End-to-end crypto price direction prediction framework built on the Mamba Selective State Space architecture (Python/PyTorch). Covers the full lifecycle: data preparation, Triple Barrier labeling, training, evaluation & live inference — all from a single config. Linear-time sequence processing as a Transformer alternative.

This project explores Attention-Based Transformer Encoders to develop robust buy/sell classification models for financial time series. It addresses market non-stationarity and noise by combining De Prado-inspired preprocessing with a hybrid Transformer-LSTM architecture.