Directed Information Flow in Macro Prediction Markets: Separating Genuine Coupling from Common-Driver Artifacts

Estimation code, pinned reproduction environment, and aggregated result files for the paper of the same title. A transfer-entropy network is estimated over Kalshi macro prediction-market composites; an identification framework separates genuine directed coupling from common-driver artifacts using scheduled macroeconomic releases (FOMC, CPI, employment) as known-timing perturbations, and every surviving edge is cross-validated against the nonparametric Kraskov-Stoegbauer-Grassberger estimator. The aggregated CSV outputs regenerate every table and figure. The raw Kalshi candlestick data is not redistributed, consistent with Kalshi's data terms; the repository documents the API fetch needed to rebuild it.