Are Whitepaper Claims Reflected in Market Structure? A Contamination-Aware Pipeline and a Power-Limited Null

Working paper DAI-2508. Do the functional narratives in cryptocurrency whitepapers correspond to how their tokens behave in markets? We develop a content-verified, contamination-aware pipeline for measuring structural correspondence between project narratives and market structure, and report two results. The first is a cautionary one. An apparent entity-level signal in an earlier version of our corpus — specialised tokens appearing to align more strongly than broad infrastructure tokens — was entirely an artefact of corpus contamination: roughly a quarter of the documents were failed-download stubs or wrong-document whitepapers (for example, a “Cosmos” entry that was in fact Binance Smart Chain text), and the apparent ordering does not survive content verification. We report it as a contamination diagnosis, not a finding. The second is an honest null. Combining zero-shot NLP classification of 43 content-verified whitepapers across 10 semantic categories with seven cross-sectional market-structure statistics (hourly data, 17,543 timestamps, 2023–2024), aligned via Procrustes rotation and Tucker’s congruence coefficient φ, we do not detect a significant claims–market alignment in this n=43 sample (dimension-matched φ=0.303, primary estimate; zero-padded φ=0.223, a conservative display floor; both non-significant). A positive-control and power analysis shows the binding constraint is the low reliability of the text instrument: at mean inter-method reliability the minimum detectable effect is φ≈0.66. Our results are inconsistent with strong alignment (φ≥0.70), which the design has over 80% power to detect, but cannot distinguish weak alignment from none. We frame the contribution as a method plus a cautionary tale for text-based studies of narrative–market correspondence, which routinely operate below an unstated detectability floor. v4.0.0 — major revision: clean content-verified 43-whitepaper corpus (after detecting ~25% contamination), removal of the tensor factor-analysis leg, reframing as a contamination-aware method plus a power-limited null. Reproducible code and data bundled (CC BY 4.0). v5.0.0 (2026-07-11): Matches arXiv v6 content. Power analysis corrected: the minimum detectable congruence under the matched-geometry (statistics-based) positive control is φ ≈ 0.44 (the previously headlined 0.66 is the factor-based control’s value; both are now reported and labeled); realistic-power column relabeled (96% detection at φ = 0.50); Appendix J deduplicated against Appendix D.7 (CCA and RV coefficient results were already reported there); effective-rank conventions reconciled (participation ratio 6.5 vs Shannon entropy 7.6, same matrix); the nuclear-norm/congruence derivation rewritten (Q* maximizes the trace form; φ at Q* is scale-invariant and distinct from the norm); §4.5 attribution chain aligned with the pipeline; Samieifar & Baur (2021) added to related work. The v4.0.0 code-and-data archive is retained unchanged.