High-Frequency Trading Latency Optimization Middleware

Create a Node.js middleware for optimizing API communication in high-frequency trading environments, focusing on minimizing latency and maximizing throughput. Implement a custom connection pooling mechanism that can handle over 10,000 concurrent API requests with sub-10ms response times. Design a sophisticated caching layer using Redis that intelligently predicts and pre-loads market data, reducing API call redundancy. Include advanced circuit breaker patterns to automatically handle API failures and prevent cascading system errors.