Relevance
8/10
Highly relevant as it provides rigorous evidence against quantum kernel advantage in cross-sectional stock return prediction, identifies methodological pitfalls (survivorship bias, insufficient power, weak baselines) that plague applied QML-finance literature, and proposes concrete protocol standards for evaluating quantum advantage claims. The finding that penalized linear models dominate the equity cross-section is directly actionable for quant practitioners. The anatomy of spurious advantage is a cautionary tale for any ML-based trading strategy evaluation.
Implementation Complexity
7/10
Moderate-to-high complexity. Requires quantum circuit simulation (PennyLane), kernel ridge regression with Nyström approximation, walk-forward evaluation with factor rotation, multiple classical baselines (ridge, XGBoost, MLP, NN3), interaction catalog construction, bandwidth diagnostics, geometric difference computation, and multiplicity-corrected inference. The 14-notebook pipeline with data construction from raw exchange/financial-statement data adds significant engineering overhead. However, at 8 qubits the quantum kernel is classically computable, reducing hardware requirements.
Reproducibility
4/5
Full pipeline implemented in fourteen Jupyter notebooks with all intermediate artifacts persisted to disk. Quantum simulation uses PennyLane. Code and intermediate data for full replication available from the author. However, no public GitHub repository is explicitly linked, and data availability is 'on request' rather than open access.