Replication Materials: Disposition-Effect Measurement in Cryptocurrency Perpetual Futures

This replication package accompanies a study of the disposition effect and measurement granularity in cryptocurrency perpetual-futures trading records. It contains the frozen research design, the execution- and episode-level analysis pipeline, derived wallet-level data, robustness analyses, order-level consolidation, full-flow checks, and a separate validation sample drawn from all trading addresses. Version 2.0.0 preserves the original frozen design and historical pipeline materials while adding the final manuscript-facing analyses and reporting files. These additions include denominator-precision checks, the fill-to-episode consolidation ladder, estimation-sample ex-post results, all-address validation, exploratory regression outputs, final figure source files, and a machine-readable crosswalk linking reported results to deposited inputs and scripts. The central measurement analysis examines how behavioral inference changes when the same underlying executions are consolidated from fills to orders, reduction spells, position generations, time-gap clusters, and coin-day realization episodes. Historical intermediate outputs are retained for auditability and are clearly distinguished from the final reported results. Raw trade executions and account statements are not redistributed because of their size. They are publicly available from the Hyperliquid ecosystem archive released by Artemis (Amazon S3 requester-pays), and the deposited code documents the acquisition and reconstruction procedure. Deposited derived data contain public on-chain wallet addresses only; no personally identifying information was collected. See README.md, REPRODUCTION_NOTES.md, SPECIFICATION_HISTORY.md, and CHANGELOG.md for package structure, reproduction instructions, and version history.