Seed Snapshot
A seed snapshot is an initial full order book state used as the starting point for applying incremental updates and reconstructing the book over time.
In market data systems, incremental order book updates usually describe changes to an existing book rather than the complete state. A consumer therefore needs a known starting point. The seed snapshot supplies that starting point by giving a full or sufficiently complete view of the order book at a specific time or sequence position.
After the seed snapshot is loaded, later updates can be applied in the correct replay order to maintain an accurate local order book. If the seed snapshot is stale, incomplete, or not aligned with the update stream, the reconstructed book may be incorrect.
Seed snapshots are commonly used in order book reconstruction, historical replay, data recovery, and systems that need to restart from a known state after a sequence gap or connection loss.
Why It Matters
Seed snapshots matter because incremental updates are only meaningful when applied to the correct initial state. Without a valid seed snapshot, a consumer cannot reliably reconstruct the current order book or replay historical order book changes.
They help developers reduce bandwidth, restart processing safely, and verify that replayed market data remains consistent with the expected book state.
What is a seed snapshot in market data?
A seed snapshot is the initial order book state used to start processing an update stream. It usually contains price levels, quantities, and sometimes sequence or timestamp metadata that identifies where replay should begin.
How is a seed snapshot used with incremental updates?
A consumer first loads the seed snapshot, then applies incremental order book updates that occur after the snapshot point. The updates must be applied in the correct order and without missing sequence numbers.
What happens if the seed snapshot is wrong?
If the seed snapshot does not match the update stream, the reconstructed order book may become inaccurate. The consumer may need to request a new snapshot, detect a sequence gap, or reset the replay process.
Real-World Example
A trading application downloads a full BTC/USD order book snapshot at sequence number 10,000. It then applies incremental updates starting at sequence number 10,001. The snapshot is the seed state, and each update changes that state until the application has a live reconstructed book.