Anomaly Detection in Financial Trading Patterns

Design a sophisticated Python script for detecting complex anomalies in financial trading data. Implement a multi-layered approach using unsupervised machine learning techniques including isolation forests, local outlier factor, and autoencoder neural networks. Create a comprehensive analysis that identifies both point anomalies and contextual anomalies in high-frequency trading datasets. Include performance optimization for processing large-scale financial time series data with minimal computational overhead.