intelligent-trading-bot

Why we included this project

What makes this bot worth a look is that it treats live trading as the real test of a machine-learning pipeline, not backtest performance. Training happens offline in batch mode while prediction runs online, and the project takes care that both modes use identical derived features so the model keeps operating consistently over months of live use. Crypto traders and quant developers can extend the feature set with custom Python functions alongside standard technical indicators and switch between one-minute, hourly, and daily timeframes from a single configuration. That same config drives the online service, which fetches fresh data, computes signals, and routes them to Telegram, an API endpoint, a database, or actual trade execution, while the backtesting path periodically retrains models on historic data. It is a solid, well-documented reference for anyone wiring model training, feature engineering, and live signal delivery into their own trading stack.