# QuickFIX/J four-flow local parity preflight

This is a separate-process localhost diagnostic, not a Stage 5 latency result.
It used the current QuickFIX/J 2.3.1 adapter, a separate acceptor and
initiator, five warmups, 20 measured samples, one run, and 100/s offered rate
for each canonical flow. The acceptor was configured to exit after session
logout; both processes exited zero on every flow.

| Flow | Measured samples | Wire audit |
|---|---:|---|
| `nos_er` | 20 | pass |
| `md` | 20 | pass |
| `md_nos_er` | 20 | pass |
| `tick_to_trade` | 20 | pass |

The [result table](evidence/quickfixj-local-20260930/results.tsv) and
[retained evidence](evidence/quickfixj-local-20260930/) include the client
raw samples, CSV summaries, manifests, canonical cycle dumps, process logs,
and per-flow configuration. The canonical `35=8` is a new-order ACK with
`150=0`, `39=0`, `14=0`, and the full `151` leaves quantity. The
`tick_to_trade` dump has the measured `35=W → 35=D` pair without a reply.
`contract_audit.audit_dump` passed all four dumps.

The existing 750/s SCAN local profile remains the only valid lower-rate
QuickFIX/J evidence cited by the operational contract. This small parity
preflight does not establish 10,000/s capacity, physical DAC traversal, or
report-grade Stage 5 status.

## Three-run server sample boundary check

A separate [three-run local preflight](evidence/quickfixj-multirun-local-20260930/results.tsv)
used 20 measured samples per run, five warmups per run, and 100/s. Both
processes exited zero and the wire audit passed in all four flows. The
tick-to-trade [server raw series](evidence/quickfixj-multirun-local-20260930/tick_to_trade/server_tick_to_trade.bin)
has exactly three runs of 20 samples each. Before the QuickFIX/J client sent a
pause command at each run boundary, that series had 25, 25, and 20 samples:
the first two included the next run's five warmups. The correction discards
those warmups on the server clock before each measured phase. The small local
run's stability gate failed, so its latency figures are diagnostic only.
