A real-time machine learning app that predicts short-term price movement probabilities for major cryptocurrencies.
Built with production-level structure, trained on engineered features, and deployed on Streamlit Cloud.
🔗 Live Demo: cryptosignalprediction.streamlit.app
Given a coin like Bitcoin, Ethereum, or Solana, the app:
- Fetches the last 10 days of real-time market data
- Engineers financial features like volatility, momentum, and price change
- Uses a trained ML model to estimate the probability of upward movement in the next 3 days
- Provides a simple, user-friendly prediction message
- ✅ Real-time CoinGecko data integration
- ✅ Feature engineering on price, volume, volatility, momentum
- ✅ Trained XGBoost classifier with multicoin support
- ✅ Model persisted via
joblib
and served via Streamlit
- Bitcoin
- Ethereum
- Solana
- Dogecoin
- Shiba Inu