TradePre

TradePre is a desktop-based trading prediction system that collects OHLCV data, engineers technical indicators, creates R1-before-S1 labels, trains multiple machine learning models, validates predictions with walk-forward backtesting, and runs a daily prediction workflow. It also includes a Claude/LLM synthesis layer and a Flet desktop interface for setup, training, and signal review.

Traders and analysts need a structured way to test whether price is more likely to hit bullish or bearish pivot levels first, instead of relying only on manual chart reading or untested assumptions.

Collects market data, builds technical features, labels historical outcomes, trains ensemble ML models, backtests predictions, runs daily prediction workflows, and presents signals through a desktop interface.

Built as a hands-on AI/ML trading system project, covering workflow design, Python implementation, model pipeline, backtesting logic, desktop interface, and iterative testing.