AVET: A Provenance-Aware Venue Emulator Trace Dataset for Event Markets

Venue integrations are often tested through endpoint mocks, isolated lifecycle fixtures, or final-balance assertions. Those artifacts omit the ordered evidence needed to reproduce how a venue moved from market data to order acceptance, matching, settlement, close, resolution, redemption, or operational failure. This paper introduces the Axient Venue Emulator Trace Dataset (AVET), a deterministic and provenance-aware format for event-venue emulator traces, and reports the AVET v0.2.0 corpus. An AVET trace is a self-contained ordered event stream with source provenance, logical and observed clocks, causation identifiers, typed payloads, publication boundaries, and layered cryptographic identities. The specification distinguishes synthetic, rights-cleared derived, raw-permitted, hash-only, and private-not-distributed source classes. AVET v0.2.0 contains 60 fully synthetic source-generated traces from an immutable emulator release: twelve each in BOOK, ORDER, SETTLEMENT, LIFECYCLE, and FAULT. The corpus contains 240 events, fourteen event types, seven sequence families, and a 40/10/10 core, stress, and negative split. All traces pass strict offline schema and semantic validation; the distributed candidate includes a separately labelled two-mutation smoke test; and Python and Node.js agree on 12,000 recursive key-order checks. Event-stream hashes and trace digests are publicly recomputable, while scenario-bundle and terminal-state hashes are explicitly labelled source commitments. AVET is not empirical evidence about a live venue and does not estimate latency, liquidity, availability, fill quality, or settlement reliability. Its contribution is a reusable trace object, source-anchored synthetic corpus, explicit data-rights boundary, exact replay tooling, and publication protocol for venue adapters, audit tools, and future empirical calibration.