TensorFlow / Keras – LSTM models
- Pandas, NumPy – data manipulation
- SciPy – optimization algorithms
- Matplotlib, Seaborn – visualizations
- Streamlit – optional dashboard interface
- React + TypeScript
- Tailwind CSS
shadcn/uicomponent library- Vite (fast dev/build tool)
QF-Forex/
├── Project.ipynb # Jupyter notebook (data prep, modeling, plotting)
├── app.py # Streamlit app (optional UI)
├── models/ # Saved LSTM model files
├── src/ # React frontend source code
├── Graphs/ # Generated visualization images
├── *.csv # Historical Forex data
└── README.md # Project documentation
- Node.js (v18+)
- Python 3.8+
- Git
git clone https://github.com/AP4549/Forex-Future.git- Backend Setup (Python)
python -m venv venv
source venv/bin/activate # or venv\Scripts\activate on Windows
pip install -r requirements.txt- Frontend Setup (React)
npm install
npm run dev
- (Optional) Streamlit Interface
streamlit run app.py
⚙️ Usage 📊 Data Collection Historical Forex data is fetched from:
Alpha Vantage API
Yahoo Finance API
The data is cleaned, merged, and stored as .csv files for modeling.
🧠 Model Training Each currency pair is trained using a separate LSTM model, designed to capture time-series patterns for price forecasting.
💼 Portfolio Optimization A built-in mean-variance optimizer calculates the ideal asset weights to maximize the Sharpe ratio under defined risk constraints.
🔁 Arbitrage Detection Real-time and historical price comparisons from different APIs are used to detect arbitrage opportunities in the Forex market.
📡 Data Sources Alpha Vantage (10-year historical data)
Yahoo Finance (recent historical + real-time data)
📄 License This project is licensed under the MIT License – see the LICENSE file for full details.