Design a low-latency trading platform

Problem Statement

Design a high-frequency trading (HFT) platform capable of ingesting market data from multiple exchanges, executing automated strategies, and managing orders. The system must operate with microsecond-level latency while maintaining rigorous pre-trade risk controls.

Functional Requirements

1. Market Data Feed Handler: Consume UDP multicast feeds from multiple exchanges, normalize data, and publish it to an internal messaging bus.
2. Strategy Engine: Subscribe to normalized market data, execute trading logic based on current state, and emit order signals.
3. Order Management System (OMS): Manage the lifecycle of orders, track fills, and handle cancellations/modifications.
4. Pre-trade Risk Engine: Validate every outgoing order against limits (e.g., maximum position size, daily loss limit, fat-finger checks) before sending to the exchange.

Non-Functional Requirements

• Latency: End-to-end latency from market data packet arrival to order gateway departure must be under 50 microseconds.
• Throughput: Support spikes of up to 1 million messages per second.
• Determinism: Minimize jitter caused by garbage collection or thread context switching.
• Reliability: The system must be fault-tolerant; a single component failure should not result in an unauthorized market position.

Expected Deliverables

• A high-level system architecture diagram (described in text).
• Selection of transport protocols and data serialization formats (e.g., SBE, Protobuf, shared memory).
• Discussion on how to minimize 'stop-the-world' latency spikes.
• A strategy for how risk checks can be performed without significantly bloating the hot path.
• Approaches for logging and monitoring without impacting performance.