Explain the largest multi-millisecond latency stalls

Determine the cause of the separate population of multi-millisecond stalls observed in the socket-to-book latency measurements.

Background

The live CME market-data study characterizes the typical latency floor, serialization slope, queue occupancy, and packet-position effects, but reports a separate population of exceptionally large stalls reaching the millisecond range. The paper explicitly states that these observations remain unexplained within the study, leaving their cause unresolved.

References

And a separate population of multi-millisecond stalls --- the largest latencies in the dataset (the max column of Table~\ref{tab:mddist}) --- is real and remains unexplained here.

— Adapting the Actor Model of Concurrency for High-Frequency Trading: Synchronous Message Delivery (fast_send) and a Tick-to-Book Latency Study  (2609.21173 - Maciejewski, 18 Sep 2026) in Section 13, “Limits of this measurement” paragraph