Data conflation (conflated vs unconflated feeds)
Conflation is the practice of combining several market data updates into one, typically by sending only the latest state of a price level or quote at intervals instead of every change. It saves bandwidth and processing but hides intermediate states: quotes that flickered, liquidity that appeared and vanished between two snapshots, or the sequence of individual trades.
Senzoukria · Glossary · Updated September 2026
Definition
An unconflated feed delivers every event in order. A conflated feed delivers the current state at a lower rate: if a level changed five times since the last update, the consumer receives only the fifth value. Conflation can apply to quotes, to depth and, in aggregated form, to trades.
Worked example
A price level changes 120 times in one second as orders are added and cancelled. An unconflated feed delivers 120 updates. A feed conflated to 10 updates per second delivers at most 10 states and never shows the other 110. A display refreshing 30 times per second can show at most 30 states, even when all 120 were received and applied.
What conflation hides
- Flickering orders and short-lived liquidity, which matter to anyone studying pulling and stacking.
- The exact order of quote changes relative to trades.
- Individual trades, when trades are aggregated into one per interval or per price.
- Queue dynamics that only a full event stream can reconstruct.
Feed conflation versus display refresh
The two are often confused. If the feed is unconflated, every event reaches the application and is applied to its state; the screen then shows that state at its refresh rate. Totals built from events, such as volume at price and delta, remain exact. If the feed itself is conflated, the missing events never reach the application and no display can recover them.
In Senzoukria
The footprint aggregates every trade it receives. Display paths are refreshed at bounded rates: the Databento book is published to the heatmap at about 30 updates a second, and the broker's position stream is coalesced to at most 4 updates a second per instrument, with opening, closing and reversing a position published immediately. Those are display choices applied after the data arrived, not a conflation of the feed.
Common mistakes
- Studying spoofing or pulled liquidity on a conflated depth feed.
- Blaming a display's frame rate for events the feed never sent.
- Comparing trade counts from an aggregated trade stream with an unaggregated one.
Related
In the same section
- Data vs software cost
- Settlement price
- Data feed
- Price limits
- Data freshness
- Daily loss limit
- Data gap
- d1 and d2
This page in other languages
Frequently asked questions
- Is a conflated feed less accurate?
- Its values are correct at the moments it reports, but it omits what happened in between. For price and totals that can be fine; for studying order book dynamics it is a real loss.
- How do I know whether my feed is conflated?
- Ask the provider, or check its documentation for update rates and whether it delivers every book event. A feed that updates at a fixed cadence regardless of activity is a sign of conflation.