BTC Up/Down 5min ML Ensemble
Uses an ML ensemble of LGBM, XGB, and LSTM models to predict BTC price direction every 5 minutes and trades on Polymarket.
This strategy leverages machine learning to trade Bitcoin's short-term price direction on Polymarket. It employs an ensemble of three models: LightGBM (LGBM), XGBoost (XGB), and Long Short-Term Memory (LSTM) networks. The models are trained on historical BTC price data and technical indicators to predict whether the price will go up or down in the next 5-minute window. The bot is auto-trained via GitHub Actions, meaning it periodically retrains on new data without manual intervention. It places bets on Polymarket's BTC Up/Down 5-minute markets when the ensemble's confidence exceeds a threshold. Entry is at the start of the 5-minute window, and exit is at the end or when the position is closed. A live dashboard on GitHub Pages shows performance, likely tracking win rate and P&L. The large repo size (599MB) suggests significant training data or model artifacts. The strategy is fully automated and runs continuously, making it a statistical arbitrage approach based on predictive modeling.