From Public Evidence to Contractual Outcome: First and Stable Decidability on Kalshi

Battle Count: 50

The paper is relevant to prediction market trading and settlement but is not directly about quantitative trading strategies, portfolio construction, or algorithmic execution. Its primary contribution is a measurement framework for when public evidence becomes contractually sufficient to determine a prediction market outcome. For traders, understanding the gap between first decidability and venue determination (LFD) could inform settlement risk, information timing, and the value of early position unwinding. The paper explicitly states no price study or cross-venue result is reported. The relevance is indirect—through understanding settlement latency and information observability in prediction markets rather than through trading signal generation or execution optimization.

The methodology involves a complex multi-layered protocol: formal rule-evidence graph construction with versioned rule objects, source hierarchy tracking, correction/revision graphs, deterministic stratified sampling with square-root weights and largest-remainder adjustment, blind packet construction with strict no-outcome-leakage boundaries, two-phase adjudication (blind review then venue linkage), interval/partial identification estimation, cluster bootstrapping, and a registered support disposition mechanism. The infrastructure requires continuous observation of official source releases, rule versioning, lifecycle frames, and reconnect receipts. The human adjudication component adds significant operational complexity. However, no computational model training is involved—the complexity is in protocol design, data governance, and measurement infrastructure rather than algorithmic implementation.

The paper provides a highly detailed and registered protocol including deterministic sampling algorithms, blinding boundaries, adjudication classifications, and a KMVE dataset (doi: 10.17632/fn65786cg6.1). However, no empirical results are yet reported—the core measurement task (blind adjudication of decidability clocks) remains unexecuted. The prospective enrollment is active but the final sample is conditional on support thresholds at enrollment close. Replication materials are referenced but the actual registry is generated at enrollment close. The historical feasibility test is complete and reproducible, but the prospective study is still in progress.