Machine Learning Feature Store for Predictive Trading

Create a high-performance feature store using Apache Cassandra and Node.js that supports real-time feature engineering for machine learning trading models. Design a schema that can ingest multiple data streams (market data, sentiment analysis, economic indicators), implement efficient feature versioning, and support point-in-time correct feature retrieval for backtesting trading algorithms. Include mechanisms for feature drift detection and automated model retraining.