The Replacement Principle: When Monitoring Backfires in Delegated Portfolio Management

Description

This replication package supports the paper "The Replacement Principle: When Monitoring Backfires in Delegated Portfolio Management." It contains Python 3 scripts, a shell pipeline (run_all_empirics.sh), public and derived data, and a frozen snapshot of all headline result JSONs and figure PDFs. The empirical analysis draws on SEC EDGAR Form 13F-HR filings, EDHEC hedge-fund indices, PivotalPath strategy returns, ETF and Fama-French / Q5 / AQR benchmark data, and Numerai public tournament leaderboards. No proprietary or restricted data are used. Outputs include the 104-strategy panel regressions, 13F turnover-tail tests, the Bartik 2SLS specification, the Numerai cross-section, the E13 position-level replacement-trap test, policy-frontier simulations, and aggregation-loop stress simulations. See code/README.md and code/DATA_PROVENANCE.md for source attributions, processing notes, and a list of required final-release metadata.

Steps to reproduce

Extract code.zip, data.zip, and logs.zip into the same parent directory. Open a terminal in the code/ folder, run "pip install -r requirements.txt", then run "bash run_all_empirics.sh" to regenerate all headline outputs. The frozen reference outputs are pre-loaded in logs/ for inspection without re-running the scripts; run "python3 code/verify_replication.py" for a structural integrity check

Institutions

• Texas Tech University

  Texas, Lubbock

Categories

Licence