Relevance
9/10
Highly relevant for quantitative trading in corporate bonds. The paper demonstrates that most published corporate bond factor strategies (momentum, reversal, volatility, macro betas) are artifacts of measurement error and look-ahead bias rather than genuine alpha. For practitioners, this means: (1) Short-term reversal strategies are largely unimplementable (91% of premium is bias); (2) Momentum strategies require infeasible ex-post filtering to appear profitable; (3) Only credit-spread-based value factors retain meaningful risk-adjusted returns after correction; (4) The open-source PyBondLab software and bias-corrected data enable practitioners to build implementable strategies; (5) The NSE framework highlights that portfolio construction choices can flip factor signs, critical for strategy robustness. The paper essentially provides a roadmap for credible corporate bond factor investing.
Implementation Complexity
6/10
Moderate complexity. The core methodology (signal gap, return gap, ex-ante filtering) is conceptually straightforward and implemented in PyBondLab. However, full replication requires: (1) Access to raw TRACE enhanced and 144A transaction data; (2) Mergent FISD for bond characteristics; (3) Understanding of the multi-stage data pipeline (transaction cleaning, decimal shift correction, bounce-back filter, distressed bond filters, daily aggregation, monthly return computation); (4) Proper implementation of within-firm sorts and multiple portfolio construction variants; (5) Newey-West standard errors and FDR corrections. The open-source software significantly reduces implementation burden, but the data infrastructure requirements (TRACE, FISD, WRDS) represent a barrier for independent replication.
Reproducibility
5/5
Exceptional reproducibility. The paper provides: (1) Open source TRACE data pipeline on GitHub (github.com/Alexander-M-Dickerson/trace-data-pipeline); (2) Companion software PyBondLab on PyPI (pypi.org/project/PyBondLab/); (3) Full replication code on GitHub; (4) Pre-formed factors and bond-month data for all 108 signals at openbondassetpricing.com; (5) Data also available on WRDS Contributed Data (daily and monthly); (6) All filter thresholds are configurable; (7) Companion data reports automatically generated documenting every cleaning step; (8) Detailed parameter tables in appendix. The entire pipeline from raw TRACE transactions to monthly factor returns is reproducible.