Market Data Normalization Interview Guide
Market data normalization interview guide for raw feeds, canonical schemas, timestamps, symbol mapping, validation, and tradeoffs.
Candidates building or discussing multi-venue market data systems.
Normalization creates a common interface
Different feeds may represent symbols, book updates, trades, and statuses differently. A normalized layer turns raw messages into internal events that strategies and tools can consume.
Some detail can be lost
Normalization is useful, but it can hide venue-specific semantics. A strong answer says which fields must stay raw or traceable for debugging and compliance.
Concrete example
Two venues may encode trading halts differently. A normalized status event can help consumers, but the original raw message should remain available for investigation.
Validation protects downstream users
Check symbol maps, timestamps, sequence gaps, impossible prices, duplicate events, and stale feeds before publishing normalized data as trusted.
Common mistakes
Candidates often treat normalization as simple renaming. In trading systems, schema design affects correctness, latency, replay, and incident debugging.
Practice the pattern
Use the LeetQuidity curriculum and calibration to turn this topic into a focused practice plan.