PayoffAlpha: Reproducibility Package for a Knowledge-Based Evidence-Governance Architecture
Description
This dataset provides the reproducibility package for the manuscript “PayoffAlpha: A Knowledge-Based Evidence-Governance Architecture for Adaptive Discovery Systems.” It contains deterministic Python scripts, locked configurations, derived non-price ledgers, boundary-state decisions, mutation and threshold tests, fair MCDM comparisons, performance benchmarks, and robustness evidence. Licensed Tushare source market data are not redistributed; retrieval and reconstruction instructions are included. The validation scripts require Python 3.11 or later and no third-party packages.
Files
Steps to reproduce
Requirements: - CPython 3.11 or later - No third-party Python packages - No network connection or market-data API credentials are required for the included validation From the extracted dataset directory, run: python scripts/verify_kbs_tara_knowledge_base.py python scripts/evaluate_kbs_expert_system_controls.py python scripts/evaluate_kbs_mcdm_performance.py The first command verifies the locked knowledge base, evidence permissions, deterministic inference, all 6,912 boundary states, and the ensemble-member SHA-256 lock. The second command reproduces the controller comparison, five rule-mutation tests, and fourteen threshold-perturbation experiments. The third command reproduces the fair TOPSIS/ELECTRE/TARA comparison and system-performance benchmarks. Each command should terminate successfully and report "status": "pass". The supplied MANIFEST.sha256 can be used to verify the integrity of all released files. Licensed Tushare market data are not included. Reconstructing the full market backtests requires a licensed Tushare account and a local Qlib-compatible data provider; see DATA_RETRIEVAL.md. The included return-blind expert-system experiments can be reproduced without those data.
Institutions
- Zhejiang University of Finance and EconomicsZhejiang, Hangzhou