The research question
An order-level event stream records changes to individual orders. A depth snapshot sums the displayed quantity at each price at one moment. When the former is reduced to the latter or to a feature vector, which distinctions disappear? The preprint studies that question through venue-constrained state transitions and the information available to an observer. Its central object is an event-state path, rather than a price series alone.
The motivating issue is concrete: cancelling an order ahead of a named resting order can change its queue position, while cancelling the same quantity behind it does not. Both events can produce the same aggregate depth update.
Event semantics
Which transitions are admissible?
event + stateRepresentation
Which distinctions are lost?
state → observationValidation
Does the distinction matter for the task?
hypothesis + evidenceThree separate layers
Semantics and observation. Distinguish the full venue state, the state reconstructed from a feed, and information available to a live strategy. A replay is exact only relative to its feed, parser and venue rules.
Projection and task sufficiency. A projection groups detailed states into a shared observation. If a target differs across states in that group, the observation alone cannot determine it exactly. The question is sufficiency for a specified task.
Candidate features and validation. Residual identity loss, queue-ahead fragility and cancellation topology are candidate feature families. Their usefulness requires separate tests with information boundaries, latency and costs made explicit.
Inspect the smallest example
INTERACTIVE / QUEUE POSITION
Same depth. Different position.
Three unit orders share one price in a first-in, first-out (FIFO) queue. I is the order we observe. Choose whether to cancel U ahead of I, or V behind it.
BEFORE
Total depth 3 · Quantity ahead of I: 1
AFTER
Total depth2
Quantity ahead of I0
Cancelling U changes the quantity ahead of I from 1 to 0; total depth falls from 3 to 2.
Assumptions: one price, three displayed unit orders, FIFO and one full cancellation; no hidden quantity or new events. This example does not calculate fill probability.
Scope and limitations
This is a preprint, not a peer-reviewed journal publication or a trading-performance record. It does not establish that richer data always improves decisions. Additional detail can increase estimation variance, processing cost or latency; some information may only become available retrospectively.
Hidden liquidity, replace semantics, feed gaps, auctions and a strategy’s own actions can require additional modelling. A useful empirical programme should test one target and one feature family at a time, after validating replay and information availability.
Read and cite the source
The linked Zenodo record identifies this manuscript as version v2, published on May 30, 2026. An SSRN copy is also linked below. Repository copies can be updated independently; check the record and version when citing.
Xiong, Yu (2026). Event-Semantic Limit Order Books: Projection Loss, Microstructure Signals, and Queue-Aware Execution. Preprint, May 30, 2026. https://doi.org/10.5281/zenodo.20411291